You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1339 lines
42 KiB

Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Allow serving API only via new cli command - Add new command to the cli/node: `serveapi` that alows serving the API just by connecting to the PostgreSQL database. The mode flag should me passed in order to select whether we are connecting to a synchronizer database or a coordinator database. If `coord` is chosen as mode, the coordinator endpoints can be activated in order to allow inserting l2txs and authorizations into the L2DB. Summary of the implementation details - New SQL table with 3 columns (plus `item_id` pk). The table only contains a single row with `item_id` = 1. Columns: - state: historydb.StateAPI in JSON. This is the struct that is served via the `/state` API endpoint. The node will periodically update this struct and store it int he DB. The api server will query it from the DB to serve it. - config: historydb.NodeConfig in JSON. This struct contains node configuration parameters that the API needs to be aware of. It's updated once every time the node starts. - constants: historydb.Constants in JSON. This struct contains all the hermez network constants gathered via the ethereum client by the node. It's written once every time the node starts. - The HistoryDB contains methods to get and update each one of these columns individually. - The HistoryDB contains all methods that query the DB and prepare objects that will appear in the StateAPI endpoint. - The configuration used in for the `serveapi` cli/node command is defined in `config.APIServer`, and is a subset of `node.Config` in order to allow reusing the same configuration file of the node if desired. - A new object is introduced in the api: `StateAPIUpdater`, which contains all the necessary information to update the StateAPI in the DB periodically by the node. - Moved the types `SCConsts`, `SCVariables` and `SCVariablesPtr` from `syncrhonizer` to `common` for convenience.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Redo coordinator structure, connect API to node - API: - Modify the constructor so that hardcoded rollup constants don't need to be passed (introduce a `Config` and use `configAPI` internally) - Common: - Update rollup constants with proper *big.Int when required - Add BidCoordinator and Slot structs used by the HistoryDB and Synchronizer. - Add helper methods to AuctionConstants - AuctionVariables: Add column `DefaultSlotSetBidSlotNum` (in the SQL table: `default_slot_set_bid_slot_num`), which indicates at which slotNum does the `DefaultSlotSetBid` specified starts applying. - Config: - Move coordinator exclusive configuration from the node config to the coordinator config - Coordinator: - Reorganize the code towards having the goroutines started and stopped from the coordinator itself instead of the node. - Remove all stop and stopped channels, and use context.Context and sync.WaitGroup instead. - Remove BatchInfo setters and assing variables directly - In ServerProof and ServerProofPool use context instead stop channel. - Use message passing to notify the coordinator about sync updates and reorgs - Introduce the Pipeline, which can be started and stopped by the Coordinator - Introduce the TxManager, which manages ethereum transactions (the TxManager is also in charge of making the forge call to the rollup smart contract). The TxManager keeps ethereum transactions and: 1. Waits for the transaction to be accepted 2. Waits for the transaction to be confirmed for N blocks - In forge logic, first prepare a batch and then wait for an available server proof to have all work ready once the proof server is ready. - Remove the `isForgeSequence` method which was querying the smart contract, and instead use notifications sent by the Synchronizer to figure out if it's forging time. - Update test (which is a minimal test to manually see if the coordinator starts) - HistoryDB: - Add method to get the number of batches in a slot (used to detect when a slot has passed the bid winner forging deadline) - Add method to get the best bid and associated coordinator of a slot (used to detect the forgerAddress that can forge the slot) - General: - Rename some instances of `currentBlock` to `lastBlock` to be more clear. - Node: - Connect the API to the node and call the methods to update cached state when the sync advances blocks. - Call methods to update Coordinator state when the sync advances blocks and finds reorgs. - Synchronizer: - Add Auction field in the Stats, which contain the current slot with info about highest bidder and other related info required to know who can forge in the current block. - Better organization of cached state: - On Sync, update the internal cached state - On Init or Reorg, load the state from HistoryDB into the internal cached state.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator to work better under real net - cli / node - Update handler of SIGINT so that after 3 SIGINTs, the process terminates unconditionally - coordinator - Store stats without pointer - In all functions that send a variable via channel, check for context done to avoid deadlock (due to no process reading from the channel, which has no queue) when the node is stopped. - Abstract `canForge` so that it can be used outside of the `Coordinator` - In `canForge` check the blockNumber in current and next slot. - Update tests due to smart contract changes in slot handling, and minimum bid defaults - TxManager - Add consts, vars and stats to allow evaluating `canForge` - Add `canForge` method (not used yet) - Store batch and nonces status (last success and last pending) - Track nonces internally instead of relying on the ethereum node (this is required to work with ganache when there are pending txs) - Handle the (common) case of the receipt not being found after the tx is sent. - Don't start the main loop until we get an initial messae fo the stats and vars (so that in the loop the stats and vars are set to synchronizer values) - When a tx fails, check and discard all the failed transactions before sending the message to stop the pipeline. This will avoid sending consecutive messages of stop the pipeline when multiple txs are detected to be failed consecutively. Also, future txs of the same pipeline after a discarded txs are discarded, and their nonces reused. - Robust handling of nonces: - If geth returns nonce is too low, increase it - If geth returns nonce too hight, decrease it - If geth returns underpriced, increase gas price - If geth returns replace underpriced, increase gas price - Add support for resending transactions after a timeout - Store `BatchInfos` in a queue - Pipeline - When an error is found, stop forging batches and send a message to the coordinator to stop the pipeline with information of the failed batch number so that in a restart, non-failed batches are not repated. - When doing a reset of the stateDB, if possible reset from the local checkpoint instead of resetting from the synchronizer. This allows resetting from a batch that is valid but not yet sent / synced. - Every time a pipeline is started, assign it a number from a counter. This allows the TxManager to ignore batches from stopped pipelines, via a message sent by the coordinator. - Avoid forging when we haven't reached the rollup genesis block number. - Add config parameter `StartSlotBlocksDelay`: StartSlotBlocksDelay is the number of blocks of delay to wait before starting the pipeline when we reach a slot in which we can forge. - When detecting a reorg, only reset the pipeline if the batch from which the pipeline started changed and wasn't sent by us. - Add config parameter `ScheduleBatchBlocksAheadCheck`: ScheduleBatchBlocksAheadCheck is the number of blocks ahead in which the forger address is checked to be allowed to forge (apart from checking the next block), used to decide when to stop scheduling new batches (by stopping the pipeline). For example, if we are at block 10 and ScheduleBatchBlocksAheadCheck is 5, eventhough at block 11 we canForge, the pipeline will be stopped if we can't forge at block 15. This value should be the expected number of blocks it takes between scheduling a batch and having it mined. - Add config parameter `SendBatchBlocksMarginCheck`: SendBatchBlocksMarginCheck is the number of margin blocks ahead in which the coordinator is also checked to be allowed to forge, apart from the next block; used to decide when to stop sending batches to the smart contract. For example, if we are at block 10 and SendBatchBlocksMarginCheck is 5, eventhough at block 11 we canForge, the batch will be discarded if we can't forge at block 15. - Add config parameter `TxResendTimeout`: TxResendTimeout is the timeout after which a non-mined ethereum transaction will be resent (reusing the nonce) with a newly calculated gas price - Add config parameter `MaxGasPrice`: MaxGasPrice is the maximum gas price allowed for ethereum transactions - Add config parameter `NoReuseNonce`: NoReuseNonce disables reusing nonces of pending transactions for new replacement transactions. This is useful for testing with Ganache. - Extend BatchInfo with more useful information for debugging - eth / ethereum client - Add necessary methods to create the auth object for transactions manually so that we can set the nonce, gas price, gas limit, etc manually - Update `RollupForgeBatch` to take an auth object as input (so that the coordinator can set parameters manually) - synchronizer - In stats, add `NextSlot` - In stats, store full last batch instead of just last batch number - Instead of calculating a nextSlot from scratch every time, update the current struct (only updating the forger info if we are Synced) - Afer every processed batch, check that the calculated StateDB MTRoot matches the StateRoot found in the forgeBatch event.
3 years ago
Fix eth events query and sync inconsistent state - kvdb - Fix path in Last when doing `setNew` - Only close if db != nil, and after closing, always set db to nil - This will avoid a panic in the case where the db is closed but there's an error soon after, and a future call tries to close again. This is because pebble.Close() will panic if the db is already closed. - Avoid calling pebble methods when a the Storage interface already implements that method (like Close). - statedb - In test, avoid calling KVDB method if the same method is available for the StateDB (like MakeCheckpoint, CurrentBatch). - eth - In *EventByBlock methods, take blockHash as input argument and use it when querying the event logs. Previously the blockHash was only taken from the logs results *only if* there was any log. This caused the following issue: if there was no logs, it was not possible to know if the result was from the expected block or an uncle block! By querying logs by blockHash we make sure that even if there are no logs, they are from the right block. - Note that now the function can either be called with a blockNum or blockHash, but not both at the same time. - sync - If there's an error during call to Sync call resetState, which internally resets the stateDB to avoid stale checkpoints (and a corresponding invalid increase in the StateDB batchNum). - During a Sync, after very batch processed, make sure that the StateDB currentBatch corresponds to the batchNum in the smart contract log/event.
3 years ago
Update coordinator, call all api update functions - Common: - Rename Block.EthBlockNum to Block.Num to avoid unneeded repetition - API: - Add UpdateNetworkInfoBlock to update just block information, to be used when the node is not yet synchronized - Node: - Call API.UpdateMetrics and UpdateRecommendedFee in a loop, with configurable time intervals - Synchronizer: - When mapping events by TxHash, use an array to support the possibility of multiple calls of the same function happening in the same transaction (for example, a smart contract in a single transaction could call withdraw with delay twice, which would generate 2 withdraw events, and 2 deposit events). - In Stats, keep entire LastBlock instead of just the blockNum - In Stats, add lastL1BatchBlock - Test Stats and SCVars - Coordinator: - Enable writing the BatchInfo in every step of the pipeline to disk (with JSON text files) for debugging purposes. - Move the Pipeline functionality from the Coordinator to its own struct (Pipeline) - Implement shouldL1lL2Batch - In TxManager, implement logic to perform several attempts when doing ethereum node RPC calls before considering the error. (Both for calls to forgeBatch and transaction receipt) - In TxManager, reorganize the flow and note the specific points in which actions are made when err != nil - HistoryDB: - Implement GetLastL1BatchBlockNum: returns the blockNum of the latest forged l1Batch, to help the coordinator decide when to forge an L1Batch. - EthereumClient and test.Client: - Update EthBlockByNumber to return the last block when the passed number is -1.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
Update missing parts, improve til, and more - Node - Updated configuration to initialize the interface to all the smart contracts - Common - Moved BlockData and BatchData types to common so that they can be shared among: historydb, til and synchronizer - Remove hash.go (it was never used) - Remove slot.go (it was never used) - Remove smartcontractparams.go (it was never used, and appropriate structs are defined in `eth/`) - Comment state / status method until requirements of this method are properly defined, and move it to Synchronizer - Synchronizer - Simplify `Sync` routine to only sync one block per call, and return useful information. - Use BlockData and BatchData from common - Check that events belong to the expected block hash - In L1Batch, query L1UserTxs from HistoryDB - Fill ERC20 token information - Test AddTokens with test.Client - HistryDB - Use BlockData and BatchData from common - Add `GetAllTokens` method - Uncomment and update GetL1UserTxs (with corresponding tests) - Til - Rename all instances of RegisterToken to AddToken (to follow the smart contract implementation naming) - Use BlockData and BatchData from common - Move testL1CoordinatorTxs and testL2Txs to a separate struct from BatchData in Context - Start Context with BatchNum = 1 (which the protocol defines to be the first batchNum) - In every Batch, set StateRoot and ExitRoot to a non-nil big.Int (zero). - In all L1Txs, if LoadAmount is not used, set it to 0; if Amount is not used, set it to 0; so that no *big.Int is nil. - In L1UserTx, don't set BatchNum, because when L1UserTxs are created and obtained by the synchronizer, the BatchNum is not known yet (it's a synchronizer job to set it) - In L1UserTxs, set `UserOrigin` and set `ToForgeL1TxsNum`.
4 years ago
  1. package synchronizer
  2. import (
  3. "context"
  4. "database/sql"
  5. "fmt"
  6. "math/big"
  7. "sync"
  8. "time"
  9. "github.com/ethereum/go-ethereum"
  10. "github.com/hermeznetwork/hermez-node/common"
  11. "github.com/hermeznetwork/hermez-node/db/historydb"
  12. "github.com/hermeznetwork/hermez-node/db/l2db"
  13. "github.com/hermeznetwork/hermez-node/db/statedb"
  14. "github.com/hermeznetwork/hermez-node/eth"
  15. "github.com/hermeznetwork/hermez-node/log"
  16. "github.com/hermeznetwork/hermez-node/metric"
  17. "github.com/hermeznetwork/hermez-node/txprocessor"
  18. "github.com/hermeznetwork/tracerr"
  19. )
  20. const (
  21. // errStrUnknownBlock is the string returned by geth when querying an
  22. // unknown block
  23. errStrUnknownBlock = "unknown block"
  24. )
  25. var (
  26. // ErrUnknownBlock is the error returned by the Synchronizer when a
  27. // block is queried by hash but the ethereum node doesn't find it due
  28. // to it being discarded from a reorg.
  29. ErrUnknownBlock = fmt.Errorf("unknown block")
  30. )
  31. // Stats of the synchronizer
  32. type Stats struct {
  33. Eth struct {
  34. RefreshPeriod time.Duration
  35. Updated time.Time
  36. FirstBlockNum int64
  37. LastBlock common.Block
  38. LastBatchNum int64
  39. }
  40. Sync struct {
  41. Updated time.Time
  42. LastBlock common.Block
  43. LastBatch common.Batch
  44. // LastL1BatchBlock is the last ethereum block in which an
  45. // l1Batch was forged
  46. LastL1BatchBlock int64
  47. LastForgeL1TxsNum int64
  48. Auction struct {
  49. CurrentSlot common.Slot
  50. NextSlot common.Slot
  51. }
  52. }
  53. }
  54. // Synced returns true if the Synchronizer is up to date with the last ethereum block
  55. func (s *Stats) Synced() bool {
  56. return s.Eth.LastBlock.Num == s.Sync.LastBlock.Num
  57. }
  58. // TODO(Edu): Consider removing all the mutexes from StatsHolder, make
  59. // Synchronizer.Stats not thread-safe, don't pass the synchronizer to the
  60. // debugAPI, and have a copy of the Stats in the DebugAPI that the node passes
  61. // when the Sync updates.
  62. // StatsHolder stores stats and that allows reading and writing them
  63. // concurrently
  64. type StatsHolder struct {
  65. Stats
  66. rw sync.RWMutex
  67. }
  68. // NewStatsHolder creates a new StatsHolder
  69. func NewStatsHolder(firstBlockNum int64, refreshPeriod time.Duration) *StatsHolder {
  70. stats := Stats{}
  71. stats.Eth.RefreshPeriod = refreshPeriod
  72. stats.Eth.FirstBlockNum = firstBlockNum
  73. stats.Sync.LastForgeL1TxsNum = -1
  74. return &StatsHolder{Stats: stats}
  75. }
  76. // UpdateCurrentNextSlot updates the auction stats
  77. func (s *StatsHolder) UpdateCurrentNextSlot(current *common.Slot, next *common.Slot) {
  78. s.rw.Lock()
  79. s.Sync.Auction.CurrentSlot = *current
  80. s.Sync.Auction.NextSlot = *next
  81. s.rw.Unlock()
  82. }
  83. // UpdateSync updates the synchronizer stats
  84. func (s *StatsHolder) UpdateSync(lastBlock *common.Block, lastBatch *common.Batch,
  85. lastL1BatchBlock *int64, lastForgeL1TxsNum *int64) {
  86. now := time.Now()
  87. s.rw.Lock()
  88. s.Sync.LastBlock = *lastBlock
  89. if lastBatch != nil {
  90. s.Sync.LastBatch = *lastBatch
  91. }
  92. if lastL1BatchBlock != nil {
  93. s.Sync.LastL1BatchBlock = *lastL1BatchBlock
  94. s.Sync.LastForgeL1TxsNum = *lastForgeL1TxsNum
  95. }
  96. s.Sync.Updated = now
  97. s.rw.Unlock()
  98. }
  99. // UpdateEth updates the ethereum stats, only if the previous stats expired
  100. func (s *StatsHolder) UpdateEth(ethClient eth.ClientInterface) error {
  101. now := time.Now()
  102. s.rw.RLock()
  103. elapsed := now.Sub(s.Eth.Updated)
  104. s.rw.RUnlock()
  105. if elapsed < s.Eth.RefreshPeriod {
  106. return nil
  107. }
  108. lastBlock, err := ethClient.EthBlockByNumber(context.TODO(), -1)
  109. if err != nil {
  110. return tracerr.Wrap(fmt.Errorf("EthBlockByNumber: %w", err))
  111. }
  112. lastBatchNum, err := ethClient.RollupLastForgedBatch()
  113. if err != nil {
  114. return tracerr.Wrap(fmt.Errorf("RollupLastForgedBatch: %w", err))
  115. }
  116. s.rw.Lock()
  117. s.Eth.Updated = now
  118. s.Eth.LastBlock = *lastBlock
  119. s.Eth.LastBatchNum = lastBatchNum
  120. s.rw.Unlock()
  121. return nil
  122. }
  123. // CopyStats returns a copy of the inner Stats
  124. func (s *StatsHolder) CopyStats() *Stats {
  125. s.rw.RLock()
  126. sCopy := s.Stats
  127. if s.Sync.Auction.CurrentSlot.BidValue != nil {
  128. sCopy.Sync.Auction.CurrentSlot.BidValue =
  129. common.CopyBigInt(s.Sync.Auction.CurrentSlot.BidValue)
  130. }
  131. if s.Sync.Auction.CurrentSlot.DefaultSlotBid != nil {
  132. sCopy.Sync.Auction.CurrentSlot.DefaultSlotBid =
  133. common.CopyBigInt(s.Sync.Auction.CurrentSlot.DefaultSlotBid)
  134. }
  135. if s.Sync.Auction.NextSlot.BidValue != nil {
  136. sCopy.Sync.Auction.NextSlot.BidValue =
  137. common.CopyBigInt(s.Sync.Auction.NextSlot.BidValue)
  138. }
  139. if s.Sync.Auction.NextSlot.DefaultSlotBid != nil {
  140. sCopy.Sync.Auction.NextSlot.DefaultSlotBid =
  141. common.CopyBigInt(s.Sync.Auction.NextSlot.DefaultSlotBid)
  142. }
  143. if s.Sync.LastBatch.StateRoot != nil {
  144. sCopy.Sync.LastBatch.StateRoot =
  145. common.CopyBigInt(s.Sync.LastBatch.StateRoot)
  146. }
  147. s.rw.RUnlock()
  148. return &sCopy
  149. }
  150. func (s *StatsHolder) blocksPerc() float64 {
  151. syncLastBlockNum := s.Sync.LastBlock.Num
  152. if s.Sync.LastBlock.Num == 0 {
  153. syncLastBlockNum = s.Eth.FirstBlockNum - 1
  154. }
  155. return float64(syncLastBlockNum-(s.Eth.FirstBlockNum-1)) * 100.0 /
  156. float64(s.Eth.LastBlock.Num-(s.Eth.FirstBlockNum-1))
  157. }
  158. func (s *StatsHolder) batchesPerc(batchNum common.BatchNum) float64 {
  159. return float64(batchNum) * 100.0 /
  160. float64(s.Eth.LastBatchNum)
  161. }
  162. // StartBlockNums sets the first block used to start tracking the smart
  163. // contracts
  164. type StartBlockNums struct {
  165. Rollup int64
  166. Auction int64
  167. WDelayer int64
  168. }
  169. // Config is the Synchronizer configuration
  170. type Config struct {
  171. StatsRefreshPeriod time.Duration
  172. ChainID uint16
  173. }
  174. // Synchronizer implements the Synchronizer type
  175. type Synchronizer struct {
  176. ethClient eth.ClientInterface
  177. consts common.SCConsts
  178. historyDB *historydb.HistoryDB
  179. l2DB *l2db.L2DB
  180. stateDB *statedb.StateDB
  181. cfg Config
  182. initVars common.SCVariables
  183. startBlockNum int64
  184. vars common.SCVariables
  185. stats *StatsHolder
  186. resetStateFailed bool
  187. }
  188. // NewSynchronizer creates a new Synchronizer
  189. func NewSynchronizer(ethClient eth.ClientInterface, historyDB *historydb.HistoryDB,
  190. l2DB *l2db.L2DB, stateDB *statedb.StateDB, cfg Config) (*Synchronizer, error) {
  191. auctionConstants, err := ethClient.AuctionConstants()
  192. if err != nil {
  193. return nil, tracerr.Wrap(fmt.Errorf("NewSynchronizer ethClient.AuctionConstants(): %w",
  194. err))
  195. }
  196. rollupConstants, err := ethClient.RollupConstants()
  197. if err != nil {
  198. return nil, tracerr.Wrap(fmt.Errorf("NewSynchronizer ethClient.RollupConstants(): %w",
  199. err))
  200. }
  201. wDelayerConstants, err := ethClient.WDelayerConstants()
  202. if err != nil {
  203. return nil, tracerr.Wrap(fmt.Errorf("NewSynchronizer ethClient.WDelayerConstants(): %w",
  204. err))
  205. }
  206. consts := common.SCConsts{
  207. Rollup: *rollupConstants,
  208. Auction: *auctionConstants,
  209. WDelayer: *wDelayerConstants,
  210. }
  211. initVars, startBlockNums, err := getInitialVariables(ethClient, &consts)
  212. if err != nil {
  213. return nil, tracerr.Wrap(err)
  214. }
  215. log.Infow("Synchronizer syncing from smart contract blocks",
  216. "rollup", startBlockNums.Rollup,
  217. "auction", startBlockNums.Auction,
  218. "wdelayer", startBlockNums.WDelayer,
  219. )
  220. // Set startBlockNum to the minimum between Auction, Rollup and
  221. // WDelayer StartBlockNum
  222. startBlockNum := startBlockNums.Auction
  223. if startBlockNums.Rollup < startBlockNum {
  224. startBlockNum = startBlockNums.Rollup
  225. }
  226. if startBlockNums.WDelayer < startBlockNum {
  227. startBlockNum = startBlockNums.WDelayer
  228. }
  229. stats := NewStatsHolder(startBlockNum, cfg.StatsRefreshPeriod)
  230. s := &Synchronizer{
  231. ethClient: ethClient,
  232. consts: consts,
  233. historyDB: historyDB,
  234. l2DB: l2DB,
  235. stateDB: stateDB,
  236. cfg: cfg,
  237. initVars: *initVars,
  238. startBlockNum: startBlockNum,
  239. stats: stats,
  240. }
  241. return s, s.init()
  242. }
  243. // StateDB returns the inner StateDB
  244. func (s *Synchronizer) StateDB() *statedb.StateDB {
  245. return s.stateDB
  246. }
  247. // Stats returns a copy of the Synchronizer Stats. It is safe to call Stats()
  248. // during a Sync call
  249. func (s *Synchronizer) Stats() *Stats {
  250. return s.stats.CopyStats()
  251. }
  252. // AuctionConstants returns the AuctionConstants read from the smart contract
  253. func (s *Synchronizer) AuctionConstants() *common.AuctionConstants {
  254. return &s.consts.Auction
  255. }
  256. // RollupConstants returns the RollupConstants read from the smart contract
  257. func (s *Synchronizer) RollupConstants() *common.RollupConstants {
  258. return &s.consts.Rollup
  259. }
  260. // WDelayerConstants returns the WDelayerConstants read from the smart contract
  261. func (s *Synchronizer) WDelayerConstants() *common.WDelayerConstants {
  262. return &s.consts.WDelayer
  263. }
  264. // SCVars returns a copy of the Smart Contract Variables
  265. func (s *Synchronizer) SCVars() *common.SCVariables {
  266. return &common.SCVariables{
  267. Rollup: *s.vars.Rollup.Copy(),
  268. Auction: *s.vars.Auction.Copy(),
  269. WDelayer: *s.vars.WDelayer.Copy(),
  270. }
  271. }
  272. // setSlotCoordinator queries the highest bidder of a slot in the HistoryDB to
  273. // determine the coordinator that can bid in a slot
  274. func (s *Synchronizer) setSlotCoordinator(slot *common.Slot) error {
  275. bidCoord, err := s.historyDB.GetBestBidCoordinator(slot.SlotNum)
  276. if err != nil && tracerr.Unwrap(err) != sql.ErrNoRows {
  277. return tracerr.Wrap(err)
  278. }
  279. if tracerr.Unwrap(err) == sql.ErrNoRows {
  280. slot.BootCoord = true
  281. slot.Forger = s.vars.Auction.BootCoordinator
  282. slot.URL = s.vars.Auction.BootCoordinatorURL
  283. } else if err == nil {
  284. slot.BidValue = bidCoord.BidValue
  285. slot.DefaultSlotBid = bidCoord.DefaultSlotSetBid[slot.SlotNum%6]
  286. // Only if the highest bid value is greater/equal than
  287. // the default slot bid, the bidder is the winner of
  288. // the slot. Otherwise the boot coordinator is the
  289. // winner.
  290. if slot.BidValue.Cmp(slot.DefaultSlotBid) >= 0 {
  291. slot.Bidder = bidCoord.Bidder
  292. slot.Forger = bidCoord.Forger
  293. slot.URL = bidCoord.URL
  294. } else {
  295. slot.BootCoord = true
  296. slot.Forger = s.vars.Auction.BootCoordinator
  297. slot.URL = s.vars.Auction.BootCoordinatorURL
  298. }
  299. }
  300. return nil
  301. }
  302. // updateCurrentSlot updates the slot with information of the current slot.
  303. // The information about which coordinator is allowed to forge is only updated
  304. // when we are Synced.
  305. // hasBatch is true when the last synced block contained at least one batch.
  306. func (s *Synchronizer) updateCurrentSlot(slot *common.Slot, reset bool, hasBatch bool) error {
  307. // We want the next block because the current one is already mined
  308. blockNum := s.stats.Sync.LastBlock.Num + 1
  309. slotNum := s.consts.Auction.SlotNum(blockNum)
  310. syncLastBlockNum := s.stats.Sync.LastBlock.Num
  311. if reset {
  312. // Using this query only to know if there
  313. dbFirstBatchBlockNum, err := s.historyDB.GetFirstBatchBlockNumBySlot(slotNum)
  314. if err != nil && tracerr.Unwrap(err) != sql.ErrNoRows {
  315. return tracerr.Wrap(fmt.Errorf("historyDB.GetFirstBatchBySlot: %w", err))
  316. } else if tracerr.Unwrap(err) == sql.ErrNoRows {
  317. hasBatch = false
  318. } else {
  319. hasBatch = true
  320. syncLastBlockNum = dbFirstBatchBlockNum
  321. }
  322. slot.ForgerCommitment = false
  323. } else if slotNum > slot.SlotNum {
  324. // We are in a new slotNum, start from default values
  325. slot.ForgerCommitment = false
  326. }
  327. slot.SlotNum = slotNum
  328. slot.StartBlock, slot.EndBlock = s.consts.Auction.SlotBlocks(slot.SlotNum)
  329. if hasBatch && s.consts.Auction.RelativeBlock(syncLastBlockNum) < int64(s.vars.Auction.SlotDeadline) {
  330. slot.ForgerCommitment = true
  331. }
  332. // If Synced, update the current coordinator
  333. if s.stats.Synced() && blockNum >= s.consts.Auction.GenesisBlockNum {
  334. if err := s.setSlotCoordinator(slot); err != nil {
  335. return tracerr.Wrap(err)
  336. }
  337. // TODO: Remove this SANITY CHECK once this code is tested enough
  338. // BEGIN SANITY CHECK
  339. canForge, err := s.ethClient.AuctionCanForge(slot.Forger, blockNum)
  340. if err != nil {
  341. return tracerr.Wrap(fmt.Errorf("AuctionCanForge: %w", err))
  342. }
  343. if !canForge {
  344. return tracerr.Wrap(fmt.Errorf("Synchronized value of forger address for closed slot "+
  345. "differs from smart contract: %+v", slot))
  346. }
  347. // END SANITY CHECK
  348. }
  349. return nil
  350. }
  351. // updateNextSlot updates the slot with information of the next slot.
  352. // The information about which coordinator is allowed to forge is only updated
  353. // when we are Synced.
  354. func (s *Synchronizer) updateNextSlot(slot *common.Slot) error {
  355. // We want the next block because the current one is already mined
  356. blockNum := s.stats.Sync.LastBlock.Num + 1
  357. slotNum := s.consts.Auction.SlotNum(blockNum) + 1
  358. slot.SlotNum = slotNum
  359. slot.ForgerCommitment = false
  360. slot.StartBlock, slot.EndBlock = s.consts.Auction.SlotBlocks(slot.SlotNum)
  361. // If Synced, update the current coordinator
  362. if s.stats.Synced() && blockNum >= s.consts.Auction.GenesisBlockNum {
  363. if err := s.setSlotCoordinator(slot); err != nil {
  364. return tracerr.Wrap(err)
  365. }
  366. // TODO: Remove this SANITY CHECK once this code is tested enough
  367. // BEGIN SANITY CHECK
  368. canForge, err := s.ethClient.AuctionCanForge(slot.Forger, slot.StartBlock)
  369. if err != nil {
  370. return tracerr.Wrap(fmt.Errorf("AuctionCanForge: %w", err))
  371. }
  372. if !canForge {
  373. return tracerr.Wrap(fmt.Errorf("Synchronized value of forger address for closed slot "+
  374. "differs from smart contract: %+v", slot))
  375. }
  376. // END SANITY CHECK
  377. }
  378. return nil
  379. }
  380. // updateCurrentNextSlotIfSync updates the current and next slot. Information
  381. // about forger address that is allowed to forge is only updated if we are
  382. // Synced.
  383. func (s *Synchronizer) updateCurrentNextSlotIfSync(reset bool, hasBatch bool) error {
  384. current := s.stats.Sync.Auction.CurrentSlot
  385. next := s.stats.Sync.Auction.NextSlot
  386. if err := s.updateCurrentSlot(&current, reset, hasBatch); err != nil {
  387. return tracerr.Wrap(err)
  388. }
  389. if err := s.updateNextSlot(&next); err != nil {
  390. return tracerr.Wrap(err)
  391. }
  392. s.stats.UpdateCurrentNextSlot(&current, &next)
  393. return nil
  394. }
  395. func (s *Synchronizer) init() error {
  396. // Update stats parameters so that they have valid values before the
  397. // first Sync call
  398. if err := s.stats.UpdateEth(s.ethClient); err != nil {
  399. return tracerr.Wrap(err)
  400. }
  401. lastBlock := &common.Block{}
  402. lastSavedBlock, err := s.historyDB.GetLastBlock()
  403. // `s.historyDB.GetLastBlock()` will never return `sql.ErrNoRows`
  404. // because we always have the default block 0 in the DB
  405. if err != nil {
  406. return tracerr.Wrap(err)
  407. }
  408. // If we only have the default block 0,
  409. // make sure that the stateDB is clean
  410. if lastSavedBlock.Num == 0 {
  411. if err := s.stateDB.Reset(0); err != nil {
  412. return tracerr.Wrap(err)
  413. }
  414. } else {
  415. lastBlock = lastSavedBlock
  416. }
  417. if err := s.resetState(lastBlock); err != nil {
  418. s.resetStateFailed = true
  419. return tracerr.Wrap(err)
  420. }
  421. s.resetStateFailed = false
  422. log.Infow("Sync init block",
  423. "syncLastBlock", s.stats.Sync.LastBlock,
  424. "syncBlocksPerc", s.stats.blocksPerc(),
  425. "ethFirstBlockNum", s.stats.Eth.FirstBlockNum,
  426. "ethLastBlock", s.stats.Eth.LastBlock,
  427. )
  428. log.Infow("Sync init batch",
  429. "syncLastBatch", s.stats.Sync.LastBatch.BatchNum,
  430. "syncBatchesPerc", s.stats.batchesPerc(s.stats.Sync.LastBatch.BatchNum),
  431. "ethLastBatch", s.stats.Eth.LastBatchNum,
  432. )
  433. return nil
  434. }
  435. func (s *Synchronizer) resetIntermediateState() error {
  436. lastBlock, err := s.historyDB.GetLastBlock()
  437. if tracerr.Unwrap(err) == sql.ErrNoRows {
  438. lastBlock = &common.Block{}
  439. } else if err != nil {
  440. return tracerr.Wrap(fmt.Errorf("historyDB.GetLastBlock: %w", err))
  441. }
  442. if err := s.resetState(lastBlock); err != nil {
  443. s.resetStateFailed = true
  444. return tracerr.Wrap(fmt.Errorf("resetState at block %v: %w", lastBlock.Num, err))
  445. }
  446. s.resetStateFailed = false
  447. return nil
  448. }
  449. // Sync attempts to synchronize an ethereum block starting from lastSavedBlock.
  450. // If lastSavedBlock is nil, the lastSavedBlock value is obtained from de DB.
  451. // If a block is synced, it will be returned and also stored in the DB. If a
  452. // reorg is detected, the number of discarded blocks will be returned and no
  453. // synchronization will be made.
  454. // TODO: Be smart about locking: only lock during the read/write operations
  455. func (s *Synchronizer) Sync(ctx context.Context,
  456. lastSavedBlock *common.Block) (blockData *common.BlockData, discarded *int64, err error) {
  457. if s.resetStateFailed {
  458. if err := s.resetIntermediateState(); err != nil {
  459. return nil, nil, tracerr.Wrap(err)
  460. }
  461. }
  462. var nextBlockNum int64 // next block number to sync
  463. if lastSavedBlock == nil {
  464. // Get lastSavedBlock from History DB
  465. lastSavedBlock, err = s.historyDB.GetLastBlock()
  466. if err != nil && tracerr.Unwrap(err) != sql.ErrNoRows {
  467. return nil, nil, tracerr.Wrap(err)
  468. }
  469. // If we don't have any stored block, we must do a full sync
  470. // starting from the startBlockNum
  471. if tracerr.Unwrap(err) == sql.ErrNoRows || lastSavedBlock.Num == 0 {
  472. nextBlockNum = s.startBlockNum
  473. lastSavedBlock = nil
  474. }
  475. }
  476. if lastSavedBlock != nil {
  477. nextBlockNum = lastSavedBlock.Num + 1
  478. if lastSavedBlock.Num < s.startBlockNum {
  479. return nil, nil, tracerr.Wrap(
  480. fmt.Errorf("lastSavedBlock (%v) < startBlockNum (%v)",
  481. lastSavedBlock.Num, s.startBlockNum))
  482. }
  483. }
  484. ethBlock, err := s.ethClient.EthBlockByNumber(ctx, nextBlockNum)
  485. if tracerr.Unwrap(err) == ethereum.NotFound {
  486. return nil, nil, nil
  487. } else if err != nil {
  488. return nil, nil, tracerr.Wrap(fmt.Errorf("EthBlockByNumber: %w", err))
  489. }
  490. log.Debugf("ethBlock: num: %v, parent: %v, hash: %v",
  491. ethBlock.Num, ethBlock.ParentHash.String(), ethBlock.Hash.String())
  492. if err := s.stats.UpdateEth(s.ethClient); err != nil {
  493. return nil, nil, tracerr.Wrap(err)
  494. }
  495. log.Debugw("Syncing...",
  496. "block", nextBlockNum,
  497. "ethLastBlock", s.stats.Eth.LastBlock,
  498. )
  499. // Check that the obtained ethBlock.ParentHash == prevEthBlock.Hash; if not, reorg!
  500. if lastSavedBlock != nil {
  501. if lastSavedBlock.Hash != ethBlock.ParentHash {
  502. // Reorg detected
  503. log.Debugw("Reorg Detected",
  504. "blockNum", ethBlock.Num,
  505. "block.parent(got)", ethBlock.ParentHash, "parent.hash(exp)", lastSavedBlock.Hash)
  506. lastDBBlockNum, err := s.reorg(lastSavedBlock)
  507. if err != nil {
  508. return nil, nil, tracerr.Wrap(err)
  509. }
  510. discarded := lastSavedBlock.Num - lastDBBlockNum
  511. metric.Reorgs.Inc()
  512. return nil, &discarded, nil
  513. }
  514. }
  515. defer func() {
  516. // If there was an error during sync, reset to the last block
  517. // in the historyDB because the historyDB is written last in
  518. // the Sync method and is the source of consistency. This
  519. // allows resetting the stateDB in the case a batch was
  520. // processed but the historyDB block was not committed due to an
  521. // error.
  522. if err != nil {
  523. if err2 := s.resetIntermediateState(); err2 != nil {
  524. log.Errorw("sync revert", "err", err2)
  525. }
  526. }
  527. }()
  528. // Get data from the rollup contract
  529. rollupData, err := s.rollupSync(ethBlock)
  530. if err != nil {
  531. return nil, nil, tracerr.Wrap(err)
  532. }
  533. // Get data from the auction contract
  534. auctionData, err := s.auctionSync(ethBlock)
  535. if err != nil {
  536. return nil, nil, tracerr.Wrap(err)
  537. }
  538. // Get data from the WithdrawalDelayer contract
  539. wDelayerData, err := s.wdelayerSync(ethBlock)
  540. if err != nil {
  541. return nil, nil, tracerr.Wrap(err)
  542. }
  543. for i := range rollupData.Withdrawals {
  544. withdrawal := &rollupData.Withdrawals[i]
  545. if !withdrawal.InstantWithdraw {
  546. wDelayerTransfers := wDelayerData.DepositsByTxHash[withdrawal.TxHash]
  547. if len(wDelayerTransfers) == 0 {
  548. return nil, nil, tracerr.Wrap(fmt.Errorf("WDelayer deposit corresponding to " +
  549. "non-instant rollup withdrawal not found"))
  550. }
  551. // Pop the first wDelayerTransfer to consume them in chronological order
  552. wDelayerTransfer := wDelayerTransfers[0]
  553. wDelayerData.DepositsByTxHash[withdrawal.TxHash] =
  554. wDelayerData.DepositsByTxHash[withdrawal.TxHash][1:]
  555. withdrawal.Owner = wDelayerTransfer.Owner
  556. withdrawal.Token = wDelayerTransfer.Token
  557. }
  558. }
  559. // Group all the block data into the structs to save into HistoryDB
  560. blockData = &common.BlockData{
  561. Block: *ethBlock,
  562. Rollup: *rollupData,
  563. Auction: *auctionData,
  564. WDelayer: *wDelayerData,
  565. }
  566. err = s.historyDB.AddBlockSCData(blockData)
  567. if err != nil {
  568. return nil, nil, tracerr.Wrap(err)
  569. }
  570. batchesLen := len(rollupData.Batches)
  571. if batchesLen == 0 {
  572. s.stats.UpdateSync(ethBlock, nil, nil, nil)
  573. } else {
  574. var lastL1BatchBlock *int64
  575. var lastForgeL1TxsNum *int64
  576. for _, batchData := range rollupData.Batches {
  577. if batchData.L1Batch {
  578. lastL1BatchBlock = &batchData.Batch.EthBlockNum
  579. lastForgeL1TxsNum = batchData.Batch.ForgeL1TxsNum
  580. }
  581. }
  582. s.stats.UpdateSync(ethBlock,
  583. &rollupData.Batches[batchesLen-1].Batch,
  584. lastL1BatchBlock, lastForgeL1TxsNum)
  585. }
  586. hasBatch := false
  587. if len(rollupData.Batches) > 0 {
  588. hasBatch = true
  589. }
  590. if err = s.updateCurrentNextSlotIfSync(false, hasBatch); err != nil {
  591. return nil, nil, tracerr.Wrap(err)
  592. }
  593. for _, batchData := range rollupData.Batches {
  594. metric.LastBatchNum.Set(float64(batchData.Batch.BatchNum))
  595. metric.EthLastBatchNum.Set(float64(s.stats.Eth.LastBatchNum))
  596. log.Debugw("Synced batch",
  597. "syncLastBatch", batchData.Batch.BatchNum,
  598. "syncBatchesPerc", s.stats.batchesPerc(batchData.Batch.BatchNum),
  599. "ethLastBatch", s.stats.Eth.LastBatchNum,
  600. )
  601. }
  602. metric.LastBlockNum.Set(float64(s.stats.Sync.LastBlock.Num))
  603. metric.EthLastBlockNum.Set(float64(s.stats.Eth.LastBlock.Num))
  604. log.Debugw("Synced block",
  605. "syncLastBlockNum", s.stats.Sync.LastBlock.Num,
  606. "syncBlocksPerc", s.stats.blocksPerc(),
  607. "ethLastBlockNum", s.stats.Eth.LastBlock.Num,
  608. )
  609. return blockData, nil, nil
  610. }
  611. // reorg manages a reorg, updating History and State DB as needed. Keeps
  612. // checking previous blocks from the HistoryDB against the blockchain until a
  613. // block hash match is found. All future blocks in the HistoryDB and
  614. // corresponding batches in StateBD are discarded. Returns the last valid
  615. // blockNum from the HistoryDB.
  616. func (s *Synchronizer) reorg(uncleBlock *common.Block) (int64, error) {
  617. blockNum := uncleBlock.Num
  618. var block *common.Block
  619. for blockNum >= s.startBlockNum {
  620. ethBlock, err := s.ethClient.EthBlockByNumber(context.Background(), blockNum)
  621. if err != nil {
  622. return 0, tracerr.Wrap(fmt.Errorf("ethClient.EthBlockByNumber: %w", err))
  623. }
  624. block, err = s.historyDB.GetBlock(blockNum)
  625. if err != nil {
  626. return 0, tracerr.Wrap(fmt.Errorf("historyDB.GetBlock: %w", err))
  627. }
  628. if block.Hash == ethBlock.Hash {
  629. log.Debugf("Found valid block: %v", blockNum)
  630. break
  631. }
  632. blockNum--
  633. }
  634. total := uncleBlock.Num - block.Num
  635. log.Debugw("Discarding blocks", "total", total, "from", uncleBlock.Num, "to", block.Num+1)
  636. // Set History DB and State DB to the correct state
  637. if err := s.historyDB.Reorg(block.Num); err != nil {
  638. return 0, tracerr.Wrap(err)
  639. }
  640. if err := s.resetState(block); err != nil {
  641. s.resetStateFailed = true
  642. return 0, tracerr.Wrap(err)
  643. }
  644. s.resetStateFailed = false
  645. return block.Num, nil
  646. }
  647. func getInitialVariables(ethClient eth.ClientInterface,
  648. consts *common.SCConsts) (*common.SCVariables, *StartBlockNums, error) {
  649. rollupInit, rollupInitBlock, err := ethClient.RollupEventInit(consts.Auction.GenesisBlockNum)
  650. if err != nil {
  651. return nil, nil, tracerr.Wrap(fmt.Errorf("RollupEventInit: %w", err))
  652. }
  653. auctionInit, auctionInitBlock, err := ethClient.AuctionEventInit(consts.Auction.GenesisBlockNum)
  654. if err != nil {
  655. return nil, nil, tracerr.Wrap(fmt.Errorf("AuctionEventInit: %w", err))
  656. }
  657. wDelayerInit, wDelayerInitBlock, err := ethClient.WDelayerEventInit(consts.Auction.GenesisBlockNum)
  658. if err != nil {
  659. return nil, nil, tracerr.Wrap(fmt.Errorf("WDelayerEventInit: %w", err))
  660. }
  661. rollupVars := rollupInit.RollupVariables()
  662. auctionVars := auctionInit.AuctionVariables(consts.Auction.InitialMinimalBidding)
  663. wDelayerVars := wDelayerInit.WDelayerVariables()
  664. return &common.SCVariables{
  665. Rollup: *rollupVars,
  666. Auction: *auctionVars,
  667. WDelayer: *wDelayerVars,
  668. }, &StartBlockNums{
  669. Rollup: rollupInitBlock,
  670. Auction: auctionInitBlock,
  671. WDelayer: wDelayerInitBlock,
  672. }, nil
  673. }
  674. func (s *Synchronizer) resetState(block *common.Block) error {
  675. rollup, auction, wDelayer, err := s.historyDB.GetSCVars()
  676. // If SCVars are not in the HistoryDB, this is probably the first run
  677. // of the Synchronizer: store the initial vars taken from config
  678. if tracerr.Unwrap(err) == sql.ErrNoRows {
  679. vars := s.initVars
  680. log.Info("Setting initial SCVars in HistoryDB")
  681. if err = s.historyDB.SetInitialSCVars(&vars.Rollup, &vars.Auction, &vars.WDelayer); err != nil {
  682. return tracerr.Wrap(fmt.Errorf("historyDB.SetInitialSCVars: %w", err))
  683. }
  684. s.vars.Rollup = *vars.Rollup.Copy()
  685. s.vars.Auction = *vars.Auction.Copy()
  686. s.vars.WDelayer = *vars.WDelayer.Copy()
  687. // Add initial boot coordinator to HistoryDB
  688. if err := s.historyDB.AddCoordinators([]common.Coordinator{{
  689. Forger: s.initVars.Auction.BootCoordinator,
  690. URL: s.initVars.Auction.BootCoordinatorURL,
  691. EthBlockNum: s.initVars.Auction.EthBlockNum,
  692. }}); err != nil {
  693. return tracerr.Wrap(err)
  694. }
  695. } else if err != nil {
  696. return tracerr.Wrap(err)
  697. } else {
  698. s.vars.Rollup = *rollup
  699. s.vars.Auction = *auction
  700. s.vars.WDelayer = *wDelayer
  701. }
  702. batch, err := s.historyDB.GetLastBatch()
  703. if err != nil && tracerr.Unwrap(err) != sql.ErrNoRows {
  704. return tracerr.Wrap(fmt.Errorf("historyDB.GetLastBatchNum: %w", err))
  705. }
  706. if tracerr.Unwrap(err) == sql.ErrNoRows {
  707. batch = &common.Batch{}
  708. }
  709. err = s.stateDB.Reset(batch.BatchNum)
  710. if err != nil {
  711. return tracerr.Wrap(fmt.Errorf("stateDB.Reset: %w", err))
  712. }
  713. lastL1BatchBlockNum, err := s.historyDB.GetLastL1BatchBlockNum()
  714. if err != nil && tracerr.Unwrap(err) != sql.ErrNoRows {
  715. return tracerr.Wrap(fmt.Errorf("historyDB.GetLastL1BatchBlockNum: %w", err))
  716. }
  717. if tracerr.Unwrap(err) == sql.ErrNoRows {
  718. lastL1BatchBlockNum = 0
  719. }
  720. lastForgeL1TxsNum, err := s.historyDB.GetLastL1TxsNum()
  721. if err != nil && tracerr.Unwrap(err) != sql.ErrNoRows {
  722. return tracerr.Wrap(fmt.Errorf("historyDB.GetLastL1BatchBlockNum: %w", err))
  723. }
  724. if tracerr.Unwrap(err) == sql.ErrNoRows || lastForgeL1TxsNum == nil {
  725. n := int64(-1)
  726. lastForgeL1TxsNum = &n
  727. }
  728. s.stats.UpdateSync(block, batch, &lastL1BatchBlockNum, lastForgeL1TxsNum)
  729. if err := s.updateCurrentNextSlotIfSync(true, false); err != nil {
  730. return tracerr.Wrap(err)
  731. }
  732. return nil
  733. }
  734. // rollupSync retrieves all the Rollup Smart Contract Data that happened at
  735. // ethBlock.blockNum with ethBlock.Hash.
  736. func (s *Synchronizer) rollupSync(ethBlock *common.Block) (*common.RollupData, error) {
  737. blockNum := ethBlock.Num
  738. var rollupData = common.NewRollupData()
  739. // var forgeL1TxsNum int64
  740. // Get rollup events in the block, and make sure the block hash matches
  741. // the expected one.
  742. rollupEvents, err := s.ethClient.RollupEventsByBlock(blockNum, &ethBlock.Hash)
  743. if err != nil && err.Error() == errStrUnknownBlock {
  744. return nil, tracerr.Wrap(ErrUnknownBlock)
  745. } else if err != nil {
  746. return nil, tracerr.Wrap(fmt.Errorf("RollupEventsByBlock: %w", err))
  747. }
  748. // No events in this block
  749. if rollupEvents == nil {
  750. return &rollupData, nil
  751. }
  752. var nextForgeL1TxsNum int64 // forgeL1TxsNum for the next L1Batch
  753. nextForgeL1TxsNumPtr, err := s.historyDB.GetLastL1TxsNum()
  754. if err != nil {
  755. return nil, tracerr.Wrap(err)
  756. }
  757. if nextForgeL1TxsNumPtr != nil {
  758. nextForgeL1TxsNum = *nextForgeL1TxsNumPtr + 1
  759. } else {
  760. nextForgeL1TxsNum = 0
  761. }
  762. // Get L1UserTX
  763. rollupData.L1UserTxs, err = getL1UserTx(rollupEvents.L1UserTx, blockNum)
  764. if err != nil {
  765. return nil, tracerr.Wrap(err)
  766. }
  767. // Get ForgeBatch events to get the L1CoordinatorTxs
  768. for _, evtForgeBatch := range rollupEvents.ForgeBatch {
  769. batchData := common.NewBatchData()
  770. position := 0
  771. // Get the input for each Tx
  772. forgeBatchArgs, sender, err := s.ethClient.RollupForgeBatchArgs(evtForgeBatch.EthTxHash,
  773. evtForgeBatch.L1UserTxsLen)
  774. if err != nil {
  775. return nil, tracerr.Wrap(fmt.Errorf("RollupForgeBatchArgs: %w", err))
  776. }
  777. batchNum := common.BatchNum(evtForgeBatch.BatchNum)
  778. var l1UserTxs []common.L1Tx
  779. // Check if this is a L1Batch to get L1 Tx from it
  780. if forgeBatchArgs.L1Batch {
  781. // Get L1UserTxs with toForgeL1TxsNum, which correspond
  782. // to the L1UserTxs that are forged in this batch, so
  783. // that stateDB can process them.
  784. // First try to find them in HistoryDB.
  785. l1UserTxs, err = s.historyDB.GetUnforgedL1UserTxs(nextForgeL1TxsNum)
  786. if err != nil {
  787. return nil, tracerr.Wrap(err)
  788. }
  789. // Apart from the DB, try to find them in this block.
  790. // This could happen because in a block there could be
  791. // multiple batches with L1Batch = true (although it's
  792. // a very rare case). If not found in the DB and the
  793. // block doesn't contain the l1UserTxs, it means that
  794. // the L1UserTxs queue with toForgeL1TxsNum was closed
  795. // empty, so we leave `l1UserTxs` as an empty slice.
  796. for _, l1UserTx := range rollupData.L1UserTxs {
  797. if *l1UserTx.ToForgeL1TxsNum == nextForgeL1TxsNum {
  798. l1UserTxs = append(l1UserTxs, l1UserTx)
  799. }
  800. }
  801. position = len(l1UserTxs)
  802. }
  803. l1TxsAuth := make([]common.AccountCreationAuth,
  804. 0, len(forgeBatchArgs.L1CoordinatorTxsAuths))
  805. // Get L1 Coordinator Txs
  806. for i := range forgeBatchArgs.L1CoordinatorTxs {
  807. l1CoordinatorTx := forgeBatchArgs.L1CoordinatorTxs[i]
  808. l1CoordinatorTx.Position = position
  809. // l1CoordinatorTx.ToForgeL1TxsNum = &forgeL1TxsNum
  810. l1CoordinatorTx.UserOrigin = false
  811. l1CoordinatorTx.EthBlockNum = blockNum
  812. l1CoordinatorTx.BatchNum = &batchNum
  813. l1Tx, err := common.NewL1Tx(&l1CoordinatorTx)
  814. if err != nil {
  815. return nil, tracerr.Wrap(err)
  816. }
  817. batchData.L1CoordinatorTxs = append(batchData.L1CoordinatorTxs, *l1Tx)
  818. position++
  819. // Create a slice of account creation auth to be
  820. // inserted later if not exists
  821. if l1CoordinatorTx.FromEthAddr != common.RollupConstEthAddressInternalOnly {
  822. l1CoordinatorTxAuth := forgeBatchArgs.L1CoordinatorTxsAuths[i]
  823. l1TxsAuth = append(l1TxsAuth, common.AccountCreationAuth{
  824. EthAddr: l1CoordinatorTx.FromEthAddr,
  825. BJJ: l1CoordinatorTx.FromBJJ,
  826. Signature: l1CoordinatorTxAuth,
  827. })
  828. }
  829. // fmt.Println("DGB l1coordtx")
  830. }
  831. // Insert the slice of account creation auth
  832. // only if the node run as a coordinator
  833. if s.l2DB != nil && len(l1TxsAuth) > 0 {
  834. err = s.l2DB.AddManyAccountCreationAuth(l1TxsAuth)
  835. if err != nil {
  836. return nil, tracerr.Wrap(err)
  837. }
  838. }
  839. // Insert all the txs forged in this batch (l1UserTxs,
  840. // L1CoordinatorTxs, PoolL2Txs) into stateDB so that they are
  841. // processed.
  842. // Set TxType to the forged L2Txs
  843. for i := range forgeBatchArgs.L2TxsData {
  844. if err := forgeBatchArgs.L2TxsData[i].SetType(); err != nil {
  845. return nil, tracerr.Wrap(err)
  846. }
  847. }
  848. // Transform L2 txs to PoolL2Txs
  849. // NOTE: This is a big ugly, find a better way
  850. poolL2Txs := common.L2TxsToPoolL2Txs(forgeBatchArgs.L2TxsData)
  851. if int(forgeBatchArgs.VerifierIdx) >= len(s.consts.Rollup.Verifiers) {
  852. return nil, tracerr.Wrap(fmt.Errorf("forgeBatchArgs.VerifierIdx (%v) >= "+
  853. " len(s.consts.Rollup.Verifiers) (%v)",
  854. forgeBatchArgs.VerifierIdx, len(s.consts.Rollup.Verifiers)))
  855. }
  856. tpc := txprocessor.Config{
  857. NLevels: uint32(s.consts.Rollup.Verifiers[forgeBatchArgs.VerifierIdx].NLevels),
  858. MaxTx: uint32(s.consts.Rollup.Verifiers[forgeBatchArgs.VerifierIdx].MaxTx),
  859. ChainID: s.cfg.ChainID,
  860. MaxFeeTx: common.RollupConstMaxFeeIdxCoordinator,
  861. MaxL1Tx: common.RollupConstMaxL1Tx,
  862. }
  863. tp := txprocessor.NewTxProcessor(s.stateDB, tpc)
  864. // ProcessTxs updates poolL2Txs adding: Nonce (and also TokenID, but we don't use it).
  865. processTxsOut, err := tp.ProcessTxs(forgeBatchArgs.FeeIdxCoordinator,
  866. l1UserTxs, batchData.L1CoordinatorTxs, poolL2Txs)
  867. if err != nil {
  868. return nil, tracerr.Wrap(err)
  869. }
  870. if s.stateDB.CurrentBatch() != batchNum {
  871. return nil, tracerr.Wrap(fmt.Errorf("stateDB.BatchNum (%v) != "+
  872. "evtForgeBatch.BatchNum = (%v)",
  873. s.stateDB.CurrentBatch(), batchNum))
  874. }
  875. if s.stateDB.MT.Root().BigInt().Cmp(forgeBatchArgs.NewStRoot) != 0 {
  876. return nil, tracerr.Wrap(fmt.Errorf("stateDB.MTRoot (%v) != "+
  877. "forgeBatchArgs.NewStRoot (%v)",
  878. s.stateDB.MT.Root().BigInt(), forgeBatchArgs.NewStRoot))
  879. }
  880. // Transform processed PoolL2 txs to L2 and store in BatchData
  881. l2Txs, err := common.PoolL2TxsToL2Txs(poolL2Txs) // NOTE: This is a big uggly, find a better way
  882. if err != nil {
  883. return nil, tracerr.Wrap(err)
  884. }
  885. // Set TxID, BlockNum, BatchNum and Position to the forged L2Txs
  886. for i := range l2Txs {
  887. if err := l2Txs[i].SetID(); err != nil {
  888. return nil, tracerr.Wrap(err)
  889. }
  890. l2Txs[i].EthBlockNum = blockNum
  891. l2Txs[i].BatchNum = batchNum
  892. l2Txs[i].Position = position
  893. position++
  894. }
  895. batchData.L2Txs = l2Txs
  896. // Set the BatchNum in the forged L1UserTxs
  897. for i := range l1UserTxs {
  898. l1UserTxs[i].BatchNum = &batchNum
  899. }
  900. batchData.L1UserTxs = l1UserTxs
  901. // Set batchNum in exits
  902. for i := range processTxsOut.ExitInfos {
  903. exit := &processTxsOut.ExitInfos[i]
  904. exit.BatchNum = batchNum
  905. }
  906. batchData.ExitTree = processTxsOut.ExitInfos
  907. for i := range processTxsOut.CreatedAccounts {
  908. createdAccount := &processTxsOut.CreatedAccounts[i]
  909. createdAccount.Nonce = 0
  910. createdAccount.Balance = big.NewInt(0)
  911. createdAccount.BatchNum = batchNum
  912. }
  913. batchData.CreatedAccounts = processTxsOut.CreatedAccounts
  914. batchData.UpdatedAccounts = make([]common.AccountUpdate, 0,
  915. len(processTxsOut.UpdatedAccounts))
  916. for _, acc := range processTxsOut.UpdatedAccounts {
  917. batchData.UpdatedAccounts = append(batchData.UpdatedAccounts,
  918. common.AccountUpdate{
  919. EthBlockNum: blockNum,
  920. BatchNum: batchNum,
  921. Idx: acc.Idx,
  922. Nonce: acc.Nonce,
  923. Balance: acc.Balance,
  924. })
  925. }
  926. slotNum := int64(0)
  927. if ethBlock.Num >= s.consts.Auction.GenesisBlockNum {
  928. slotNum = (ethBlock.Num - s.consts.Auction.GenesisBlockNum) /
  929. int64(s.consts.Auction.BlocksPerSlot)
  930. }
  931. // Get Batch information
  932. batch := common.Batch{
  933. BatchNum: batchNum,
  934. EthBlockNum: blockNum,
  935. ForgerAddr: *sender,
  936. CollectedFees: processTxsOut.CollectedFees,
  937. FeeIdxsCoordinator: forgeBatchArgs.FeeIdxCoordinator,
  938. StateRoot: forgeBatchArgs.NewStRoot,
  939. NumAccounts: len(batchData.CreatedAccounts),
  940. LastIdx: forgeBatchArgs.NewLastIdx,
  941. ExitRoot: forgeBatchArgs.NewExitRoot,
  942. SlotNum: slotNum,
  943. }
  944. nextForgeL1TxsNumCpy := nextForgeL1TxsNum
  945. if forgeBatchArgs.L1Batch {
  946. batch.ForgeL1TxsNum = &nextForgeL1TxsNumCpy
  947. batchData.L1Batch = true
  948. nextForgeL1TxsNum++
  949. }
  950. batchData.Batch = batch
  951. rollupData.Batches = append(rollupData.Batches, *batchData)
  952. }
  953. // Get Registered Tokens
  954. for _, evtAddToken := range rollupEvents.AddToken {
  955. var token common.Token
  956. token.TokenID = common.TokenID(evtAddToken.TokenID)
  957. token.EthAddr = evtAddToken.TokenAddress
  958. token.EthBlockNum = blockNum
  959. if consts, err := s.ethClient.EthERC20Consts(evtAddToken.TokenAddress); err != nil {
  960. log.Warnw("Error retreiving ERC20 token constants", "addr", evtAddToken.TokenAddress)
  961. token.Name = "ERC20_ETH_ERROR"
  962. token.Symbol = "ERROR"
  963. token.Decimals = 1
  964. } else {
  965. token.Name = cutStringMax(consts.Name, 20)
  966. token.Symbol = cutStringMax(consts.Symbol, 10)
  967. token.Decimals = consts.Decimals
  968. }
  969. rollupData.AddedTokens = append(rollupData.AddedTokens, token)
  970. }
  971. for _, evt := range rollupEvents.UpdateBucketWithdraw {
  972. rollupData.UpdateBucketWithdraw = append(rollupData.UpdateBucketWithdraw,
  973. common.BucketUpdate{
  974. EthBlockNum: blockNum,
  975. NumBucket: evt.NumBucket,
  976. BlockStamp: evt.BlockStamp,
  977. Withdrawals: evt.Withdrawals,
  978. })
  979. }
  980. for _, evt := range rollupEvents.Withdraw {
  981. rollupData.Withdrawals = append(rollupData.Withdrawals, common.WithdrawInfo{
  982. Idx: common.Idx(evt.Idx),
  983. NumExitRoot: common.BatchNum(evt.NumExitRoot),
  984. InstantWithdraw: evt.InstantWithdraw,
  985. TxHash: evt.TxHash,
  986. })
  987. }
  988. for _, evt := range rollupEvents.UpdateTokenExchange {
  989. if len(evt.AddressArray) != len(evt.ValueArray) {
  990. return nil, tracerr.Wrap(fmt.Errorf("in RollupEventUpdateTokenExchange "+
  991. "len(AddressArray) != len(ValueArray) (%v != %v)",
  992. len(evt.AddressArray), len(evt.ValueArray)))
  993. }
  994. for i := range evt.AddressArray {
  995. rollupData.TokenExchanges = append(rollupData.TokenExchanges,
  996. common.TokenExchange{
  997. EthBlockNum: blockNum,
  998. Address: evt.AddressArray[i],
  999. ValueUSD: int64(evt.ValueArray[i]),
  1000. })
  1001. }
  1002. }
  1003. varsUpdate := false
  1004. for _, evt := range rollupEvents.UpdateForgeL1L2BatchTimeout {
  1005. s.vars.Rollup.ForgeL1L2BatchTimeout = evt.NewForgeL1L2BatchTimeout
  1006. varsUpdate = true
  1007. }
  1008. for _, evt := range rollupEvents.UpdateFeeAddToken {
  1009. s.vars.Rollup.FeeAddToken = evt.NewFeeAddToken
  1010. varsUpdate = true
  1011. }
  1012. for _, evt := range rollupEvents.UpdateWithdrawalDelay {
  1013. s.vars.Rollup.WithdrawalDelay = evt.NewWithdrawalDelay
  1014. varsUpdate = true
  1015. }
  1016. // NOTE: We skip the event rollupEvents.SafeMode because the
  1017. // implementation RollupEventsByBlock already inserts a non-existing
  1018. // RollupEventUpdateBucketsParameters into UpdateBucketsParameters with
  1019. // all the bucket values at 0 and SafeMode = true
  1020. for _, evt := range rollupEvents.UpdateBucketsParameters {
  1021. for i, bucket := range evt.ArrayBuckets {
  1022. s.vars.Rollup.Buckets[i] = common.BucketParams{
  1023. CeilUSD: bucket.CeilUSD,
  1024. Withdrawals: bucket.Withdrawals,
  1025. BlockWithdrawalRate: bucket.BlockWithdrawalRate,
  1026. MaxWithdrawals: bucket.MaxWithdrawals,
  1027. }
  1028. }
  1029. s.vars.Rollup.SafeMode = evt.SafeMode
  1030. varsUpdate = true
  1031. }
  1032. if varsUpdate {
  1033. s.vars.Rollup.EthBlockNum = blockNum
  1034. rollupData.Vars = s.vars.Rollup.Copy()
  1035. }
  1036. return &rollupData, nil
  1037. }
  1038. func cutStringMax(s string, max int) string {
  1039. if len(s) > max {
  1040. return s[:max]
  1041. }
  1042. return s
  1043. }
  1044. // auctionSync gets information from the Auction Contract
  1045. func (s *Synchronizer) auctionSync(ethBlock *common.Block) (*common.AuctionData, error) {
  1046. blockNum := ethBlock.Num
  1047. var auctionData = common.NewAuctionData()
  1048. // Get auction events in the block
  1049. auctionEvents, err := s.ethClient.AuctionEventsByBlock(blockNum, &ethBlock.Hash)
  1050. if err != nil && err.Error() == errStrUnknownBlock {
  1051. return nil, tracerr.Wrap(ErrUnknownBlock)
  1052. } else if err != nil {
  1053. return nil, tracerr.Wrap(fmt.Errorf("AuctionEventsByBlock: %w", err))
  1054. }
  1055. // No events in this block
  1056. if auctionEvents == nil {
  1057. return &auctionData, nil
  1058. }
  1059. // Get bids
  1060. for _, evt := range auctionEvents.NewBid {
  1061. bid := common.Bid{
  1062. SlotNum: evt.Slot,
  1063. BidValue: evt.BidAmount,
  1064. Bidder: evt.Bidder,
  1065. EthBlockNum: blockNum,
  1066. }
  1067. auctionData.Bids = append(auctionData.Bids, bid)
  1068. }
  1069. // Get Coordinators
  1070. for _, evt := range auctionEvents.SetCoordinator {
  1071. coordinator := common.Coordinator{
  1072. Bidder: evt.BidderAddress,
  1073. Forger: evt.ForgerAddress,
  1074. URL: evt.CoordinatorURL,
  1075. EthBlockNum: blockNum,
  1076. }
  1077. auctionData.Coordinators = append(auctionData.Coordinators, coordinator)
  1078. }
  1079. varsUpdate := false
  1080. for _, evt := range auctionEvents.NewSlotDeadline {
  1081. s.vars.Auction.SlotDeadline = evt.NewSlotDeadline
  1082. varsUpdate = true
  1083. }
  1084. for _, evt := range auctionEvents.NewClosedAuctionSlots {
  1085. s.vars.Auction.ClosedAuctionSlots = evt.NewClosedAuctionSlots
  1086. varsUpdate = true
  1087. }
  1088. for _, evt := range auctionEvents.NewOutbidding {
  1089. s.vars.Auction.Outbidding = evt.NewOutbidding
  1090. varsUpdate = true
  1091. }
  1092. for _, evt := range auctionEvents.NewDonationAddress {
  1093. s.vars.Auction.DonationAddress = evt.NewDonationAddress
  1094. varsUpdate = true
  1095. }
  1096. for _, evt := range auctionEvents.NewBootCoordinator {
  1097. s.vars.Auction.BootCoordinator = evt.NewBootCoordinator
  1098. s.vars.Auction.BootCoordinatorURL = evt.NewBootCoordinatorURL
  1099. varsUpdate = true
  1100. // Add new boot coordinator
  1101. auctionData.Coordinators = append(auctionData.Coordinators, common.Coordinator{
  1102. Forger: evt.NewBootCoordinator,
  1103. URL: evt.NewBootCoordinatorURL,
  1104. EthBlockNum: blockNum,
  1105. })
  1106. }
  1107. for _, evt := range auctionEvents.NewOpenAuctionSlots {
  1108. s.vars.Auction.OpenAuctionSlots = evt.NewOpenAuctionSlots
  1109. varsUpdate = true
  1110. }
  1111. for _, evt := range auctionEvents.NewAllocationRatio {
  1112. s.vars.Auction.AllocationRatio = evt.NewAllocationRatio
  1113. varsUpdate = true
  1114. }
  1115. for _, evt := range auctionEvents.NewDefaultSlotSetBid {
  1116. if evt.SlotSet > 6 { //nolint:gomnd
  1117. return nil, tracerr.Wrap(fmt.Errorf("unexpected SlotSet in "+
  1118. "auctionEvents.NewDefaultSlotSetBid: %v", evt.SlotSet))
  1119. }
  1120. s.vars.Auction.DefaultSlotSetBid[evt.SlotSet] = evt.NewInitialMinBid
  1121. s.vars.Auction.DefaultSlotSetBidSlotNum = s.consts.Auction.SlotNum(blockNum) +
  1122. int64(s.vars.Auction.ClosedAuctionSlots)
  1123. varsUpdate = true
  1124. }
  1125. // NOTE: We ignore NewForgeAllocated
  1126. // NOTE: We ignore NewForge because we're already tracking ForgeBatch event from Rollup
  1127. // NOTE: We ignore HEZClaimed
  1128. if varsUpdate {
  1129. s.vars.Auction.EthBlockNum = blockNum
  1130. auctionData.Vars = s.vars.Auction.Copy()
  1131. }
  1132. return &auctionData, nil
  1133. }
  1134. // wdelayerSync gets information from the Withdrawal Delayer Contract
  1135. func (s *Synchronizer) wdelayerSync(ethBlock *common.Block) (*common.WDelayerData, error) {
  1136. blockNum := ethBlock.Num
  1137. wDelayerData := common.NewWDelayerData()
  1138. // Get wDelayer events in the block
  1139. wDelayerEvents, err := s.ethClient.WDelayerEventsByBlock(blockNum, &ethBlock.Hash)
  1140. if err != nil && err.Error() == errStrUnknownBlock {
  1141. return nil, tracerr.Wrap(ErrUnknownBlock)
  1142. } else if err != nil {
  1143. return nil, tracerr.Wrap(fmt.Errorf("WDelayerEventsByBlock: %w", err))
  1144. }
  1145. // No events in this block
  1146. if wDelayerEvents == nil {
  1147. return &wDelayerData, nil
  1148. }
  1149. for _, evt := range wDelayerEvents.Deposit {
  1150. wDelayerData.Deposits = append(wDelayerData.Deposits, common.WDelayerTransfer{
  1151. Owner: evt.Owner,
  1152. Token: evt.Token,
  1153. Amount: evt.Amount,
  1154. })
  1155. wDelayerData.DepositsByTxHash[evt.TxHash] =
  1156. append(wDelayerData.DepositsByTxHash[evt.TxHash],
  1157. &wDelayerData.Deposits[len(wDelayerData.Deposits)-1])
  1158. }
  1159. for _, evt := range wDelayerEvents.Withdraw {
  1160. wDelayerData.Withdrawals = append(wDelayerData.Withdrawals, common.WDelayerTransfer{
  1161. Owner: evt.Owner,
  1162. Token: evt.Token,
  1163. Amount: evt.Amount,
  1164. })
  1165. }
  1166. for _, evt := range wDelayerEvents.EscapeHatchWithdrawal {
  1167. wDelayerData.EscapeHatchWithdrawals = append(wDelayerData.EscapeHatchWithdrawals,
  1168. common.WDelayerEscapeHatchWithdrawal{
  1169. EthBlockNum: blockNum,
  1170. Who: evt.Who,
  1171. To: evt.To,
  1172. TokenAddr: evt.Token,
  1173. Amount: evt.Amount,
  1174. })
  1175. }
  1176. varsUpdate := false
  1177. for range wDelayerEvents.EmergencyModeEnabled {
  1178. s.vars.WDelayer.EmergencyMode = true
  1179. s.vars.WDelayer.EmergencyModeStartingBlock = blockNum
  1180. varsUpdate = true
  1181. }
  1182. for _, evt := range wDelayerEvents.NewWithdrawalDelay {
  1183. s.vars.WDelayer.WithdrawalDelay = evt.WithdrawalDelay
  1184. varsUpdate = true
  1185. }
  1186. for _, evt := range wDelayerEvents.NewEmergencyCouncil {
  1187. s.vars.WDelayer.EmergencyCouncilAddress = evt.NewEmergencyCouncil
  1188. varsUpdate = true
  1189. }
  1190. for _, evt := range wDelayerEvents.NewHermezGovernanceAddress {
  1191. s.vars.WDelayer.HermezGovernanceAddress = evt.NewHermezGovernanceAddress
  1192. varsUpdate = true
  1193. }
  1194. if varsUpdate {
  1195. s.vars.WDelayer.EthBlockNum = blockNum
  1196. wDelayerData.Vars = s.vars.WDelayer.Copy()
  1197. }
  1198. return &wDelayerData, nil
  1199. }
  1200. func getL1UserTx(eventsL1UserTx []eth.RollupEventL1UserTx, blockNum int64) ([]common.L1Tx, error) {
  1201. l1Txs := make([]common.L1Tx, len(eventsL1UserTx))
  1202. for i := range eventsL1UserTx {
  1203. eventsL1UserTx[i].L1UserTx.EthBlockNum = blockNum
  1204. // Check validity of L1UserTx
  1205. l1Tx, err := common.NewL1Tx(&eventsL1UserTx[i].L1UserTx)
  1206. if err != nil {
  1207. return nil, tracerr.Wrap(err)
  1208. }
  1209. l1Txs[i] = *l1Tx
  1210. }
  1211. return l1Txs, nil
  1212. }