feat: update wd-mon to work for OptimismPortal2

.changeset/dry-bobcats-obey.md

@@ -0,0 +1,5 @@
+---
+'@eth-optimism/common-ts': patch
+---
+
+Adds a new validator for address types.

.changeset/ninety-pugs-clap.md

@@ -0,0 +1,5 @@
+---
+'@eth-optimism/chain-mon': minor
+---
+
+Updates wd-mon inside chain-mon to support FPAC.

packages/chain-mon/src/wd-mon/constants.ts

@@ -0,0 +1,21 @@
+import { L2ChainID } from '@eth-optimism/sdk'
+
+// TODO: Consider moving to `@eth-optimism/constants` and generating from superchain registry.
+// @see https://github.com/ethereum-optimism/optimism/pull/9041
+
+/**
+ * Mapping of L2ChainIDs to the L1 block numbers where the wd-mon service should start looking for
+ * withdrawals by default. L1 block numbers here are based on the block number in which the
+ * OptimismPortal proxy contract was deployed to L1.
+ */
+export const DEFAULT_STARTING_BLOCK_NUMBERS: {
+  [ChainID in L2ChainID]?: number
+} = {
+  [L2ChainID.OPTIMISM]: 17365802 as const,
+  [L2ChainID.OPTIMISM_GOERLI]: 8299684 as const,
+  [L2ChainID.OPTIMISM_SEPOLIA]: 4071248 as const,
+  [L2ChainID.BASE_MAINNET]: 17482143 as const,
+  [L2ChainID.BASE_GOERLI]: 8411116 as const,
+  [L2ChainID.BASE_SEPOLIA]: 4370901 as const,
+  [L2ChainID.ZORA_MAINNET]: 17473938 as const,
+}

packages/chain-mon/src/wd-mon/service.ts

@@ -6,30 +6,35 @@ import {
   validators,
   waitForProvider,
 } from '@eth-optimism/common-ts'
-import { CrossChainMessenger } from '@eth-optimism/sdk'
+import { getOEContract, DEFAULT_L2_CONTRACT_ADDRESSES } from '@eth-optimism/sdk'
 import { getChainId, sleep } from '@eth-optimism/core-utils'
 import { Provider } from '@ethersproject/abstract-provider'
-import { Event } from 'ethers'
+import { ethers } from 'ethers'
 import dateformat from 'dateformat'
 
 import { version } from '../../package.json'
-import { getLastFinalizedBlock as getLastFinalizedBlock } from './helpers'
+import { DEFAULT_STARTING_BLOCK_NUMBERS } from './constants'
 
 type Options = {
   l1RpcProvider: Provider
   l2RpcProvider: Provider
+  optimismPortalAddress: string
+  l2ToL1MessagePasserAddress: string
   startBlockNumber: number
+  eventBlockRange: number
   sleepTimeMs: number
 }
 
 type Metrics = {
+  highestBlockNumber: Gauge
   withdrawalsValidated: Gauge
   isDetectingForgeries: Gauge
   nodeConnectionFailures: Gauge
 }
 
 type State = {
-  messenger: CrossChainMessenger
+  portal: ethers.Contract
+  messenger: ethers.Contract
   highestUncheckedBlockNumber: number
   faultProofWindow: number
   forgeryDetected: boolean
@@ -54,12 +59,30 @@ export class WithdrawalMonitor extends BaseServiceV2<Options, Metrics, State> {
           validator: validators.provider,
           desc: 'Provider for interacting with L2',
         },
+        optimismPortalAddress: {
+          validator: validators.address,
+          default: null,
+          desc: 'Address of the OptimismPortal proxy contract on L1',
+          public: true,
+        },
+        l2ToL1MessagePasserAddress: {
+          validator: validators.address,
+          default: DEFAULT_L2_CONTRACT_ADDRESSES.BedrockMessagePasser as string,
+          desc: 'Address of the L2ToL1MessagePasser contract on L2',
+          public: true,
+        },
         startBlockNumber: {
           validator: validators.num,
           default: -1,
           desc: 'L1 block number to start checking from',
           public: true,
         },
+        eventBlockRange: {
+          validator: validators.num,
+          default: 2000,
+          desc: 'Number of blocks to query for events over per loop',
+          public: true,
+        },
         sleepTimeMs: {
           validator: validators.num,
           default: 15000,
@@ -68,6 +91,11 @@ export class WithdrawalMonitor extends BaseServiceV2<Options, Metrics, State> {
         },
       },
       metricsSpec: {
+        highestBlockNumber: {
+          type: Gauge,
+          desc: 'Highest block number (checked and known)',
+          labels: ['type'],
+        },
         withdrawalsValidated: {
           type: Gauge,
           desc: 'Latest L1 Block (checked and known)',
@@ -99,38 +127,41 @@ export class WithdrawalMonitor extends BaseServiceV2<Options, Metrics, State> {
       name: 'L2',
     })
 
-    this.state.messenger = new CrossChainMessenger({
-      l1SignerOrProvider: this.options.l1RpcProvider,
-      l2SignerOrProvider: this.options.l2RpcProvider,
-      l1ChainId: await getChainId(this.options.l1RpcProvider),
-      l2ChainId: await getChainId(this.options.l2RpcProvider),
-      bedrock: true,
-    })
+    // Need L2 chain ID to resolve contract addresses.
+    const l2ChainId = await getChainId(this.options.l2RpcProvider)
 
-    // Not detected by default.
-    this.state.forgeryDetected = false
+    // Create the OptimismPortal contract instance. If the optimismPortal option is not provided
+    // then the SDK will attempt to resolve the address automatically based on the L2 chain ID. If
+    // the SDK isn't aware of the L2 chain ID then it will throw an error that makes it clear the
+    // user needs to provide this value explicitly.
+    this.state.portal = getOEContract('OptimismPortal', l2ChainId, {
+      signerOrProvider: this.options.l1RpcProvider,
+      address: this.options.optimismPortalAddress,
+    })
 
-    this.state.faultProofWindow =
-      await this.state.messenger.getChallengePeriodSeconds()
-    this.logger.info(
-      `fault proof window is ${this.state.faultProofWindow} seconds`
-    )
+    // Create the L2ToL1MessagePasser contract instance. If the l2ToL1MessagePasser option is not
+    // provided then we'll use the default address which typically should be correct. It's very
+    // unlikely that any user would change this address so this should work in 99% of cases. If we
+    // really wanted to be extra safe we could do some sanity checks to make sure the contract has
+    // the interface we need but doesn't seem important for now.
+    this.state.messenger = getOEContract('L2ToL1MessagePasser', l2ChainId, {
+      signerOrProvider: this.options.l2RpcProvider,
+      address: this.options.l2ToL1MessagePasserAddress,
+    })
 
-    // Set the start block number.
+    // Previous versions of wd-mon would try to pick the starting block number automatically but
+    // this had the possibility of missing certain withdrawals if the service was restarted at the
+    // wrong time. Given the added complexity of finding a starting point automatically after FPAC,
+    // it's much easier to simply start a fixed block number than trying to do something fancy. Use
+    // the default configured in this service or use zero if no default is defined.
+    this.state.highestUncheckedBlockNumber = this.options.startBlockNumber
     if (this.options.startBlockNumber === -1) {
-      // We default to starting from the last finalized block.
-      this.state.highestUncheckedBlockNumber = await getLastFinalizedBlock(
-        this.options.l1RpcProvider,
-        this.state.faultProofWindow,
-        this.logger
-      )
-    } else {
-      this.state.highestUncheckedBlockNumber = this.options.startBlockNumber
+      this.state.highestUncheckedBlockNumber =
+        DEFAULT_STARTING_BLOCK_NUMBERS[l2ChainId] || 0
     }
 
-    this.logger.info(`starting L1 block height`, {
-      startBlockNumber: this.state.highestUncheckedBlockNumber,
-    })
+    // Default state is that forgeries have not been detected.
+    this.state.forgeryDetected = false
   }
 
   // K8s healthcheck
@@ -143,94 +174,131 @@ export class WithdrawalMonitor extends BaseServiceV2<Options, Metrics, State> {
   }
 
   async main(): Promise<void> {
-    // Get current block number
+    // Get the latest L1 block number.
     let latestL1BlockNumber: number
     try {
       latestL1BlockNumber = await this.options.l1RpcProvider.getBlockNumber()
     } catch (err) {
+      // Log the issue so we can debug it.
       this.logger.error(`got error when connecting to node`, {
         error: err,
         node: 'l1',
         section: 'getBlockNumber',
       })
+
+      // Increment the metric so we can detect the issue.
       this.metrics.nodeConnectionFailures.inc({
         layer: 'l1',
         section: 'getBlockNumber',
       })
-      await sleep(this.options.sleepTimeMs)
-      return
+
+      // Sleep for a little to give intermittent errors a chance to recover.
+      return sleep(this.options.sleepTimeMs)
     }
 
-    // See if we have a new unchecked block
+    // Update highest block number metrics so we can keep track of how the service is doing.
+    this.metrics.highestBlockNumber.set({ type: 'known' }, latestL1BlockNumber)
+    this.metrics.highestBlockNumber.set(
+      { type: 'checked' },
+      this.state.highestUncheckedBlockNumber
+    )
+
+    // Check if the RPC provider is behind us for some reason. Can happen occasionally,
+    // particularly if connected to an RPC provider that load balances over multiple nodes that
+    // might not be perfectly in sync.
     if (latestL1BlockNumber <= this.state.highestUncheckedBlockNumber) {
-      // The RPC provider is behind us, wait a bit
-      await sleep(this.options.sleepTimeMs)
-      return
+      // Sleep for a little to give the RPC a chance to catch up.
+      return sleep(this.options.sleepTimeMs)
     }
 
+    // Generally better to use a relatively small block range because it means this service can be
+    // used alongside many different types of L1 nodes. For instance, Geth will typically only
+    // support a block range of 2000 blocks out of the box.
+    const toBlockNumber = Math.min(
+      this.state.highestUncheckedBlockNumber + this.options.eventBlockRange,
+      latestL1BlockNumber
+    )
+
+    // Useful to log this stuff just in case we get stuck or something.
     this.logger.info(`checking recent blocks`, {
       fromBlockNumber: this.state.highestUncheckedBlockNumber,
-      toBlockNumber: latestL1BlockNumber,
+      toBlockNumber,
     })
 
-    // Perform the check
-    let proofEvents: Event[]
+    // Query for WithdrawalProven events within the specified block range.
+    let events: ethers.Event[]
     try {
-      // The query includes events in the blockNumbers given as the last two arguments
-      proofEvents =
-        await this.state.messenger.contracts.l1.OptimismPortal.queryFilter(
-          this.state.messenger.contracts.l1.OptimismPortal.filters.WithdrawalProven(),
-          this.state.highestUncheckedBlockNumber,
-          latestL1BlockNumber
-        )
+      events = await this.state.portal.queryFilter(
+        this.state.portal.filters.WithdrawalProven(),
+        this.state.highestUncheckedBlockNumber,
+        toBlockNumber
+      )
     } catch (err) {
+      // Log the issue so we can debug it.
       this.logger.error(`got error when connecting to node`, {
         error: err,
         node: 'l1',
         section: 'querying for WithdrawalProven events',
       })
+
+      // Increment the metric so we can detect the issue.
       this.metrics.nodeConnectionFailures.inc({
         layer: 'l1',
         section: 'querying for WithdrawalProven events',
       })
-      // connection error, wait then restart
-      await sleep(this.options.sleepTimeMs)
-      return
+
+      // Sleep for a little to give intermittent errors a chance to recover.
+      return sleep(this.options.sleepTimeMs)
     }
 
-    for (const proofEvent of proofEvents) {
-      const exists =
-        await this.state.messenger.contracts.l2.BedrockMessagePasser.sentMessages(
-          proofEvent.args.withdrawalHash
-        )
-      const block = await proofEvent.getBlock()
-      const now = new Date(block.timestamp * 1000)
-      const dateString = dateformat(
-        now,
-        'mmmm dS, yyyy, h:MM:ss TT',
-        true // use UTC time
-      )
-      const provenAt = `${dateString} UTC`
+    // Go over all the events and check if the withdrawal hash actually exists on L2.
+    for (const event of events) {
+      // Could consider using multicall here but this is efficient enough for now.
+      const hash = event.args.withdrawalHash
+      const exists = await this.state.messenger.sentMessages(hash)
+
+      // Hopefully the withdrawal exists!
       if (exists) {
-        this.metrics.withdrawalsValidated.inc()
+        // Unlike below we don't grab the timestamp here because it adds an unnecessary request.
         this.logger.info(`valid withdrawal`, {
-          withdrawalHash: proofEvent.args.withdrawalHash,
-          provenAt,
+          withdrawalHash: event.args.withdrawalHash,
         })
+
+        // Bump the withdrawals metric so we can keep track.
+        this.metrics.withdrawalsValidated.inc()
       } else {
+        // Grab and format the timestamp so it's clear how much time is left.
+        const block = await event.getBlock()
+        const ts = `${dateformat(
+          new Date(block.timestamp * 1000),
+          'mmmm dS, yyyy, h:MM:ss TT',
+          true
+        )} UTC`
+
+        // Uh oh!
         this.logger.error(`withdrawalHash not seen on L2`, {
-          withdrawalHash: proofEvent.args.withdrawalHash,
-          provenAt,
+          withdrawalHash: event.args.withdrawalHash,
+          provenAt: ts,
         })
+
+        // Change to forgery state.
         this.state.forgeryDetected = true
         this.metrics.isDetectingForgeries.set(1)
-        return
+
+        // Return early so that we never increment the highest unchecked block number and therefore
+        // will continue to loop on this forgery indefinitely. We probably want to change this
+        // behavior at some point so that we keep scanning for additional forgeries since the
+        // existence of one forgery likely implies the existence of many others.
+        return sleep(this.options.sleepTimeMs)
       }
     }
 
-    this.state.highestUncheckedBlockNumber = latestL1BlockNumber + 1
+    // Increment the highest unchecked block number for the next loop.
+    this.state.highestUncheckedBlockNumber = toBlockNumber
 
-    // If we got through the above without throwing an error, we should be fine to reset.
+    // If we got through the above without throwing an error, we should be fine to reset. Only case
+    // where this is relevant is if something is detected as a forgery accidentally and the error
+    // doesn't happen again on the next loop.
     this.state.forgeryDetected = false
     this.metrics.isDetectingForgeries.set(0)
   }