feat(sequencer): catchup from base #3057
Conversation
|
The latest Buf updates on your PR. Results from workflow CI / buf-check (pull_request).
|
|
Claude finished @julienrbrt's task in 1m 43s —— View job Code Review CompleteI've completed a comprehensive review of the sequencer catch-up implementation. The PR adds a critical feature for handling sequencer restarts after downtime, but there are critical issues that must be addressed before merging. Todo List
Critical Issues 🔴1. Non-Monotonic Block Timestamps (Severity: Critical)Location: The timestamp logic during catch-up produces non-monotonic timestamps, which will cause execution layer rejection: timestamp := time.Now()
if c.catchingUp && !c.currentDAEndTime.IsZero() {
var remainingForcedTxs uint64
if len(c.cachedForcedInclusionTxs) > 0 {
remainingForcedTxs = uint64(len(c.cachedForcedInclusionTxs)) - c.checkpoint.TxIndex
}
timestamp = c.currentDAEndTime.Add(-time.Duration(remainingForcedTxs) * time.Millisecond)
}Problems:
Impact: Most execution layers (EVM, Cosmos SDK) require strictly monotonic timestamps. Non-monotonic timestamps will cause blocks to be rejected, halting the chain. Suggested Fix: Fix timestamp monotonicity → 2. Data Race on Catch-Up State Fields (Severity: Critical)Location: The fields catchingUp bool
currentDAEndTime time.TimeConcurrent Access:
Impact:
Suggested Fix: Fix data race → High Priority Issues 🟡3. Magic Number Should Be a ConstantLocation: The value // Line 565 (implied in logic but not shown)
if missedEpochs <= 1 {The catch-up detection logic relies on epoch calculations, but there's no clear timeout or threshold constant. While the epoch-based logic is reasonable, adding a named constant like Suggested Fix: Define constants at package level for any timing thresholds used in catch-up detection. 4. Test Logger Configuration InconsistencyLocation: Some tests use Suggested Approach: Use Positive Aspects ✅
Architecture & Code QualityStrengths:
Areas for Improvement:
RecommendationsBefore Merging:
Nice to Have:
SummaryThe implementation provides a solid foundation for sequencer catch-up functionality, with good test coverage and clean architecture. However, the critical issues with timestamp monotonicity and data races must be addressed before merging, as they will cause chain halts and undefined behavior in production. The syncer integration is well-done and correctly handles the incremental DA height advancement needed for catch-up blocks. The epoch-based detection is efficient and elegant. Recommendation: Do not merge until the two critical issues are resolved and verified with race detection enabled. |
Summary of ChangesHello @julienrbrt, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request implements a critical "catch-up" feature for the sequencer, designed to enhance its resilience and consistency following periods of inactivity or downtime. By intelligently replaying missed Data Availability (DA) epochs, the sequencer can resynchronize its state, ensuring that all forced inclusion transactions are processed in the correct order and with accurate timestamps, thereby maintaining a consistent view of the blockchain's history, even when operating in a decentralized environment. This mechanism allows the sequencer to seamlessly transition back to normal operation without compromising data integrity. Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Changelog
Activity
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #3057 +/- ##
==========================================
+ Coverage 61.83% 61.90% +0.07%
==========================================
Files 111 111
Lines 11123 11226 +103
==========================================
+ Hits 6878 6950 +72
- Misses 3506 3537 +31
Partials 739 739
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a robust catch-up mechanism for the sequencer, designed to handle restarts after extended downtime. While the overall approach for consuming and verifying catch-up blocks in the syncer, including the incremental advancement of DAHeight, is well-implemented and tested, the implementation of catch-up mode in the single sequencer has significant flaws. Specifically, it produces non-monotonic block timestamps when multiple blocks are generated for a single DA epoch or when empty epochs are encountered, which will likely cause the execution layer to reject blocks and halt the chain. Additionally, there is a data race on the new catch-up state fields due to a lack of synchronization primitives. Minor suggestions for code clarity and testing experience were also noted.
| timestamp := time.Now() | ||
| if c.catchingUp && !c.currentDAEndTime.IsZero() { | ||
| timestamp = c.currentDAEndTime | ||
| } |
There was a problem hiding this comment.
During catch-up mode, the sequencer uses the timestamp from the last fetched DA epoch as the block timestamp. However, multiple blocks can be produced for a single DA epoch (e.g., if the epoch contains many transactions that don't fit in one block). In this case, all blocks produced for that epoch will have the exact same timestamp (c.currentDAEndTime). Furthermore, if an empty DA epoch is encountered, c.currentDAEndTime is not updated, causing the next block to reuse the timestamp from the previous non-empty epoch.
Most blockchain execution layers (like EVM) require that each new block has a timestamp strictly greater than its parent. If the sequencer produces multiple blocks with the same timestamp, the execution layer will reject them, halting the block production process and causing a Denial of Service.
| catchingUp bool | ||
| // currentDAEndTime is the DA epoch end timestamp from the last fetched epoch. | ||
| // Used as the block timestamp during catch-up to match based sequencing behavior. | ||
| currentDAEndTime time.Time |
There was a problem hiding this comment.
The fields catchingUp and currentDAEndTime are accessed concurrently without proper synchronization. GetNextBatch and fetchNextDAEpoch (which update these fields) can be called by the block production loop, while IsCatchingUp() can be called concurrently by other components (e.g., RPC, monitoring, or metrics).
In Go, concurrent read/write access to non-atomic fields is a data race, which leads to undefined behavior. Specifically, a race on time.Time (which is a struct) can result in a corrupted timestamp being read, potentially leading to invalid blocks and a Denial of Service.
pkg/sequencers/single/sequencer.go
Outdated
| if catchUpThreshold < 30*time.Second { | ||
| catchUpThreshold = 30 * time.Second | ||
| } |
There was a problem hiding this comment.
The value 30*time.Second is used as a minimum threshold for the catch-up logic. To improve readability and maintainability, consider defining this as a named constant at the package level, for example: const minCatchUpThreshold = 30 * time.Second, and using that constant here. This makes the intent of the value clearer and centralizes its definition.
| // This test uses maxBytes to limit how many txs are fetched, triggering the unprocessed txs scenario. | ||
| func TestSequencer_CatchUp_DetectsOldEpoch(t *testing.T) { | ||
| ctx := context.Background() | ||
| logger := zerolog.New(zerolog.NewConsoleWriter()) |
There was a problem hiding this comment.
Using zerolog.New(zerolog.NewConsoleWriter()) in tests can lead to verbose output in the console during test runs. It's generally better to use a logger that integrates with the testing framework, such as zerolog.New(zerolog.NewTestWriter(t)), which only shows logs for failed tests, or zerolog.Nop() if logs are not needed. This practice is followed in other new tests in this file (e.g., TestSequencer_CatchUp_UsesDATimestamp). Applying this consistently would improve the developer experience. This comment also applies to lines 1309, 1457, and 1676.
| logger := zerolog.New(zerolog.NewConsoleWriter()) | |
| logger := zerolog.New(zerolog.NewTestWriter(t)) |
1 participant
Closes: #3016
Opus did a pretty good job. I'll be improving it.