Repository navigation
Conversation
Add opt-in batch-owned recovery pools with configurable threads per data directory and drain recovery before closing dependencies. Serialize recovery checkpoint capture and publication per directory; cover concurrent truncation, crash recovery, failures and shutdown.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Purpose
Closes #4914
Allow replicas to recover concurrently during follower transitions instead of processing every replica serially.
Brief change log
log.replica.follower-recovery.enabled(defaultfalse) andlog.replica.follower-recovery.threads-per-data-dir(default1). Enabled recovery uses temporary pools per actual data directory.Tests
117 tests passed across eight related suites on Java 11:
./mvnw -pl fluss-server -am test \ -Dtest=LogManagerTest,FollowerRecoveryExecutorTest,FollowerRecoveryTest,FollowerRecoveryShutdownTest,TabletServerShutdownTest,ReplicaManagerTest,ReplicaTest,ReplicaFetcherManagerTest \ -Dsurefire.failIfNoSpecifiedTests=falseCoverage includes directory concurrency limits, serial fallback, failure and interruption handling, shutdown ordering, and log/writer-state recovery. Three checkpoint regressions fail without the fix, including unclean recovery with a corrupted retained segment.
Spotless, Checkstyle, RAT, and configuration documentation generation passed. The full repository suite and performance benchmarks were not run.
API and Format
Two new server configuration options; no public Java API, RPC, or storage format changes. Parallel recovery is disabled by default.
Documentation
Updated the configuration reference with both options, their defaults, and concurrency semantics.
Generative AI disclosure
Generated-by: OpenAI Codex
OpenAI Codex assisted with implementation, tests, code review, and PR preparation.