Fix data race when migrating slots #324
Merged
Conversation
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
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## unstable #324 +/- ##
============================================
+ Coverage 43.38% 47.08% +3.69%
============================================
Files 37 45 +8
Lines 2971 4420 +1449
============================================
+ Hits 1289 2081 +792
- Misses 1544 2130 +586
- Partials 138 209 +71
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:
|
There was a problem hiding this comment.
Pull Request Overview
This PR aims to resolve migration inconsistencies by resetting shard migration state per-shard, ensuring slot-only migrations clear previous migration info, and serializing the MigrateSlot API to avoid concurrent conflicts.
- Clear migration state for source shard on slot-only migrations
- Serialize MigrateSlot API calls with a handler-level mutex and manual cluster lookup
- Adjust routing middleware and change loop error handling to continue on per-shard failures
Reviewed Changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| store/cluster.go | Clear migration state before moving slots when slotOnly is true |
| server/route.go | Removed RequiredCluster middleware from the migrate endpoint |
| server/api/cluster.go | Added sync.Mutex, switched to manual cluster retrieval, and serialized MigrateSlot calls |
| controller/cluster.go | Replaced early return with continue in shard status update loop |
Comments suppressed due to low confidence (4)
server/route.go:72
- The migrate route no longer includes RequiredNamespace and RequiredCluster middleware, which may bypass essential validation and context injection. Consider re-adding the appropriate middleware to enforce namespace and cluster context.
clusters.POST("/:cluster/migrate", handler.Cluster.MigrateSlot)
server/api/cluster.go:50
- [nitpick] The field name
muis generic; consider renaming it to something more descriptive likemigrateMuorclusterMuto clarify its purpose.
mu sync.Mutex
store/cluster.go:210
- [nitpick] It may be helpful to add a comment explaining why
ClearMigrateStateis only called for the source shard whenslotOnlyis true, to improve maintainability and clarity of intent.
cluster.Shards[sourceShardIdx].ClearMigrateState()
server/api/cluster.go:123
- Using a single mutex on the handler serializes all MigrateSlot calls across all clusters, potentially creating a bottleneck. Consider using a more granular lock mechanism (e.g., per-cluster) to allow concurrent migrations on different clusters.
handler.mu.Lock()
git-hulk
reviewed
Jul 13, 2025
git-hulk
approved these changes
Jul 14, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Uh oh!
There was an error while loading. Please reload this page.