[hotfix] Fix the problems using Kafka as action state store #510
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
github-actions
bot
added
priority/major
xintongsong
reviewed
Feb 1, 2026
| // Byte array should not be in the log | ||
| assertThat(eventNode.has("event")).isFalse(); | ||
| // Byte array should be serialized for ActionState persistence | ||
| assertThat(eventNode.has("event")).isTrue(); |
Contributor
There was a problem hiding this comment.
This test case should not be changed. It verified the serialization for event logs, rather than for action states. It's expected not changed. This is the cause of the CI failures.
xintongsong
approved these changes
Feb 1, 2026
Contributor
xintongsong
left a comment
xintongsong
left a comment
There was a problem hiding this comment.
There's one minor issue, which I've fixed with an extra commit. Other than that, LGTM.
xintongsong
pushed a commit
that referenced
this pull request
Feb 1, 2026
- Fix ActionState deserialization failure: Removed @JsonIgnore from PythonEvent.getEvent() so that Python event bytes are persisted in ActionState. Without this, recovered ActionState had null event bytes, causing TypeError: a bytes-like object is required, not 'NoneType' after recovery. - Fix ActionStateStore not initialized during state recovery: Moved ActionStateStore initialization to also run in initializeState(), ensuring it's available when rebuilding state from recovery markers. - Fix state key mismatch after recovery: Changed Python Event.id from random UUID to deterministic content-based UUID (MD5 hash). This ensures the same event produces the same ActionState key across restarts, enabling proper state lookup and divergence detection. - Fix Kafka consumer partition assignment: Use consumer.assign() instead of subscribe() for explicit partition control during state rebuild.
Contributor
|
Ported to release-0.2 in 9ada100 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Linked issue: #xxx
This PR fixes issues with the Kafka ActionStateStore that prevented actions from being correctly skipped after checkpoint recovery:
Fix ActionState deserialization failure: Removed
@JsonIgnorefromPythonEvent.getEvent()so that Python event bytes are persisted in ActionState. Without this, recovered ActionState hadnullevent bytes, causingTypeError: a bytes-like object is required, not 'NoneType'after recovery.Fix ActionStateStore not initialized during state recovery: Moved ActionStateStore initialization to also run in
initializeState(), ensuring it's available when rebuilding state from recovery markers.Fix state key mismatch after recovery: Changed Python
Event.idfrom random UUID to deterministic content-based UUID (MD5 hash). This ensures the same event produces the same ActionState key across restarts, enabling proper state lookup and divergence detection.Fix Kafka consumer partition assignment: Use
consumer.assign()instead ofsubscribe()for explicit partition control during state rebuild.Tests
Manually verified with local Kafka instance. E2E tests will be added in a follow-up PR.
API
No public API changes.
Documentation
doc-neededdoc-not-neededdoc-included