Fix FileSettingsService hang on error update #89630
Merged
Conversation
This file contains 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
grcevski
added
>bug
:Core/Infra/Core
|
Pinging @elastic/es-core-infra (Team:Core/Infra) |
|
Hi @grcevski, I've created a changelog YAML for you. |
grcevski
commented
Aug 25, 2022
| @@ -186,8 +188,7 @@ static boolean isNewError(ReservedStateMetadata existingMetadata, Long newStateV | |||
| || existingMetadata.errorMetadata().version() < newStateVersion); | |||
| } | |||
|
|
|||
| private void saveErrorState(ErrorState errorState) { | |||
| ClusterState clusterState = clusterService.state(); | |||
| private void saveErrorState(ClusterState clusterState, ErrorState errorState) { | |||
There was a problem hiding this comment.
This change to pass in the most recent cluster state from the cluster state update task is sort of tangential, it just ensures we hit the issue more reliably. ClusterService.state() might not be the latest state, but the one we get in the update task will be.
grcevski
commented
Aug 25, 2022
| @@ -170,6 +170,8 @@ public void onFailure(Exception e) { | |||
| if (isNewError(existingMetadata, reservedStateVersion.version())) { | |||
| logger.debug("Failed to apply reserved cluster state", e); | |||
| errorListener.accept(e); | |||
| } else { | |||
| errorListener.accept(null); | |||
There was a problem hiding this comment.
This is the actual bug, we never called the callback to ensure the async watcher is released.
|
Thanks Chris! |
grcevski
added a commit
to grcevski/elasticsearch
that referenced
this pull request
Aug 29, 2022
💚 Backport successful
|
elasticsearchmachine
pushed a commit
that referenced
this pull request
Aug 29, 2022
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.
While working on new tests for #89567 I discovered a bug where on duplicate error update we don't unlock the file settings watcher thread properly.
It's a race condition where the settings file might be updated in succession quicker than the error state is updated. We check for duplicate error state and avoid the new cluster change update event, however we didn't call the error listener will null to unlock the watcher.
This is intermittent and should've been caught by our existing error state save integration tests, but just happened to be hit with the SLM specific test I was writing.