Skip to content

Conversation

@gareth-ellis
Copy link
Member

Relying on a timeout in the force merge runner can result in other errors - e.g if a load balancer kills a connection, typically we would see an SSL related error instead. This change changes polling method to use wait_for_completion:False instead and then polls via the task API.

@gareth-ellis gareth-ellis requested a review from a team December 15, 2025 10:41
@gareth-ellis gareth-ellis changed the title Use wait_for_completion:False instead of relying on timeouts Use wait_for_completion:False instead of relying on timeouts in ForceMerge Dec 15, 2025
Copy link
Contributor

@fressi-elastic fressi-elastic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please have a look to my comments. Overall it LGTM

Copy link
Contributor

@fressi-elastic fressi-elastic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Resulting unit test failures looks related. Please fix it.

Copy link
Contributor

@gbanasiak gbanasiak left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM.

When run with wait_for_completion=true, force merge returns task ID. This task ID could be used to filter the list of force merge tasks in case there's parallel unrelated force merge running in the cluster. WDYT about adding it as part of this PR?

@gareth-ellis gareth-ellis enabled auto-merge (squash) December 18, 2025 10:15
@gareth-ellis gareth-ellis merged commit c04cac3 into elastic:master Dec 18, 2025
15 checks passed
@gareth-ellis gareth-ellis deleted the force-merge-polling branch December 18, 2025 13:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants