Fix race condition when multiple nodes reconcile S3 snapshots #10979
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Outdated suggestions cannot be applied.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.
Proposed Changes
Fix race condition when multiple nodes reconcile S3 snapshots
Don't delete s3 etcdsnapshotfiles if they are missing from s3 but less than a minute old, its possible the other node just finished uploading it and the object key has not yet become visible.
This is most easily reproduced when an on-demand etcd snapshot is requested via Rancher, as it triggers simultaneous snapshots on all etcd nodes, and they will all try to reconcile s3 objects against the etcdsnapshotfile list at the same time despite not necessarily having a consistent view of the bucket.
Types of Changes
bugfix
Verification
See linked issue
Testing
Linked Issues
User-Facing Change
Further Comments