Added support for faster shard reallocation on graceful shutdown #1005
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Outdated suggestions cannot be applied.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.
Issue #, if available: #845
Description of changes:
We have this latency issue in production where on graceful shutdown we have seen up to 40s before records resume processing on a new container.
The cause of it is that on graceful shutdown, the lease stays owned. Others wait for lease to expire before picking them up. See #845 for details, this is pretty well explained.
The changes im proposing here are:
With these changes, I was able to go below 1s before records resumed consumption.
So I would like some opinions on this. We will test this in various envs and see the impacts. If this goes through this would be a contribution by Autodesk.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.