Improve worker pool reservation to avoid starvation #36
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Outdated suggestions cannot be applied.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.
During Elasticsearch indexing, when a batch indexing error occurs we retry with exponential backoff. The worker pool checkout/checkin was managed outside of the block with the backoff
sleep()
. This caused a condition where if lots of bad documents are arriving and causing batch errors, lots of workers will be unavailable because they are doing backoff sleeps, and worker pool exhaustion is likely.This change attempts to address that by moving the worker pool checkout/checkin closer to the batch indexing API invocation, which is appropriate as the worker pool is intended to bound the concurrency that Elasticsearch sees.