Don't replace workers that exited normally#68
Merged
Conversation
reap_dead_workers now keeps normally-exited workers in the pool dict so they are still counted toward the target, preventing a reap-replace cycle that caused unbounded worker growth. The bug was introduced in funkelab#67. There is still no clean separation of failure and retry logic, since some crashes go through the try fail (if _spawn_wrapper catches the exception), while others are reaped here. But this fixes the immediate bug and the repear is still an improvement for terminations/kills not caught in python.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
reap_dead_workers now keeps normally exited workers in the pool dict
so they are still counted toward the target, preventing a reap replace
cycle that caused unbounded worker growth. The bug was introduced in
#67.
There is still an unclean separation of failure and retry logic, since
some crashes go through the retry path (if _spawn_wrapper catches the
exception), while others are reaped here. But this fixes the immediate
bug and the reaper is still an improvement for terminations/kills not
caught in python. There is a pre-existing bug that workers exiting with 0
(e.g., because of poor exception handling in some libraries) can eventually
fill the pool with None workers and stall. This does not alter behavior in that case.
LLM disclosure: while I diagnosed and debugged the root issue, LLMs were used to exclude other causes and draft fixes. All code here has been human reviewed or created.