[SPARK-55406][PYTHON][CONNECT] Reimplement the thread pool for ExecutePlanResponseReattachableIterator#54189
Open
gaogaotiantian wants to merge 7 commits intoapache:masterfrom
Open
[SPARK-55406][PYTHON][CONNECT] Reimplement the thread pool for ExecutePlanResponseReattachableIterator#54189gaogaotiantian wants to merge 7 commits intoapache:masterfrom
gaogaotiantian wants to merge 7 commits intoapache:masterfrom
Conversation
JIRA Issue Information=== Bug SPARK-55406 === This comment was automatically generated by GitHub Actions |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What changes were proposed in this pull request?
Reimplemented the thread pool mechanism for ExecutePlanResponseReattachableIterator.
ExecutePlanResponseReattachableIteratorinstances is counted withWeakSetso we know when to release the thread poolWhy are the changes needed?
The current implementation is fundamentally wrong. It tries to shutdown a thread pool for a class from client instances. We could have multiple client instances in a process and flipping the state of a global class would just cause racing issues.
For example, the old code can return a thread pool in a thread which could be shutdown before it is used.
The original reason that we "shutdown" the thread pool is because we need the client to wait for the release requests to finish. That can't justify waiting for the whole thread pool (or shut it down) because it can be shared with other clients. We actually have a deadlock caused by it.
Now we collect futures explicitly for each
ExecutePlanResponseReattachableIteratorcall and when we need to close the client, just wait for those future to be done.Notice that we will not keep these futures longer than they should be. We used
WeakSetwhich means if the future has no reference, it will still be garbage collected properly. We won't use more memory because we keep track of these. In normal cases the set will probably be empty because all the release tasks are done.Does this PR introduce any user-facing change?
No.
How was this patch tested?
Simple local test passed. Waiting for CI.
Was this patch authored or co-authored using generative AI tooling?
No.