[SPARK-47819][CONNECT][3.5] Use asynchronous callback for execution cleanup #46064
Conversation
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
dongjoon-hyun
previously requested changes
Apr 15, 2024
Member
dongjoon-hyun
left a comment
dongjoon-hyun
left a comment
There was a problem hiding this comment.
If we want this in branch-3.5, we need to change the issue type first, @xi-db and @hvanhovell .
Contributor
|
@dongjoon-hyun I updated the ticket. I am a bit mystified by this though, this was never an issue before. This particular case can be both seen as a bug/improvement. |
Member
Thank you for updating.
I agree with you. We can choose our path.
|
xi-db
changed the title
hvanhovell
approved these changes
Apr 24, 2024
hvanhovell
pushed a commit
that referenced
this pull request
Apr 24, 2024
…leanup ([Original PR](#46027)) ### What changes were proposed in this pull request? Expired sessions are regularly checked and cleaned up by a maintenance thread. However, currently, this process is synchronous. Therefore, in rare cases, interrupting the execution thread of a query in a session can take hours, causing the entire maintenance process to stall, resulting in a large amount of memory not being cleared. We address this by introducing asynchronous callbacks for execution cleanup, avoiding synchronous joins of execution threads, and preventing the maintenance thread from stalling in the above scenarios. To be more specific, instead of calling `runner.join()` in `ExecutorHolder.close()`, we set a post-cleanup function as the callback through `runner.processOnCompletion`, which will be called asynchronously once the execution runner is completed or interrupted. In this way, the maintenance thread won't get blocked on joining an execution thread. ### Why are the changes needed? In the rare cases mentioned above, performance can be severely affected. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Existing tests and a new test `Async cleanup callback gets called after the execution is closed` in `ReattachableExecuteSuite.scala`. ### Was this patch authored or co-authored using generative AI tooling? No. Closes #46064 from xi-db/SPARK-47819-async-cleanup-3.5. Authored-by: Xi Lyu <[email protected]> Signed-off-by: Herman van Hovell <[email protected]>
turboFei
pushed a commit
to turboFei/spark
that referenced
this pull request
Nov 6, 2025
…leanup (apache#383) ([Original PR](apache#46027)) ### What changes were proposed in this pull request? Expired sessions are regularly checked and cleaned up by a maintenance thread. However, currently, this process is synchronous. Therefore, in rare cases, interrupting the execution thread of a query in a session can take hours, causing the entire maintenance process to stall, resulting in a large amount of memory not being cleared. We address this by introducing asynchronous callbacks for execution cleanup, avoiding synchronous joins of execution threads, and preventing the maintenance thread from stalling in the above scenarios. To be more specific, instead of calling `runner.join()` in `ExecutorHolder.close()`, we set a post-cleanup function as the callback through `runner.processOnCompletion`, which will be called asynchronously once the execution runner is completed or interrupted. In this way, the maintenance thread won't get blocked on joining an execution thread. ### Why are the changes needed? In the rare cases mentioned above, performance can be severely affected. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Existing tests and a new test `Async cleanup callback gets called after the execution is closed` in `ReattachableExecuteSuite.scala`. ### Was this patch authored or co-authored using generative AI tooling? No. Closes apache#46064 from xi-db/SPARK-47819-async-cleanup-3.5. Authored-by: Xi Lyu <[email protected]> Signed-off-by: Herman van Hovell <[email protected]> Co-authored-by: Xi Lyu <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
(Original PR)
What changes were proposed in this pull request?
Expired sessions are regularly checked and cleaned up by a maintenance thread. However, currently, this process is synchronous. Therefore, in rare cases, interrupting the execution thread of a query in a session can take hours, causing the entire maintenance process to stall, resulting in a large amount of memory not being cleared.
We address this by introducing asynchronous callbacks for execution cleanup, avoiding synchronous joins of execution threads, and preventing the maintenance thread from stalling in the above scenarios. To be more specific, instead of calling
runner.join()inExecutorHolder.close(), we set a post-cleanup function as the callback throughrunner.processOnCompletion, which will be called asynchronously once the execution runner is completed or interrupted. In this way, the maintenance thread won't get blocked on joining an execution thread.Why are the changes needed?
In the rare cases mentioned above, performance can be severely affected.
Does this PR introduce any user-facing change?
No.
How was this patch tested?
Existing tests and a new test
Async cleanup callback gets called after the execution is closedinReattachableExecuteSuite.scala.Was this patch authored or co-authored using generative AI tooling?
No.