[SPARK-50909][PYTHON][4.0] Setup faulthandler in PythonPlannerRunners #49635
Conversation
This file contains 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
### What changes were proposed in this pull request? Setups `faulthandler` in `PythonPlannerRunner`s. It can be enabled by the same config as UDFs. - SQL conf: `spark.sql.execution.pyspark.udf.faulthandler.enabled` - It fallback to Spark conf: `spark.python.worker.faulthandler.enabled` - `False` by default ### Why are the changes needed? The `faulthandler` is not set up in `PythonPlannerRunner`s. ### Does this PR introduce _any_ user-facing change? When enabled, if Python worker crashes, it may generate thread-dump in the error message on the best-effort basis of Python process. ### How was this patch tested? Added the related tests. ### Was this patch authored or co-authored using generative AI tooling? No. Closes apache#49592 from ueshin/issues/SPARK-50909/faulthandler. Authored-by: Takuya Ueshin <[email protected]> Signed-off-by: Takuya Ueshin <[email protected]>
HyukjinKwon
approved these changes
Jan 24, 2025
There was a problem hiding this comment.
Since one of the failures is Python part, could you re-trigger once more, @ueshin ?
|
Oh, according to the log,
|
|
@dongjoon-hyun Actually the ml related tests can't pass on the GHA in my repo for some reason, as I also mentioned before. |
|
Oh, for me, it works fine until now, even in my Today's PR.
I also don't have a special configuration. It's a simple fork like the other contributors. Do you happen to know someone who has the same symptom? |
|
Trying on master with an empty commit at #49664. I remember @HyukjinKwon said it sometimes doesn't pass in his repo as well. sometimes he said vs. always in my repo, though. |
|
yeah. it's known. I have seen those test failures ONLY in fork repos.. |
|
Let's merge this in first. I will monitor the build. |
HyukjinKwon
pushed a commit
that referenced
this pull request
Jan 25, 2025
### What changes were proposed in this pull request? This is a backport of #49592. Setups `faulthandler` in `PythonPlannerRunner`s. It can be enabled by the same config as UDFs. - SQL conf: `spark.sql.execution.pyspark.udf.faulthandler.enabled` - It fallback to Spark conf: `spark.python.worker.faulthandler.enabled` - `False` by default ### Why are the changes needed? The `faulthandler` is not set up in `PythonPlannerRunner`s. ### Does this PR introduce _any_ user-facing change? When enabled, if Python worker crashes, it may generate thread-dump in the error message on the best-effort basis of Python process. ### How was this patch tested? Added the related tests. ### Was this patch authored or co-authored using generative AI tooling? No. Closes #49635 from ueshin/issues/SPARK-50909/4.0/faulthandler. Lead-authored-by: Takuya Ueshin <[email protected]> Co-authored-by: Takuya UESHIN <[email protected]> Signed-off-by: Hyukjin Kwon <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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?
This is a backport of #49592.
Setups
faulthandlerinPythonPlannerRunners.It can be enabled by the same config as UDFs.
spark.sql.execution.pyspark.udf.faulthandler.enabledspark.python.worker.faulthandler.enabledFalseby defaultWhy are the changes needed?
The
faulthandleris not set up inPythonPlannerRunners.Does this PR introduce any user-facing change?
When enabled, if Python worker crashes, it may generate thread-dump in the error message on the best-effort basis of Python process.
How was this patch tested?
Added the related tests.
Was this patch authored or co-authored using generative AI tooling?
No.