Fix Spark 3.5.0 shell classloader issue with the plugin #9500
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
Signed-off-by: Thomas Graves <[email protected]>
|
build |
abellina
previously approved these changes
Oct 20, 2023
gerashegalov
previously approved these changes
Oct 20, 2023
razajafri
approved these changes
Oct 20, 2023
razajafri
previously approved these changes
Oct 20, 2023
tgravescs
dismissed stale reviews from razajafri, gerashegalov, and abellina
The base branch was changed.
|
build |
|
I accidentally had this targetted for 23.12, changed to 23.10. @gerashegalov @abellina mind looking again |
gerashegalov
approved these changes
Oct 20, 2023
gerashegalov
added a commit
that referenced
this pull request
Oct 24, 2023
) Contributes to #5704 This PR aims to catch issues like #9500. It modifies run_pyspark_from_build mostly to avoid recreating the logic of figuring out jar location etc. Currently it may not catch this if do not have Spark 3.5.0 CI yet. But this is how it could reproduce the #9500 ```Bash $ SPARK_HOME=~/dist/spark-3.1.1-bin-hadoop3.2 SPARK_SHELL_SMOKE_TEST=1 ./integration_tests/run_pyspark_from_build.sh ... + grep -F 'res0: Array[org.apache.spark.sql.Row] = Array([4950])' res0: Array[org.apache.spark.sql.Row] = Array([4950]) + echo 'SUCCESS spark-shell smoke test...' SUCCESS spark-shell smoke test $ echo $? 0 $ SPARK_HOME=~/dist/spark-3.5.0-bin-hadoop3 SPARK_SHELL_SMOKE_TEST=1 ./integration_tests/run_pyspark_from_build.sh $ echo $? 1 SPARK_SHELL_SMOKE_TEST=1 \ PYSP_TEST_spark_shuffle_manager=com.nvidia.spark.rapids.spark311.RapidsShuffleManager \ SPARK_HOME=~/dist/spark-3.1.1-bin-hadoop3.2 \ ./integration_tests/run_pyspark_from_build.sh + echo 'SUCCESS spark-shell smoke test' SUCCESS spark-shell smoke test $ echo $? 0 SPARK_SHELL_SMOKE_TEST=1 \ PYSP_TEST_spark_shuffle_manager=com.nvidia.spark.rapids.spark350.RapidsShuffleManager \ SPARK_HOME=~/dist/spark-3.5.0-bin-hadoop3 \ ./integration_tests/run_pyspark_from_build.sh $ echo $? 1 ``` Signed-off-by: Gera Shegalov <[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.
fixes #9498
Spark 3.5.0 (apache/spark@1486835) changed the package name of the ExecutorClassLoader. Our ShimLoader code explicitly looks for it by the old package name so we have to update our ShimLoader to look for either name.
Manually tested with spark-shell on Spark 3.5.0 and older 3.3.0 and both worked.