[VL] Split CI spark unit tests and TPC-DS random-kill job across parallel runners - #12624
Merged
Conversation
zhouyuan
force-pushed
the
wip_test_split_tests
branch
2 times, most recently
from
July 26, 2026 02:40
f72bb33 to
2fbc2ba
Compare
zhouyuan
marked this pull request as ready for review
July 27, 2026 08:29
zhouyuan
changed the title
…llel runners
The Spark unit test jobs (spark-3.4 through spark-4.1) and the TPC-DS
random-kill job were running on a single runner each, taking up to ~2 hours
per job.
Changes:
- Split spark unit tests for spark34/spark35/spark35-scala213/spark40/spark41
into 3 parallel runner groups using matrix strategy, partitioned by package
prefix via -DwildcardSuites:
- Group 1: org.apache.spark.sql.streaming, org.apache.spark.GlutenSortShuffleSuite,
org.apache.gluten
- Group 2: org.apache.spark.sql.execution, org.apache.spark.sql.catalyst,
org.apache.spark.sql.errors, org.apache.spark.sql.extension
- Group 3: org.apache.spark.sql.GlutenSQL*, org.apache.spark.sql.connector,
org.apache.spark.sql.sources, org.apache.spark.sql.hive,
org.apache.spark.sql.gluten, org.apache.spark.sql.shim
- Split the slow-test jobs (extended + hive) for each spark version into 2
parallel runners.
- Split tpc-test-ubuntu-randomkill into 3 parallel runners using the built-in
--shard=N/M option in gluten-it, each runner handling ~33 of the 99 TPC-DS
queries.
- Increase executor memory in DynamicOffHeapSizingSuite to reduce flakiness.
Note: scalatest-maven-plugin uses -DwildcardSuites (not -Dtest) for suite
selection via package prefix.
Signed-off-by: Yuan <yuanzhou@apache.org>
Signed-off-by: Yuan <yuanzhou@apache.org>
zhouyuan
force-pushed
the
wip_test_split_tests
branch
from
July 27, 2026 13:19
33f8f41 to
ee59b2b
Compare
|
Run Gluten Clickhouse CI on x86 |
zhouyuan
changed the title
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
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 are proposed in this pull request?
The Spark unit test jobs (spark-3.4 through spark-4.1) and the TPC-DS
random-kill job were running on a single runner each, taking up to ~2 hours
per job. With this patch the longest task would take ~1hour.
Changes:
into 3 parallel runner groups using matrix strategy, partitioned by package
prefix via -DwildcardSuites:
org.apache.gluten
org.apache.spark.sql.errors, org.apache.spark.sql.extension
org.apache.spark.sql.sources, org.apache.spark.sql.hive,
org.apache.spark.sql.gluten, org.apache.spark.sql.shim
parallel runners.
--shard=N/M option in gluten-it, each runner handling ~33 of the 99 TPC-DS
queries.
Note: scalatest-maven-plugin uses -DwildcardSuites (not -Dtest) for suite
selection via package prefix.
How was this patch tested?
pass GHA
Was this patch authored or co-authored using generative AI tooling?