Speed up Spark + Flink unit test execution #10581
Open
+66
−23
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
|
CI failure is unrelated, reported via #10599 |
snazy
force-pushed
the
spark-flink-test-forks
branch
from
fe65a4b
to
900a001
Compare
`test` task execution for Spark and Flink is rather slow. Gradle allows forking multiple JVMs to parallelize test execution, test _classes_ are distributed amount the available test worker JVMs.
This change allows more than one parallel fork (test worker JVM). The maximum number of workers is calculated like this:`max(min(Integer.getInteger("iceberg.maxSparkTestParallelism", 2), Runtime.runtime.availableProcessors() / 2), 1)`.
The default max settings for Spark and Flink are configured in `gradle.properties` to `2`, but this can be overridden in `~/.gradle/gradle.properties`.
This change does not affect CI, because in CI the number of "CPUs" on GitHub free hosted runners is 2, divided by 2 = 1.
snazy
force-pushed
the
spark-flink-test-forks
branch
from
900a001
to
fc2bc7e
Compare
|
This change might be controversial, because at least for Spark there's risk that test workers share the same resources on the file system. |
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Outdated suggestions cannot be applied.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.
testtask execution for Spark and Flink is rather slow. Gradle allows forking multiple JVMs to parallelize test execution, test classes are distributed amount the available test worker JVMs.This change allows more than one parallel fork (test worker JVM). The maximum number of workers is calculated like this:
max(min(Integer.getInteger("iceberg.maxSparkTestParallelism", 2), Runtime.runtime.availableProcessors() / 2), 1).The default max settings for Spark and Flink are configured in
gradle.propertiesto2, but this can be overridden in~/.gradle/gradle.properties.This change does not affect CI, because in CI the number of "CPUs" on GitHub free hosted runners is 2, divided by 2 = 1.