Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Parallelized test execution #3057

Draft
wants to merge 7 commits into
base: main
Choose a base branch
from

Conversation

alecgrieser
Copy link
Collaborator

This updates the test runner to work in parallel in an attempt to speed up the build. The relational tests do not currently support it, and there are sometimes issues with deadline exceeded exceptions, especially at higher parallelisms, and that still needs to be worked out.

@foundationdb-ci
Copy link
Contributor

Result of fdb-record-layer-pr on Linux CentOS 7

  • Commit ID: 0485336
  • Duration 0:22:57
  • Result: ❌ FAILED
  • Error: Error while executing command: ./gradlew --no-daemon --console=plain -b ./build.gradle build destructiveTest -PcoreNotStrict -PreleaseBuild=false -PpublishBuild=false -PspotbugsEnableHtmlReport. Reason: exit status 1
  • Build Log terminal output (available for 30 days)
  • Build Workspace zip file of the working directory (available for 30 days)

alecgrieser and others added 5 commits January 27, 2025 11:24
This begins the process of allowing for parallel test execution. A few issues were noticed that required some changes:

* Some tests were taking up a bunch of time and benefitted from parallelized execution of tests within the fixture. Those have been updated to CONCURRENT execution mode
* A known issue with JUnit (junit-team/junit5#3108) means that if one of the tests involves a future waiting, that can look to the `ForkJoinPool` like the thread is available for work stealing, so too many tests can end up being executed at once. A new test extension was added that has a semaphore, and that appears to be enough to stop extra tests from being executed
* The server was running out of batch GRV transactions, which resulted in tests failing with "batch GRV transactions exhausted". This mainly affected indexing tests, and I was able to resolve this by upping the transaction priority, for better or worse

There are still some issues:

* A number of tests were hitting deadline exceeded exceptions. It looked like some kind of weird concurrency stuff may be going on, because there were things happening like key space path resolution while creating a record store would include stack traces from closing the test key space path manager, which seems like things were sharing objects that shouldn't have been. This affected both the `:fdb-record-layer-core:test` and `:fdb-lucene:test` tasks
* The relational layer tests are not structured to allocate unique key spaces for each test, so they immediatetly hit concurrency problems when run in a parallelized manner
In addition to adding the new class, it reworks the class
hierarchy so the tests are in a `YamlIntegrationTests` class, that
the other classes extend from, and `JDBCYamlIntegrationTests` which
disables the tests that use features unsupported by JDBC.
@alecgrieser alecgrieser force-pushed the parllelized-relational-tests branch from 0485336 to 951a9df Compare January 27, 2025 11:25
@foundationdb-ci
Copy link
Contributor

Result of fdb-record-layer-pr on Linux CentOS 7

  • Commit ID: 951a9df
  • Duration 0:45:47
  • Result: ✅ SUCCEEDED
  • Error: N/A
  • Build Log terminal output (available for 30 days)
  • Build Workspace zip file of the working directory (available for 30 days)

@foundationdb-ci
Copy link
Contributor

Result of fdb-record-layer-pr on Linux CentOS 7

  • Commit ID: fa99246
  • Duration 0:22:02
  • Result: ❌ FAILED
  • Error: Error while executing command: ./gradlew --no-daemon --console=plain -b ./build.gradle build destructiveTest -PcoreNotStrict -PreleaseBuild=false -PpublishBuild=false -PspotbugsEnableHtmlReport. Reason: exit status 1
  • Build Log terminal output (available for 30 days)
  • Build Workspace zip file of the working directory (available for 30 days)

@foundationdb-ci
Copy link
Contributor

Result of fdb-record-layer-pr on Linux CentOS 7

  • Commit ID: 91033dc
  • Duration 0:24:18
  • Result: ❌ FAILED
  • Error: Error while executing command: ./gradlew --no-daemon --console=plain -b ./build.gradle build destructiveTest -PcoreNotStrict -PreleaseBuild=false -PpublishBuild=false -PspotbugsEnableHtmlReport. Reason: exit status 1
  • Build Log terminal output (available for 30 days)
  • Build Workspace zip file of the working directory (available for 30 days)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants