Unstable Jenkins Build

[krumts]

The jenkins build on the eclipse CI is frequently failing while running the MAT tests with an error like this one:

[ERROR] Failed to execute goal org.eclipse.tycho:tycho-surefire-plugin:4.0.9:test (default-test) on project org.eclipse.mat.tests: An unexpected error occurred while launching the test runtime (process returned error code 137(SIGKILL received?)). The process logfile /home/jenkins/agent/workspace/tycho-mat-nightly/plugins/org.eclipse.mat.tests/target/work/data/.metadata/.log might contain further details. Command-line used to launch the sub-process was /opt/tools/java/openjdk/jdk-17/latest/bin/java -Dosgi.noShutdown=false -Dosgi.os=linux -Dosgi.ws=gtk -Dosgi.arch=x86_64 -ea -Xmx1536m --add-exports=java.base/jdk.internal.org.objectweb.asm=ALL-UNNAMED -Dosgi.clean=true -jar /home/jenkins/.m2/repository/p2/osgi/bundle/org.eclipse.equinox.launcher/1.6.900.v20240613-2009/org.eclipse.equinox.launcher-1.6.900.v20240613-2009.jar -data /home/jenkins/agent/workspace/tycho-mat-nightly/plugins/org.eclipse.mat.tests/target/work/data -install /home/jenkins/agent/workspace/tycho-mat-nightly/plugins/org.eclipse.mat.tests/target/work -configuration /home/jenkins/agent/workspace/tycho-mat-nightly/plugins/org.eclipse.mat.tests/target/work/configuration -application org.eclipse.tycho.surefire.osgibooter.headlesstest -testproperties /home/jenkins/agent/workspace/tycho-mat-nightly/plugins/org.eclipse.mat.tests/target/surefire.properties in working directory /home/jenkins/agent/workspace/tycho-mat-nightly/plugins/org.eclipse.mat.tests -> [Help 1]

From what I've seen in various scenarios and what I find when looking for the error, "process returned error code 137(SIGKILL received?)" should indicate that the test process was killed because of memory pressure in the environment (vm? / container?) and not because of an OOM in java.

I have no idea if we could influence the limits in the Eclipse CI environment. For now I'd try to add a change which makes the command line for the tests more configurable and would check if running the tests with a smaller value in Xmx would secceed (we still need to have enough java heap to get the tests trough).

Any other ideas are welcome!

[krumts]

I found this tip here, but I have no idea how/where to specify the mentioned additional configuration.
https://github.com/eclipse-cbi/jiro/wiki/Common-build-errors#what-is-killing-my-build-im-using-custom-containers

[krumts]

Running the tests with -Xmx1024m instead of -Xmx1536 resulted in a green build. However, I don't know yet how reproducible this is (I got some green builds last month also with the old settings), so I'll keep this issue open.

The recommendation I pasted in the previous comment appears to work only if jenkins pipelines are used, which we don't do. It might be worth spending some time on it, if I recall correctly it would enable us also to do the PR-builds. But I don't have experience with jenkins Pipelines and I can't tell when I'll get the time to look into it. If some of the other MAT committers has more experience there, help would be appreciated.

For now I hope that the new settings will lead to a more stable build. I don't have a bad feeling about the quality, as local builds are always succeeding - seems that we have a challenge specific to the CI environment.