| [jakartaee-tck-dev] Improving efficiency of TCK jobs and switching to build/run TCKs with https://www.eclipse.org/downloads/download.php?file=/ee4j/glassfish/glassfish-6.0.0-SNAPSHOT-nightly.zip |
Hi,The build/run TCK jobs [1][2][3] are now updated to use the nightly GlassFish build.
Regarding efficiency of TCK jobs, we saw a hint that our TCK jobs may not be using correct memory settings. The hint is that increasing the JNLP memory size from 2176Mi to 3Gi [4], allowed the Full Platform TCK run to completion. Previously, for several test runs the JPA tests (running only with the appmanagedNoTx vehicle) would unexpectedly fail to deploy correctly. This sounds to me like we could be experiencing thrashing of some sort.
How can we can properly ensure that for all launched JVM processes (especially long running ones), we don't use more memory than is allowed by the OS (vm/container) that we are running on.
So, how can we get OS level statistics for the jakartaee-tck Jenkins jobs? It likely involves contacting cbi-dev@xxxxxxxxxxx which is fine but I wanted to ask here first, to raise awareness and get advice.
Scott[1] https://ci.eclipse.org/jakartaee-tck/job/standalonetck-nightly-build-run-master
[2] https://ci.eclipse.org/jakartaee-tck/job/jakartaeetck-nightly-build-master
[3] https://ci.eclipse.org/jakartaee-tck/job/jakartaeetck-nightly-run-master [4] https://github.com/eclipse-ee4j/jakartaee-tck/pull/347