Open pshipton opened 5 years ago
@Mesbah-Alam can you please take a look.
The tests timed out, collected some diagnostics, and then finished afterwards. It seems the timeout occurred about 15 seconds after the tests started.
Another one here with a 4 min timeout https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_x86-64_linux_Nightly/640
and with a less than 3 min timeout https://ci.eclipse.org/openj9/job/Test_openjdk13_j9_sanity.system_x86-64_linux_Nightly/87
More: https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_sanity.system_x86-64_windows_Nightly/184 https://ci.eclipse.org/openj9/job/Test_openjdk13_j9_sanity.system_ppc64_aix_Nightly/6
I get it, I believe this was caused by the daylight savings time change that occurred. What is the code used to determine if a timeout occurs?
A human readable timeout value (e.g. 60m
) is provided in the test , such as https://github.com/AdoptOpenJDK/openjdk-systemtest/blob/master/openjdk.test.jlm/src/test.jlm/net/adoptopenjdk/test/jlm/TestJlmRemoteMemoryAuth.java#L183
Then that value gets parsed by STF here: https://github.com/AdoptOpenJDK/stf/blob/master/stf.core/src/stf.core/net/adoptopenjdk/stf/util/TimeParser.java#L40
It is possible for stf to mis-calculate the time for tests that start before the daylight saving time change takes place and are long enough to continue on after the change happens.
It is possible for stf to mis-calculate the time for tests that start before the daylight saving time change takes place and are long enough to continue on after the change happens.
Right. I'm wondering if the code can be improved so this problem can't occur. Otherwise this problem will occur every year until we get rid of daylight savings time changes. There are some java APIs which aren't affected by clock changes.
There are some java APIs which aren't affected by clock changes
I will look into it.
Thanks for discovering the issue.
FYI @lumpfish
https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_s390x_linux_Nightly/635 DaaLoadTest_all_special_27 variation: Mode612-OSRG JVM_OPTIONS: -Xcompressedrefs -Xcompressedrefs -Xgcpolicy:gencon -Xjit:enableOSR,enableOSROnGuardFailure,count=1,disableAsyncCompilation
1.DLT.stdout shows both threads completed testing.
However a timeout occurred.
Looking at the javacore's collected, tests are still running. There is movement between the first javacore and the second.