Open andrew-m-leonard opened 4 years ago
1TISIGINFO Dump Event "systhrow" (00040000) Detail "java/lang/OutOfMemoryError" "Java heap space" received
3STHSTTYPE 02:45:26:489683997 GMT j9mm.101 - J9AllocateIndexableObject() returning NULL! 268435464 bytes requested for object of class 000000000002E000 from memory space 'Generational' id=0000004E134CC340
javacore is truncated so there is no stack trace. Core/heap dumps are disabled by default for these openjdk tests, because some of them intentionally create OOM.
The test is run with -Xmx512m, which perhaps just isn't enough. Not sure why this is set, although Hotspot runs with the same. https://github.com/AdoptOpenJDK/openjdk-tests/blob/master/openjdk/playlist.xml#L292
Checking if I can recreate in a grinder 5x https://ci.eclipse.org/openj9/view/Test/job/Grinder/922/ - passed
Actually, the test is supposed to handle OOM. I think it's hanging trying to write the javacore, there is no completion message.
JVMDUMP039I Processing dump event "systhrow", detail "java/lang/OutOfMemoryError" at 2020/06/23 21:45:26 - please wait.
JVMDUMP032I JVM requested Java dump using 'C:\\Users\\jenkins\\workspace\\Test_openjdk11_j9_sanity.openjdk_x86-64_windows\\openjdk-tests\\TKG\\test_output_15929643393679\\jdk_math_j9_0\\work\\scratch\\javacore.20200623.214526.30020.0001.txt' in response to an event
This is a tricky one, probably requiring manual intervention to diagnose. When the test is hanging, use task manager to "create dump file". Or connect via the debugger and get a stack trace to see where it's hanging, but there may be no symbols, as the debug image is likely required.
@andrew-m-leonard I wonder if it's a machine problem. I haven't seen this problem on the OpenJ9 machines.
It's happened on at least 2 different machines
Failure link
https://ci.adoptopenjdk.net/job/Test_openjdk11_j9_sanity.openjdk_x86-64_windows/251/testReport/junit/java_math_BigInteger_DivisionOverflow/java/DivisionOverflow/ test-softlayer-win2012r2-x64-1
Optional info