test-aws-rhel8-x64-1 - tests which create core files fail

[lumpfish]

It looks like the openj9 functional tests which create a j9core.dmp or other core file and then examine the contents of the file are failing to create the core dump.

Example job link: https://ci.adoptopenjdk.net/job/Test_openjdk8_j9_sanity.functional_x86-64_linux/71/consoleFull

The failing test targets are

21:20:34  FAILED test targets:
21:20:34  	cmdLineTester_callsitedbgddrext_openj9_0
21:20:34  	cmdLineTester_pltest_0

The cmdLineTester_callsitedbgddrext_openj9_0 test failure message is

19:50:25  Output from test:
19:50:25   [OUT] DTFJView version 4.29.5, using DTFJ version 1.12.29003
19:50:25   [OUT] Loading image from DTFJ...
19:50:25   [OUT] 
19:50:25   [OUT] Could not load dump file and/or could not load XML file: Image file '/home/jenkins/workspace/Test_openjdk8_j9_sanity.functional_x86-64_linux/openjdk-tests/TKG/test_output_16103945291575/cmdLineTester_callsitedbgddrext_openj9_0/j9core.dmp' not found.
19:50:25   [OUT] For a list of commands, type "help"; for how to use "help", type "help help"
19:50:25   [OUT] > DDR is not enabled for this core file, '!' commands are disabled
19:50:25   [OUT] > 
19:50:25  >> Success condition was not found: [Output match: jvminit.c]

The cmdLineTester_pltest_0 test failure message is

20:13:11   [ERR] ----------------------------------------
20:13:11   [ERR] dump tests
20:13:11   [ERR] ----------------------------------------
20:13:11   [ERR] 
20:13:11   [ERR] 
20:13:11   [ERR] Starting test j9dump_verify_functiontable_slots
20:13:11   [ERR] Ending test j9dump_verify_functiontable_slots
20:13:11   [ERR] 
20:13:11   [ERR] 
20:13:11   [ERR] Starting test j9dump_test_create_dump_with_name
20:13:11   [ERR]   calling j9dump_create with filename: /home/jenkins/workspace/Test_openjdk8_j9_sanity.functional_x86-64_linux/openjdk-tests/TKG/test_output_16103945291575/cmdLineTester_pltest_0/j9dump_test_create_dump_with_name
20:13:11   [ERR] JVMPORT030W 
20:13:11   [ERR] j9dumpTest.c line  213: j9dump_test_create_dump_with_name j9dump_create returned: 1, with filename: The core file created by child process with pid = 3757128 was not found. Expected to find core file with name "/home/jenkins/workspace/Test_openjdk8_j9_sanity.functional_x86-64_linux/openjdk-tests/TKG/test_output_16103945291575/cmdLineTester_pltest_0/core.3757128"
20:13:11   [ERR] 
20:13:11   [ERR] 		LastErrorNumber: -108
20:13:11   [ERR] 		LastErrorMessage: No such file or directory
20:13:11   [ERR] 
20:13:11   [ERR] Ending test j9dump_test_create_dump_with_name
20:13:11   [ERR] 
20:13:11   [ERR] 
20:13:11   [ERR] Starting test j9dump_test_create_dump_from_signal_handler
20:13:11   [ERR]   calling j9dump_create with filename: /home/jenkins/workspace/Test_openjdk8_j9_sanity.functional_x86-64_linux/openjdk-tests/TKG/test_output_16103945291575/cmdLineTester_pltest_0/j9dump_test_create_dump_from_signal_handler
20:13:11   [ERR] JVMPORT030W 
20:13:11   [ERR] j9dumpTest.c line  311: j9dump_test_create_dump_from_signal_handler j9dump_create returned: 1, with filename: The core file created by child process with pid = 3757143 was not found. Expected to find core file with name "/home/jenkins/workspace/Test_openjdk8_j9_sanity.functional_x86-64_linux/openjdk-tests/TKG/test_output_16103945291575/cmdLineTester_pltest_0/core.3757143"
20:13:11   [ERR] 		LastErrorNumber: -108
20:13:11   [ERR] 		LastErrorMessage: No such file or directory
20:13:11   [ERR] 
20:13:11   [ERR] Ending test j9dump_test_create_dump_from_signal_handler
20:13:11   [ERR] 
20:13:11   [ERR] 
20:13:11   [ERR] Starting test j9dump_test_create_dump_with_NO_name
20:13:11   [ERR]   calling j9dump_create with empty filename
20:13:11   [ERR] JVMPORT030W 
20:13:11   [ERR] j9dumpTest.c line  153: j9dump_test_create_dump_with_NO_name j9dump_create returned: 1, with filename: The core file created by child process with pid = 3757157 was not found. Expected to find core file with name "core.3757157"
20:13:11   [ERR] 		LastErrorNumber: -108
20:13:11   [ERR] 		LastErrorMessage: No such file or directory
20:13:11   [ERR] 
20:13:11   [ERR] Ending test j9dump_test_create_dump_with_NO_name
20:13:11   [ERR] 
20:13:11   [ERR] Dump test done, failures detected.

The same tests pass on other machines - e.g. https://ci.adoptopenjdk.net/job/Test_openjdk8_j9_sanity.functional_x86-64_linux_xl/72/consoleFull

[sxa]

I suspect they're ending up in /var/lib/systemd/coredump/ as defined in /proc/sys/kernel/core_pattern ... Depends if we want to adjust the default on the system or handle the situation differently (We may well want to adjust the system default if only because they're in a different format - LZ4 compressed)

[sxa]

I have made an adjustment to /etc/sysctl.conf to set kernel.core_pattern = core.%p
Running a test at https://ci.adoptopenjdk.net/job/Test_openjdk8_j9_sanity.functional_x86-64_linux/78
Will need to ensure that's the best place to set it and ensure we add it to the playbooks if we choose to go ahead with this as a final solution, but I'm a little reluctant to reset system defaults like this particuarly if it changes the behaviour of someone's system when they run our playbooks so I may put this in but guarded by the adoptopenjdk tag

FYA @smlambert

[sxa]

(Job above succeeded so that resolves the issue)

[lumpfish]

Closing - these tests are now passing on test-aws-rhel8-x64-1.

[sxa]

As per statement above, I do not yet considered this closed as the playbooks have (to my knowledge) not been updated to handle this.