Ansible request for Linux core settings on docker test hosts

[sxa]

Please put the name of the software product (and affected platforms if relevant) in the title of this issue

• Missing install
• Bug in ansible playbook
• Request for new playbook addition

Details: The default setting in /proc/sys/kernel/core_pattern on Ubuntu is to have it set to |/usr/share/apport/apport %p %s %c %d %P %E. This prevents processes dumping core in docker by default. A fix is to set it to core which appears to cause the behavior we want. We should probably understand the implications and any side effects of doing this before implementing it, but it gives us a viable solution that I've implemented on three machines I've been experimenting with as part of #1809 (test-ibmcloud-ubuntu1604-x64-1, test-packet-ubuntu1804-armv8-1, and the new 20.04 host I'm working with)

If added to the playbooks it will need to be done in a way that persists across reboots (what I've done on the two referenced machines will not at the moment, so it will get lost if they are restarted)

[andrew-m-leonard]

Looks like this may have fixed an issue I was seeing on node test-docker-fedora33-ppc64le-1 with: https://ci.adoptopenjdk.net/job/Test_openjdk11_j9_sanity.functional_ppc64le_linux/99/consoleFull
failing to produce dumps,

11:59:48  Testing: Create core file
11:59:48  Test start time: 2021/04/13 10:59:48 Coordinated Universal Time
11:59:49  Running command: "/home/jenkins/workspace/Test_openjdk11_j9_sanity.functional_ppc64le_linux/openjdkbinary/j2sdk-image/bin/java"  -Xjit -Xgcpolicy:gencon -Xnocompressedrefs  -Xmx256m -cp /home/jenkins/workspace/Test_openjdk11_j9_sanity.functional_ppc64le_linux/openjdk-tests/TKG/../../jvmtest/functional/cmdline_options_testresources/cmdlinetestresources.jar:/home/jenkins/workspace/Test_openjdk11_j9_sanity.functional_ppc64le_linux/openjdk-tests/TKG/../../jvmtest/functional/cmdLineTests/gcCheck/gcCheck.jar -Xdump:system:file=j9core.dmp,events=vmstop,request=exclusive+compact org.openj9.test.nogc.Main
11:59:49  Time spent starting: 53 milliseconds
11:59:56  Time spent executing: 6279 milliseconds
11:59:56  Test result: FAILED
11:59:56  Output from test:
11:59:56   [ERR] Error: Could not find or load main class org.openj9.test.nogc.Main
11:59:56   [ERR] Caused by: java.lang.ClassNotFoundException: org.openj9.test.nogc.Main
11:59:56   [ERR] JVMDUMP039I Processing dump event "vmstop", detail "#0000000000000001" at 2021/04/13 10:59:49 - please wait.
11:59:56   [ERR] JVMDUMP032I JVM requested System dump using '/home/jenkins/workspace/Test_openjdk11_j9_sanity.functional_ppc64le_linux/openjdk-tests/TKG/output_16183033589351/cmdLineTester_GCCheck_0/j9core.dmp' in response to an event
11:59:56   [ERR] JVMPORT030W /proc/sys/kernel/core_pattern setting "|/usr/share/apport/apport %p %s %c %d %P %E" specifies that the core dump is to be piped to an external program.  Attempting to rename either core or core.3103966.
11:59:56   [ERR] 
11:59:56   [ERR] JVMDUMP012E Error in System dump: The core file created by child process with pid = 3103966 was not found. Expected to find core file with name "/home/jenkins/workspace/Test_openjdk11_j9_sanity.functional_ppc64le_linux/openjdk-tests/TKG/output_16183033589351/cmdLineTester_GCCheck_0/core"
11:59:56   [ERR] JVMDUMP013I Processed dump event "vmstop", detail "#0000000000000001".
11:59:56  >> Success condition was not found: [Output match: System dump written]

seems to work now re-running via a Grinder: https://ci.adoptopenjdk.net/job/Grinder/48/

[sxa]

Resolved on that host (Until next reboot ...)