See: https://jenkins.mw.lab.eng.bos.redhat.com/hudson/view/EAP6/view/EAP6-AS-Testsuite/job/eap-61-as-testsuite-Win-matrix-OracleJDK6/55/jdk=java16_default,label_exp=w2k8%26%26x86/console https://jenkins.mw.lab.eng.bos.redhat.com/hudson/view/EAP6/view/EAP6-AS-Testsuite/job/eap-61-as-testsuite-Win-matrix-OracleJDK6/58/jdk=java16_default,label_exp=w2k8%26%26x86/console
It's highly unlikely that this particular test case hangs just by itself. Looking at the output, it's more like the previous test case (AccessConstraintUtilizationTestCase) that failed didn't tear down correctly. Will look into it. Given that I wrote the test, this is devel_ack+ as far as I'm concerned. Can't do that for real, obviously :-)
OK, I was wrong, I didn't write AccessConstraintUtilizationTestCase. But I'll still look into it.
OK, so here's what's going on. In both cases, some other test failed to tear down correctly. This can be inferred from: 1. Looking at failing tests that precede the RBAC TS. In both cases, IIOP tests fail because some other server is running and they have Arquillian configured to disallow this situation. The error messages in IIOP tests are pretty clear about this. 2. Looking at the failing test in the RBAC TS in second case. The test fails because it can't find a management resource /core-service=management/access=authorization/constraint=sensitivity-classification/type=messaging -- this can only happen when the server doesn't have the messaging subsystem. The RBAC TS uses standalone-full profile, which has messaging enabled, so there must be some server already running in wrong profile. The purpose of this bug is for the CliInterfaceStandardRolesBasicTestCase test to fail fast and not try to wait indefinitely for something that's not gonna happen.
I'm not sure if I found the same problem that caused this issue, but I did find an error in the implementation of Arquillian managed container in WildFly/EAP. In certain situations, similar to this one, the container startup timeout is prolonged cca 50x (default is 1 minute, so in the erroneous situation, it's 50 wasted minutes). Need to work on it upstream, not sure about hitting/missing 6.2.
Linking an upstream JIRA that I believe is essentially the same problem: https://issues.jboss.org/browse/WFLY-1458
This issue was not spotted in EAP 6.2.0.ER7.