I simulated the ordering of the tests as it was in the failing CI job - you can check here, it's hacky way: https://github.com/ochaloup/wildfly/tree/WFLY-9871-reordering-testcases for decreasing number of thread to be used the io subsystem could be changed. <subsystem xmlns= "urn:jboss:domain:io:2.0" > <worker name= " default " task-max-threads= "3" io-threads= "3" /> <buffer-pool name= " default " /> </subsystem> Tried to run with elytron as it was the case where intermittent failures happened. No idea about relationship of the issue and the elytron. cd testsuite/integration/xts mvn test -Delytron