The shutdown thread may kill a wrong process in the ManagedContainer classes. If the managed container is killed in background (e.g. test kills the WildFly container for some reason) then stopping the container creates the shutdown thread which points to the non-existing (already stopped process).
https://github.com/wildfly/wildfly-arquillian/blob/2.2.0.Final/container-managed/src/main/java/org/jboss/as/arquillian/container/managed/ManagedDeployableContainer.java#L259
But as the container process does not exist anymore there is nobody interrupting the thread
https://github.com/wildfly/wildfly-arquillian/blob/2.2.0.Final/container-managed/src/main/java/org/jboss/as/arquillian/container/managed/ManagedDeployableContainer.java#L290
and just the catch clause is executed
https://github.com/wildfly/wildfly-arquillian/blob/2.2.0.Final/container-managed/src/main/java/org/jboss/as/arquillian/container/managed/ManagedDeployableContainer.java#L296

The shutdown thread waits till stop container timeout (ie. by default for 60 seconds) and then it destroys process. But the `process` instance may be already pointing to another instance of container which was started afterwards. Then such container is suddenly stopped without any reason

INFO  [org.jboss.as.server] (Thread-2) WFLYSRV0220: Server shutdown has been requested via an OS signal

This is copied from https://github.com/wildfly/wildfly-arquillian/issues/163