You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
During the reconfiguration test case the test code changes the config of the keystone service via the CRD then gets the keystone pods with oc get pods -n openstack -l service=keystone -o name Due to the reconfiguration the keystone service pod is restarted. So there is a chance that the still terminating old Pod is returned by oc get pods -n openstack -l service=keystone -o name see below in the logs.
So this is probably a race condition between the test changing the CRD and the operator + k8s reacting to that change and replacing the Pod. Eventually the only running pod will be the one with the new configuration. But the cluster might need time to get there. More time than what the kuttl test takes to assert that the reconfiguration take effect on the Pod. So in theory the above code can lead to the case where the kuttl test fails the assert as it still sees the old Pod only as the cluster haven't reacted to the CRD changes yet.
In this particular case the race lead to false positive assert. The oc logs command returned an error due to multiple pod names are passed to it, but the assert script exited with 0 status.
During the reconfiguration test case the test code changes the config of the keystone service via the CRD then gets the keystone pods with
oc get pods -n openstack -l service=keystone -o name
Due to the reconfiguration the keystone service pod is restarted. So there is a chance that the still terminating old Pod is returned byoc get pods -n openstack -l service=keystone -o name
see below in the logs.The text was updated successfully, but these errors were encountered: