-
Bug
-
Resolution: Done
-
Critical
-
OSSM 2.4.5, OSSM 2.5.0
-
False
-
None
-
False
-
-
Sometimes (usually in federation test when the whole test suite is running[1]) the cleanup step is stuck (a namespace is stuck in the terminating phase) and it is terminated after 1h timeout with causes test failures even though the test passes.
I have looked into the cluster and some namespace (usually west-mesh) was stuck in a terminating state because some finalizers blocked it from being deleted (service.kubernetes.io/load-balancer-cleanup). When I listed all remaining resources in that project, it was service/east-mesh-ingress. (The workaround is to delete the finalizer from that resource which unblocks deleting namespace but it is not a solution to what the test suite should do).
This is not happens every time and it happens only on PSI OpenStack with IPI OCP
Maybe the cleanup method should be more complex and if some resources need to be deleted before other resources (because it uses their finalizers), the order should be defined (just guessing).
=====
[1]
examples (see test results):
https://master-jenkins-csb-servicemesh.apps.ocp-c1.prod.psi.redhat.com/job/maistra/job/maistra-test-tool/1153/
https://master-jenkins-csb-servicemesh.apps.ocp-c1.prod.psi.redhat.com/job/maistra/job/maistra-test-tool/1154/
https://master-jenkins-csb-servicemesh.apps.ocp-c1.prod.psi.redhat.com/job/maistra/job/maistra-test-tool/1155/
- relates to
-
OSSM-5826 Create a workaround for stuck namespaces to the jenkins pipelines
- Closed
- mentioned on