Description of problem: The containers started by CRI-O do not have dependencies that ensure correct ordering at shutdown as a result of which system services such openvswitch may shutdown before something like kube-apiserver which then can't reach openshift-apiserver over pod network preventing a clean shutdown. Here are some example logs: Apr 14 03:37:39.816073 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: openvswitch.service: Succeeded. Apr 14 03:37:39.816804 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: Stopped Open vSwitch ... Apr 14 03:39:49.289223 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: Stopped libcontainer container 6d13682416c7a6f9acbddeb1cac5d23042f1b1700b17cd2e2d7659c947e2979c. Apr 14 03:39:49.300516 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: crio-6d13682416c7a6f9acbddeb1cac5d23042f1b1700b17cd2e2d7659c947e2979c.scope: Consumed 16min 32.974s CPU time Apr 14 03:39:49.300976 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: Reached target Shutdown. Apr 14 03:39:49.307468 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: Reached target Final Step. Apr 14 03:39:49.316853 ci-op-81677qy8-db044-bb8sk-master-2 systemd[1]: Starting Reboot... Version-Release number of selected component (if applicable): How reproducible: Seen in CI Expected results: The system services such as openvswitch should only be shut down after all the container scopes. Additional info:
https://github.com/cri-o/cri-o/pull/4766
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: OpenShift Container Platform 4.8.2 bug fix and security update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2021:2438