Details
-
Epic
-
Resolution: Unresolved
-
Critical
-
None
-
openshift-4.11, openshift-4.12, openshift-4.13, openshift-4.14
-
Resilient Cert Refresh
-
False
-
None
-
False
-
Not Selected
-
To Do
-
OCPSTRAT-543 - Shutdown/Resume of managed OSD/ROSA clusters
-
-
0
-
0%
-
0
-
0
Description
Hive is working on a feature which will enable OCP cluster to hibernate and resume the cluster on demand
Without proper handling, it is possible that hibernating and resuming a cluster can lead to undesired behavior in MCO like expired certs . To make sure that cluster can be safely hibernated and resumed in automated fashion, we need to know:
- Which certs expiry can cause issue and how frequently they get renewed?
- Will MCO be able to successfully resume update of any expired certs when cluster resumes from hibernation
- What happens when certs like kubelet ca expires during hibernation. Does kube apiserver automatically generates new cert bundle when cluster resumes.
- How to tell that MCP update is in progress
- In 4.13, for paused pool cert update will happen automatically. Can we detect if a cert update is in progress for paused pool?
- With 4.14, workflow for cert update will get decoupled from renderd MC. In this case how we will detect that applying cert is in progress?
- How long cluster can be safely hibernated?
Related Hive user story https://issues.redhat.com/browse/HIVE-2224
