-
Bug
-
Resolution: Can't Do
-
Major
-
None
-
4.13.z
-
None
-
None
-
False
-
Description of problem:
Customers are encountering the following bug in 4.13.z installs (so far), this caused install failures for a supported version. The the fix in 4.14 needs to be backported to 4.13 to avoid users from encountering this bug. https://issues.redhat.com/browse/OCPBUGS-2474
Version-Release number of selected component (if applicable):
How reproducible:
Intermittent, unsure about the frequency. Please have a look at the linked/closed ocpbugs ticket for more information as QA was done in there.
Steps to Reproduce:
1. Install a cluster with version < 4.14 2. 3.
Actual results:
Almost the same with what is describe in the following KCS that is also pointing to the linked/closed ocpbugs ticket/bugzilla. https://access.redhat.com/solutions/6976343 . Static pods failed to start namely kube-apiserver and guard pods. kube-apiserver static pods failed to start and gets stuck.
Expected results:
kube-apiserver should have a running state
Additional info:
During install, ARO checks for cluster operators health and especially kube-apiserver. The following log we get when this install failure happens in many clusters. Please see the KCS/bugzilla, the errors are almost exactly the same; "reason": "GuardController_SyncError::MissingStaticPodController_SyncError::StaticPods_Error", "message": "GuardControllerDegraded: Missing operand on node <redacted>\nMissingStaticPodControllerDegraded: static pod lifecycle failure - static pod: \"kube-apiserver\" in namespace: \"openshift-kube-apiserver\" for revision: 6 on node: \"<redacted>\" didn't show up, waited: 4m45s\nStaticPodsDegraded: pod/kube-apiserver<redacted> \"kube-apiserver\" is terminated: Completed: \nStaticPodsDegraded: <redacted> \"kube-apiserver-cert-regeneration-controller\" is terminated: Completed: