-
Bug
-
Resolution: Won't Do
-
Major
-
None
-
4.12.z
-
Quality / Stability / Reliability
-
False
-
-
None
-
None
-
No
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem:
kube-scheduler pod on a freshly installed cluster is stuck in Completed and never makes it into running. Issue seems to be similar to OCPBUGS-94 oc get pods -n openshift-kube-scheduler NAME READY STATUS RESTARTS AGE installer-5-ip-10-0-129-102.us-east-2.compute.internal 0/1 Completed 0 155m installer-5-ip-10-0-189-140.us-east-2.compute.internal 0/1 Completed 0 157m installer-5-ip-10-0-203-126.us-east-2.compute.internal 0/1 Completed 0 154m installer-6-ip-10-0-189-140.us-east-2.compute.internal 0/1 Completed 0 151m openshift-kube-scheduler-guard-ip-10-0-129-102.us-east-2.compute.internal 1/1 Running 0 154m openshift-kube-scheduler-guard-ip-10-0-189-140.us-east-2.compute.internal 0/1 Running 0 156m openshift-kube-scheduler-guard-ip-10-0-203-126.us-east-2.compute.internal 1/1 Running 0 153m openshift-kube-scheduler-ip-10-0-129-102.us-east-2.compute.internal 3/3 Running 15 (4m48s ago) 154m openshift-kube-scheduler-ip-10-0-189-140.us-east-2.compute.internal 0/3 Completed 0 156m openshift-kube-scheduler-ip-10-0-203-126.us-east-2.compute.internal 3/3 Running 15 (3m18s ago) 153m journalctl -u kubelet | grep openshift-kube-scheduler-ip-10-0-189-140.us-east-2.compute.internal | head -n 50 Jan 12 08:13:55 ip-10-0-189-140 kubenswrapper[2226]: E0112 08:13:55.614430 2226 pod_workers.go:965] "Error syncing pod, skipping" err="failed to delete cgroup paths for [kubepods burstable podbb541fad0475f80a0d69e7660bc99190] : unable to destroy cgroup paths for cgroup [kubepods burstable podbb541fad0475f80a0d69e7660bc99190] : Timed out while waiting for systemd to remove kubepods-burstable-podbb541fad0475f80a0d69e7660bc99190.slice" pod="openshift-kube-scheduler/openshift-kube-scheduler-ip-10-0-189-140.us-east-2.compute.internal" podUID=bb541fad0475f80a0d69e7660bc99190
Version-Release number of selected component (if applicable):
Seen on 4.12.11 and 4.12.25
How reproducible:
Sporadic
Steps to Reproduce:
1.
2.
3.
Actual results:
kube-scheduler pod is stuck in Completed state.
Expected results:
kube-scheduler pod makes it into running state.
Additional info:
Attaching: must gather, sosreport
