-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
4.12.z, 4.14.z
-
None
-
Important
-
No
-
False
-
-
Description of problem:
Openshift cluster upgrade from 4.12.10 to 4.12.30 failing because pod version-4.12.30-xxx is in CreateContainerConfigError. Also tested in 4.14
Steps to Reproduce:
- Deploy new 4.12.10 cluster
$ oc get clusterversion NAME VERSION AVAILABLE PROGRESSING SINCE STATUS version 4.12.10 True False 12m Cluster version is 4.12.10
- Create the following SCC
--- allowHostDirVolumePlugin: true allowHostIPC: false allowHostNetwork: false allowHostPID: false allowHostPorts: false allowPrivilegeEscalation: true allowPrivilegedContainer: true allowedCapabilities: null apiVersion: security.openshift.io/v1 defaultAddCapabilities: null fsGroup: type: RunAsAny groups: [] kind: SecurityContextConstraints metadata: name: scc-hostpath-cnf-cat-1 priority: null readOnlyRootFilesystem: false requiredDropCapabilities: - KILL - MKNOD - SETUID - SETGID runAsUser: type: MustRunAsNonRoot seLinuxContext: type: MustRunAs supplementalGroups: type: RunAsAny users: [] volumes: - configMap - downwardAPI - emptyDir - hostPath - persistentVolumeClaim - projected - secret
- Upgrade to 4.12.30
$ oc adm upgrade --to=4.12.30 $ oc get pod -n openshift-cluster-version NAME READY STATUS RESTARTS AGE cluster-version-operator-85db98885c-jt25z 1/1 Running 0 41m version-4.12.30-vw4pm-l2nng 0/1 Init:CreateContainerConfigError 0 42s $ oc get events | grep Failed 10s Warning Failed pod/version-4.12.30-p6k4r-nmn6m Error: container has runAsNonRoot and image will run as root (pod: "version-4.12.30-p6k4r-nmn6m_openshift-cluster-version(4d1704d9-ca34-4aa3-86e1-1742e8cead0c)", container: cleanup) $ oc get pod version-4.12.30-97nbr-88mxp -o yaml |grep scc openshift.io/scc: scc-hostpath-cnf-cat-1
As a workaround, we can remove the scc "scc-hostpath-cnf-cat-1" and the pod version-xxx and the upgrade worked. Customer has created custom scc for use of applications.
$ oc get pod version-4.12.30-nmskz-d5x2c -o yaml | grep scc
openshift.io/scc: node-exporter
$ oc get pod
NAME READY STATUS RESTARTS AGE
cluster-version-operator-6cb5557f8f-v65vb 1/1 Running 0 54s
version-4.12.30-nmskz-d5x2c 0/1 Completed 0 67s
There's an old bug https://issues.redhat.com/browse/OCPBUGSM-47192 which was fixed setting readOnlyRootFilesystem to false, but in this case the scc is still failing.
---
container.SecurityContext = &corev1.SecurityContext{
Privileged: pointer.BoolPtr(true),
ReadOnlyRootFilesystem: pointer.BoolPtr(false),
}
---
- is duplicated by
-
-
- Closed
-
- links to