Page MenuHomePhabricator

Helm install fails in CI namespace: apparmor failed to apply profile
Closed, ResolvedPublic

Description

Attempting to install the blubberoid helm chart in the ci namespace is failing.

kubectl describe pod reveals this error message:

Failed create pod sandbox: rpc error: code = Unknown desc = failed to start sandbox container for pod "blubberoid-4ipjhmel-66d65bd7cd-pbr9x": Error response from daemon: invalid header field value "oci runtime error: container_linux.go:247: starting container process caused \"process_linux.go:359: container init caused \\\"apparmor failed to apply profile: no such file or directory\\\"\"\n"

Event Timeline

jeena renamed this task from Helm install fails in CI namespace to Helm install fails in CI namespace: apparmor failed to apply profile.Feb 2 2021, 12:27 AM
JMeybohm edited projects, added serviceops, Prod-Kubernetes; removed SRE.

Most likely my fault as I installed the apparmor package as part of T228967.
Will take a look!

Joe triaged this task as High priority.Feb 2 2021, 9:10 AM

Change 661083 had a related patch set uploaded (by JMeybohm; owner: JMeybohm):
[operations/puppet@production] k8s::kubelet: Ensure apparmor is purged on old k8s nodes

https://gerrit.wikimedia.org/r/661083

Fortunately, this issue had not effected prod clusters. So it must be something in the combination of kernel / k8s version / apparmor that goes wrong here.
As we're moving away from that combination I think it's not worth it to spend more time investigating. We'll just purge the apparmor package from old k8s nodes again.

Change 661083 merged by JMeybohm:
[operations/puppet@production] k8s::kubelet: Ensure apparmor is purged on old k8s nodes

https://gerrit.wikimedia.org/r/661083

Did a puppet run on the affected nodes and scaled blubberoid pods on each of them. Looks good to me.