Description of problem:
Apr 16 02:47:54.240 W ns/e2e-volumemode-7208 pod/csi-hostpathplugin-0 node/ci-op-8lmbm-w-b-g294t.c.openshift-gce-devel-ci.internal reason/FailedPreStopHook Exec lifecycle hook ([/bin/sh -c rm -rf /registration/csi-hostpath /registration/csi-hostpath-reg.sock]) for Container "node-driver-registrar" in Pod "csi-hostpathplugin-0_e2e-volumemode-7208(c3a0be58-f762-4513-ad18-e4e8d183d547)" failed - error: rpc error: code = Unknown desc = command error: time="2020-04-16T02:47:54Z" level=error msg="exec failed: container_linux.go:349: starting container process caused \"exec: \\\"/bin/sh\\\": stat /bin/sh: no such file or directory\""\nexec failed: container_linux.go:349: starting container process caused "exec: \"/bin/sh\": stat /bin/sh: no such file or directory"\n, stdout: , stderr: , exit code -1, message: ""
A CSI tests is cleaning up and deletes csi-hostpath driver pod. The driver pod has PreStop hook to delete driver's socket, however, there is no /bin/sh in the node driver registrar image (we're using distroless) and the hook fails.
The test itself succeeds, however, logs are full of hard to understand errors, which complicates debugging of other failures.
Version-Release number of selected component (if applicable):
Note: this is e2e tests we got from upstream, i.e. it uses upstream node registrar image that is distroless.
Our OCP images are based on RHEL and they have a shell.
Upstream PR: https://github.com/kubernetes-csi/csi-driver-host-path/pull/176
"backport" to kubernetes/kubernetes: https://github.com/kubernetes/kubernetes/pull/90234
Checked in this e2e log: https://storage.googleapis.com/origin-ci-test/pr-logs/pull/openshift_cluster-etcd-operator/338/pull-ci-openshift-cluster-etcd-operator-master-e2e-gcp/1312/artifacts/e2e-gcp/e2e.log
Did not find this issue, so marked this bug as verified.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.