UPSTREAM: 69215: Fix flake in CSI plugin e2e test

[bertinatto]

This PR fixes a flake in the CSI plugin test.

During the test, the container hostpath-driver exposes the socket /csi/csi.sock so that other containers can connect to and perform CSI operations.

Since hostpath-driver container is privileged, other containers can't access the socket because they don't have the proper permission for it. As a result, the other containers need to be privileged as well.

The flake only happens in OKD, so I haven't submitted a PR to upstream.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1622670

[bertinatto]

/sig storage

[bertinatto]

/assign @wongma7

[bertinatto]

CC @tsmetana

[wongma7]

/lgtm
I think it is harmless to carry this patch for now but I wonder too why it is needed here but not upstream. What is the openshift equivalent of /var/lib/kubelet? Maybe it has something to do with the setgid bit on /var/lib/kubelet?

[bertinatto]

I think the problem lies in the fact that Openshift maps hostPath volumes to external XFS partitions instead of mapping it to the host's filesystem. This is the output of the command mount executed inside the container:

/ # mount | grep csi
/dev/mapper/rhel-root on /csi type xfs (rw,seclabel,relatime,attr2,inode64,noquota)

I believe in this case the files stored in the mount point are subject to different security policies than they would be if they were stored in the host (i.e., as it happens in Kubernetes).

[1] https://github.com/bertinatto/kubernetes/blob/516876b232c8fa45689f0754a5865aad1aaa2f7b/test/e2e/storage/csi_objects.go#L330

[bertinatto]

/assign @gnufied

[gnufied]

Just to confirm - we are sure that external-provisioner pod isn't already privileged right? I think what may be happening is - on upstream in absence of selinux the other pods are able to access that socket but in case of selinux since we don't peform relabeling of hostpaths, other pods can't access socket unless privileged. It seems like an oversight in my opinion to have driver pod privileged and other pods not. Lets submit this to upstream and see what people think?

[bertinatto]

Just to confirm - we are sure that external-provisioner pod isn't already privileged right?

Yes, I inspected the container during the test and it wasn't privileged.

Also, here is a test run with commit 5e39b only; it fails with the message described in the Bugzilla ticket above:

https://openshift-gce-devel.appspot.com/build/origin-ci-test/pr-logs/pull/21121/pull-ci-openshift-origin-master-e2e-gcp/107#sig-storage-csi-volumes-csi-plugin-test-using-csi-driver-hostpath-should-provision-storage-suiteopenshiftconformanceparallel-suitek8s

Lets submit this to upstream and see what people think?

Sure, I can try submitting the patch upstream.

[bertinatto]

Submitted upstream: kubernetes/kubernetes#69215

[bertinatto]

The upstream PR [1] was merged, so I updated this one to carry that patch instead.

[1] kubernetes/kubernetes#69215

[bertinatto]

/test cmd

[gnufied]

/lgtm

[gnufied]

/assign @knobunc @smarterclayton for final merge.

@bertinatto thanks for fixing this!

[knobunc]

/lgtm
/approve

[openshift-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: bertinatto, gnufied, knobunc, wongma7

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• test/extended/OWNERS [knobunc]
• vendor/k8s.io/kubernetes/test/e2e/storage/OWNERS [gnufied]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[knobunc]

/assign @smarterclayton

[openshift-bot]

/retest

Please review the full test history for this PR and help us cut down flakes.

[openshift-bot]

/retest

Please review the full test history for this PR and help us cut down flakes.

[openshift-bot]

/retest

Please review the full test history for this PR and help us cut down flakes.

[openshift-bot]

/retest

Please review the full test history for this PR and help us cut down flakes.

[openshift-ci-robot]

@bertinatto: The following test failed, say /retest to rerun them all:

Test name	Commit	Details	Rerun command
ci/openshift-jenkins/extended_builds	e9302aa	link	/test extended_builds

Full PR test history. Your PR dashboard. Please help us cut down on flakes by linking to an open issue when you hit one in your PR.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here.