flake: conformance: EmptyDir volumes should support (root,0644,default) hangs

[mmilata]

The test gets stuck on:

• [SLOW TEST:9.323 seconds]
[k8s.io] EmptyDir volumes
/data/src/github.com/openshift/origin/vendor/k8s.io/kubernetes/test/e2e/framework/framework.go:793
  should support (root,0644,default) [Conformance]
  /data/src/github.com/openshift/origin/vendor/k8s.io/kubernetes/test/e2e/common/empty_dir.go:97

Seen here: https://ci.openshift.redhat.com/jenkins/job/test_pull_requests_origin_conformance/8030/console

[gnufied]

According to the test logs, this e2e test is not failing there. It failed on:

[k8s.io] DisruptionController [Feature:PodDisruptionbudget]
/data/src/github.com/openshift/origin/vendor/k8s.io/kubernetes/test/e2e/framework/framework.go:793
  should update PodDisruptionBudget status [It]
  /data/src/github.com/openshift/origin/vendor/k8s.io/kubernetes/test/e2e/disruption.go:102

  Expected error:
      <*errors.errorString | 0xc8200dd780>: {
          s: "timed out waiting for the condition",
      }
      timed out waiting for the condition
  not to have occurred

  /data/src/github.com/openshift/origin/vendor/k8s.io/kubernetes/test/e2e/disruption.go:100

The EmptyDir specs actually passed successfully:

STEP: Creating a pod to test emptydir 0644 on node default medium
Oct 31 08:52:54.558: INFO: Waiting up to 5m0s for pod pod-f0b8ac18-9f68-11e6-bc5e-0edf46e0b6b0 status to be success or failure
Oct 31 08:52:54.560: INFO: No Status.Info for container 'test-container' in pod 'pod-f0b8ac18-9f68-11e6-bc5e-0edf46e0b6b0' yet
Oct 31 08:52:54.560: INFO: Waiting for pod pod-f0b8ac18-9f68-11e6-bc5e-0edf46e0b6b0 in namespace 'e2e-tests-emptydir-b4ij8' status to be 'success or failure'(found phase: "Pending", readiness: false) (1.95877ms elapsed)
Oct 31 08:52:56.562: INFO: Nil State.Terminated for container 'test-container' in pod 'pod-f0b8ac18-9f68-11e6-bc5e-0edf46e0b6b0' in namespace 'e2e-tests-emptydir-b4ij8' so far
Oct 31 08:52:56.562: INFO: Waiting for pod pod-f0b8ac18-9f68-11e6-bc5e-0edf46e0b6b0 in namespace 'e2e-tests-emptydir-b4ij8' status to be 'success or failure'(found phase: "Running", readiness: true) (2.004294609s elapsed)
STEP: Saw pod success
Oct 31 08:52:58.568: INFO: Trying to get logs from node 172.18.0.150 pod pod-f0b8ac18-9f68-11e6-bc5e-0edf46e0b6b0 container test-container: <nil>
[AfterEach] [k8s.io] EmptyDir volumes
  /data/src/github.com/openshift/origin/vendor/k8s.io/kubernetes/test/e2e/framework/framework.go:134
Oct 31 08:52:58.584: INFO: Waiting up to 1m0s for all nodes to be ready
STEP: Destroying namespace "e2e-tests-emptydir-b4ij8" for this suite.
Oct 31 08:53:03.624: INFO: namespace: e2e-tests-emptydir-b4ij8, resource: bindings, ignored listing per whitelist

[gnufied]

Also the podDisruption test is being discussed here - #11114

[mmilata]

@gnufied apologies, I haven't noticed the PodDisruptionBudget failure.

Still, the issue here was that the tests got stuck after the EmptyDir test finished. It's not obvious from the logs, however you can see that the job ran 5+ hours. Could this be caused by the earlier failure?