Bug 1397968

Description of problem:

As an OpenShift admin, I need to be able to tell the recycler to ignore certain files or folders.

Related issue: https://github.com/openshift/origin/issues/8721

When a pod is deleted and the pv  has to be freed up, I can see that there is a recycler pod that comes in and tries to make the PV available. At many organizations, every mount has a .snapshot folder that is readonly as given to us by the storage team. Since the recycler pod fails to remove it, the recycler pod itself goes to failed state and the PV never goes to available.  The PV then also stays in Failed state. This leads me to remove the recycler pod, remove the PV and recreate the PV.

Version-Release number of selected component (if applicable):

[admin@ga1-1 metrics]$ openshift version
openshift v3.3.1.3
kubernetes v1.3.0+52492b4
etcd 2.3.0+git

How reproducible:

Always

Expected results:

I expect to define a --exclude somewhere so the recycler pod ignores .snapshot folder

Additional info:

[admin@ga1-1 metrics]$ oc get all
NAME                      READY     STATUS    RESTARTS   AGE
recycler-for-infra-pv-1   1/1       Running   0          4s
recycler-for-infra-pv-3   0/1       Error     0          19s
[sasharm4@cae-ga1-1 metrics]$ oc logs recycler-for-infra-pv-3
error: recycle failed: /scrub/.snapshot (drwxrwxrwx), calling walkFn: remove /scrub/.snapshot: read-only file system

[admin@ga1-1 metrics]$ oc get pv
NAME         CAPACITY   ACCESSMODES   STATUS      CLAIM                                 REASON    AGE
infra-pv-1   50Gi       RWO           Failed      openshift-infra/metrics-cassandra-1             1h
infra-pv-2   50Gi       RWO           Available                                                   1h
infra-pv-3   50Gi       RWO           Failed      openshift-infra/metrics-cassandra-1             1h

Submitted on behalf of client.