A CEPH RWX PV is shared between deployment and statefull set (same namespace). A
Both applications share the volume, and the volume has correct permissions.

The Statefull set application is running with "runAsUser=1000", where the other was running with default openshift uid set from namespace range.

Until couple of weeks back the solution worked just fine.
When the issue started, customer saw behavior, when one of the application's pods were deleted/recreated - then the other application lost access to the volume with error "Permission Denied".

We checked the linux permissions, and the directory/mount is set to "drwxrwsr-x" and owned by "root:1000".
All files into the volume are "rwx" to user, group and other. 

This behavior is visible on DEV and PROD environment.
Currently, not easy manual workaround is in place.

OpenShift 4.11.39

- currently working on reproducer

1.
2.
3.