Tim, Steve, and Enzo –
The following does reflect an internal case. Thank you, Enzo, for addressing it. I am opening this XSO to ensure we have this documented for the community facing side, as well.
This is a request to add (within the toolstack layers) the checking of:
- Existence of stale mount points
- What SR-type they tie to
- Making sure XenCenter reflects any stale or disconnected NFS points
The environment where this stems from is 6.5 SP1, just like my own test environments, but the issue is:
- NFS mount points go stale
- These mount points are usually ISO repositories
- XenCenter never informs/shows the status of the NFS mount points as "in a questionable or unusable state"
Example data in logs to support this are:
(general grep from kern.log):
nfs: server abcd.efg.com has timed out
or, more specifically:
Jan 10 12:47:16 localhost xapi: [error|vm-prod-033|11655184 UNIX /var/xapi/xapi|SR.scan D:ffffffffff|storage_access] SR.scan failed SR:OpaqueRef:ffffffff-1111-fffff-1111-f1f1f1f1f1f1f1f1 code=SR_BACKEND_FAILURE_47 params=[; The SR is not available [opterr=no such directory /var/run/sr-mount/ffffffff-1111-fffff-1111-f1f1f1f1f1f1f1f1]; ]
or, the best:
df: `/var/run/sr-mount/bb6fb5ab-6461-4d12-5d94-6457a6d859b6': Stale NFS file handle
So, thanks for reviewing this again as from my tests and in working with two clients, this would greatly help. I like to call it "Operation NARC on All SR Behavior".