I’m having a problem with clients dropping their xsan volumes, usually on a restart of the client. The fix I used is below, but I’d live to have abetter method. Here’s what I know:
The MDCs are running 10.7.4, as are the clients.
There is a ghost directory that I have deleted, but it returns from the dead, still x’ed out.
There’s been no change to fiber cabling.
The wwnn’s on the client match those on the qlogic switch for the client’s zone
All LUN’s are visible on the client via Disk Utility
Mount san volume fails from xsan admin and from Xsan in user prefs.
I’ve restarted the client
I’ve removed the client from the san and then re-added it
I’ve disabled the san pref and re-enabled it.
Both Xsan Admin and cvadmin act strangely. (Note that I try not to run cvadmin and Xsan admin at the same time, and I don’t intentionally leave them running on both MDCs at the same time. This has always been a recipe for squirrely behavior.) I believe there is a bug related to a disconnect between cvadmin and the Xsan Admin GUI.
In the Mount list, Xsan Admin shows all clients as “not mounted”
Here’s a terminal session showing cvadmin on the primary MDC:
>
ohaephqxs002:~ aepadmin$ sudo cvadmin
Password:
Xsan Administrator
Enter command(s)
For command help, enter “help” or “?”.
List FSS
File System Services (* indicates service is in control of FS):
1> Video[0] located on ohaephqxs001.aepsc.meta:49257 (pid 232)
2>*Video[1] located on ohaephqxs002.aepsc.meta:55017 (pid 35048)
Select FSM “Video”
Admin Tap Connection to FSM failed: [errno 54]: Connection reset by peer
FSM may have too many connections active.
Cannot select FSS “Video”
>
What fixes it is to force a failover of the MDCs. This immediately mounts the problem client and restores cvadmin to normal behavior; also Xsan Admin correctly shows clients as mounted.