Hi folks,

We have a library sharing environment in a STK SL3000. The library is FC
connected, as are the drives:

12x HP LTO-6
8x HP LTO-5

The library manager runs RHEL6.6 and TSM 6.3.4.300. There is a library
client that ran RHEL5 and TSM 6.3.4.300, and had no problems. Our HBAs are
all QLogic (QLE2562 and QLE2564) and we use the latest drivers and SAN
discovery libraries from qlogic.com.

We recently upgraded the client to RHEL6.6 (TSM is still 6.3.4.300) and are
seeing *very* slow dismounts. For instance, before the upgrade, we could
issue DISMOUNT VOLUME on the library client and the volume would immediately
disappear from QUERY MOUNT output on the client, and would go from IDLE to
DISMOUNTING on the library manager. The drive would also immediately enter
the busyRewind state in the library management interface.

Now, issuing a DISMOUNT VOLUME will cause the volume to enter DISMOUNTING
on the library manager, but will remain in an IDLE state on the library
client. The library management interface never shows the drive enter
busyRewind; it remains in Loaded.

Eventually, the library manager wrests control from the library client, but
this can take several hours. I suspect the manager never gets its SCSI 
reservation
back from the client and ends up resetting the drive, but there's nothing
in the activity log indicating any errors. I've tried adjusting
MOUNTRETENTION on our device classes down to 1, but it doesn't seem to have
helped things at all.

Has anyone else seen this behavior or have insight into what's going on?

Thanks!

--
-- Skylar Thompson (skylar2 AT u.washington DOT edu)
-- Genome Sciences Department, System Administrator
-- Foege Building S046, (206)-685-7354
-- University of Washington School of Medicine