Oh this is a good one. We were having problems with hung  Netbackup
processes. We could only kill them with a reboot also.
We are running 3.2, Brocade, 9840, and Solaris 7 & 8. The problem was only
on the Sun side. The problem was that avrd looks for FCAW SG devices
in /devices/sbusXXX/fcawXX/sgXX. This was not a problem all the time, but
when the process would hang, it would take a drive with it. Veritas said we
need to run the sg.build and the sg.install scripts (page 45 media manager
config guide). After this we still did not have the SG devices created. We
decided to upgrade the JNI HBA driver to the latest level and after another
reboot, the SG devices were there. Do more problems and no more hung
processes.

Mike





chad.haywood AT storability DOT [email protected] on 02/20/2002 
03:12:10
PM

Sent by:  veritas-bu-admin AT mailman.eng.auburn DOT edu


To:   veritas-bu AT mailman.eng.auburn DOT edu
cc:
Subject:  [Veritas-bu] hung tldd processes on Solaris NBU servers


Hi,

We manage several NBU 3.4.1 servers running on Solaris 7 & 8 E3500s using
STK L700 libraries containing a dozen fibre channel 9840A tape drives (
directly attached to Brocade fabrics, no Crossroads type FC<->SCSI
bridges).
Several times a week we wind up with down drives due to hung tldd
processes.  We can't get rid of these processes even with a kill -9, only
rebooting seems to clean them up.
Veritas is telling us it's a Solaris system call problem.  Is anyone else
experiencing anything similar on Solaris (or any other UNIX variant)?

Thanks,
Chad
_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu