Okay:
Here's the scoop
My client has fibre attached (real fibre attached) STK9840's and STK9940's
hanging off of a Sun Server running Solaris 8. The client's assumed if a drive
needed to be replaced it was a simple "hot-swap", however since AVRD looks at
the /dev/rmt's every 15 seconds (or whatever the time value), it hangs on the
drive that was unplugged and basically "freezes" after serveral days.
Eventually, Netbackup begins to show signs of a slow death, Tapes are not
DisMounted properly or at all and AVRD just HANGS and doesn't even respond to a
kill with extreme prejudice (-9). Everything else dies normally when we ask
for it to terminate.
So, here's the question.
Is this a problem with AVRD that will be fixed in future releases?
Is there a way to let AVRD "give-up" on a drive and leave it completely alone
and not affect all other drives in the configuration?
Sun has said, based on all of the logs they reviewed, etc. that this is treated
as a "SCSI Failure", so if it is a SCSI Failure, shouldn't AVRD behave a little
better?
If not, that's cool and I'll just let my client know that they have to follow
the right procedures...like waiting for their shop to complete all requested
jobs before they can change drives.
Thanks in advance!
David
|