I have a weird issue happening on my master.  I have written a script that
queries a master or masters for all clients that have not been successfully
backed up since the beginning of our window, then checks to see if any of the
ones that don't have a backup are still running, and then, if no backup and not
running, it reruns the jobs.  All that logic works fine.  And, it's even kicking
off the jobs fine, but for some reason if I try to kick of a larger number (40
or 50+), all the currently running jobs fail with a status 50, because one of
either bpcd or bprd crashes and drops core.

I am using this command:

bpbackup -c $CLASS -i -s $SCHEDULE -S $MASTER -h $CLIENT -t $CLASSTYPE

All of the parameters are provided from the script interpreting logs, and I've
verified that these are correct.

So, it's looking like I'm overloading the server in some way, but I have never
seen such a thing happen.  I tried putting a 10 second delay between starting
each job, but that didn't help either.

Why would kicking off a large number manual jobs in this fashion (one at a time)
break a process?  It does not seem to have this problem if they are launched
from the X interface.

Oh, this is Netbackup 3.4 at patchlevel 645 (yes, I know I'm behind in patches,
but everything has to be certified before it gets on the floor, and that's an
undertaking).  

Any ideas or suggestions would be most appreciated.

-- 
        Scott Garrett     mailto:sgarrett AT technomancer DOT com
                                          http://www.technomancer.com

    *** No matter what you do, you'll never outsmart the squirrels.