Howdy.
We're using Networker to backup the server [an 8400 5/300] and, at this
time, 11 other clients (DU, Ultrix and SGI) in two groups. Our tape
drive is a TZ877 7 cartridge tape jukebox.
Question 1: I started the 2nd group just yesterday, trying it out with
just one client to see if I set it up correctly. Well, it works.. with
one major problem. One particular machine [an Ultrix DEC 5000/200] is
set up to backup /, /usr and /var. The backups of / and /usr are done
with no problem, but the backup of /var consistently dies. It has never
done the full dump, so it keeps trying to do it. Eventually I will get
a message saying:
[host]:/var has been idle for 30 minutes as of [date]/
[host]:/var is being abandoned by asavegroup
It then says the backup of [host] was unsuccesfully completed.
Group control will call the backup "finished" and will not restart it.
But I cannot dismount the current tape, as nwadmin believes the tape is
still being written to. The only way to get at the tape (that I've
found) is to use /sbin/init.d/nsrd to stop and then restart the NSR
daemons.
Might anyone know what is causing this? What I can do or try to correct
it? I'd really like to make sure /var is backed up, because of the mail
spool.
Question 2: Does anyone know if DEC distributes an NSR Solaris client?
Question 3: When we change tapes in the jukebox, we've found we must
run the inventory program to get it to recognize the new tapes. Is
there away to avoid this?
Question 4: I've been thinking about disaster recovery with NSR should
the -server- get trashed. What are other folks doing about ensuring
that they can restore their server and backup service with NSR should
their server fail?
We had an experience a few months ago where we lost our original tape
indexes [they were overwritten] and we were unable to use the tapes
because NSR would not recognize the labels without a db to match them
to! I am considering doing an off-system non-NSR [probably tar :)]
backup of the /var/nsr directory. Has anyone done anything like this
and had success restoring things after a disaster?
Many thanks for your time and answers.
----------------------------------------------------------------
Esther Filderman moose+_at_cmu.edu
System Mangler Pittsburgh SuperDuperComputing Center
Received on Sat Mar 16 1996 - 22:07:03 NZDT