Backup and restore on solaris 7 and 8 systems

[shawkins]

Hello all,

I have been given the task of implimenting a backup and restore solution in our development environment. The environment consists of 9 solaris servers and only 3 of the systems have tape drives.

I was going to approach the solution with nfs mounts receiving ufsdumps from the various servers and then backup the ufsdump images to the local drives, however the network is only 10mb and the servers are proxy, dns and mail servers. we clog the network.

My boss wants a solution that can provide a backup and restore facility via the network. I said SAN, he said wake up. No funding for SAN.

I looked at Veritas Netbackup and Legato Networker but I can't find out if I will still have to reload the base os to restore the systems if they crash..
Also I am concerned about the bandwidth requirements to backup over the network.

Is anyone currently performing backups in a similiar env? I have experience with Omniback and Tivoli but using them in thier manufacturer's environments. ie Omniback on 9000's and tivoli on rs6000's.

Any suggestions would be greatly appreciated.

Thanks,

Scott

 

[bi]

Take a look at OmniBack again. Their version 4.x says something about Solaris support -- and not just in the context of being able to back up Solaris systems, which you could do with Omniback 3.x at least. I never looked more closely, but they might have device drivers for Solaris now.

In any event, backing up all those systems over a 10 MB network is going to take a long time.

 

[newmangj]

Also look at the O'Reilly book, "UNIX Backup and Recovery" and the hostdump script provided by the author of that book.

 

[jad]

try looking at ufsdump and ufsrestore.

we use those commands to do incremental backups, running using 'cron' jobs.

if you want i'll paste you a copy of our basic backup script.

 

[MaryYeager]

Jad, please post your script. Thanks!

 

[jad]

this is a script run once a day at approx 11pm at night at our place, runs monday to friday, tape ejects on thrusday evening (after a level 1 backup) so that we don't have a lot of backups in a machine over the weekend.

we have this file in a directory called '/backup' as /backup/dumpdaily

#! /sbin/sh
#
# Backup Procedure "@(#)dumpdaily	1.3 07/08/02 (c) Scientific Analysis Instruments (SAI)"
#
if [ "x$1" != "x" ] && [ "$1" = "0" -o $1 -gt 1 -a $1 -lt 10 ] ; then Level=$1 ; else
DAY=`date +%a`;

case  $DAY in 
  "Mon") Level=3 ;;
  "Tue") Level=4 ;;
  "Wed") Level=5 ;;
  "Thu") Level=1 ;;
  "Fri") Level=2 ;;
esac

fi

# Tape Device
DEVICE=/dev/rmt/0n
#

# Monday    Level3
# Tuesday   Level4
# Wednesday Level5
# Thursday  Level1    Change tape today 
# Friday    Level2
#
#
#
#	/
ufsdump ${Level}fu $DEVICE /dev/dsk/c0t0d0s0
#
#	/usr/openwin
ufsdump ${Level}fu $DEVICE /dev/dsk/c0t0d0s1
#
#	/var
ufsdump ${Level}fu $DEVICE /dev/dsk/c0t0d0s3
#
#	/opt
ufsdump ${Level}fu $DEVICE /dev/dsk/c0t0d0s5
#
#	/usr
ufsdump ${Level}fu $DEVICE /dev/dsk/c0t0d0s6
#
#	/export/home
ufsdump ${Level}fu $DEVICE /dev/dsk/c0t0d0s7

# eject only after Level 1 dump. (oops)
#

if [ "$Level" = "1" -o "$Level" = "0" ] ; then 

#	eject tape
mt -f $DEVICE offline 

fi

we have rolling backup tapes. we have 10 weekly tapes, and every 10 weeks we do a level 0 backup 'reboot -- -s' wait for reboot, log in as root and run '/backup/dumpdaily 0 ; exit' and keep that tape indefinately, we then start at week 1's tape again.

the entry in our crontab is:

00 23 * * 1-5 /backup/dumpdaily

you can add all your slices by looking at 'df -F ufs -k' or looking at your /etc/vfstab

hope it helps.

Jon