SUMMARY: Databases & Solstice Backup (Networker)

From: Dan Penrod (penrod@wcnewmedia.com)
Date: Thu Jul 11 1996 - 11:04:56 CDT


Hey managers:

Thanks for all the replies.
In brief my original query said that we've installed
Sun's Solstice Backup (Networker) and want to be able
to run backups that are reliable for our Oracle database.
How do we get Networker to reliably backup the database?

The answer was pretty much "You can't get there from here."
Replies fell into about 3 categories;
1. "We use Networker and we shut down our database
   all night, every night."
2. "We do on-line "hot" backups but we don't use Networker."
3. "There's a Networker "add-on" for Oracle but we've
   never used it because it's way too expensive and
   besides I don't think Sun supports it on their
   Solstice Backup version of Networker."

As I mentioned before, we cannot take down the database
everytime we want to run backups. We're under contract
to be on-line, up-and-running, ready-to-access, 24 hrs,
round-the-clock... you get the picture.

There is a good reference that I used to nail down our
final solution. "ORACLE DBA Handbook" by Kevin Loney and
Oracle Press (Tm), which appears to be the last word on
the subject. Chapter 10 tells you that you can back up
the transaction logs on-line, you can export the database
on-line (both which constitute restorable backups), and
you can shut down the database to backup (Cold backup).

Our solution was to write a good'ol fashion bourne shell
script which implements a good'ol fashion unix dump (backup),
and implement Oracles "Hot (Archivelog) Backup" in combo
with an "EXPORT" backup. With this solution I get a good
system backup and my DBA gets a good database backup, which
all runs in one backup process at night. I've included my
script at the bottom (below sig) for your perusal, It's
probably a better explanation than I can verbiate anyways.
In my script, step 2 is where the archive log occurs and
step 5 is where the export occurs (although only once a week).

Many, many, many thanks to the following managers. Without
their feedback I'd still be digging through the crappy .ps
file docs that come with Solstice Backup looking for a
solution...

bbice@persistence.com (Brent Bice)
  "We shut-down, btw there's an add-on"
John Bradley <john.bradley@sunman.chinalake.navy.mil>
  "We shut-down at night via cron job"
"Roger B.A. Klorese" <rogerk@veritas.com>
  "Legato's the co, Network the product, Solstice Sun's name"
"Plesha, Tom A." <tplesha@seacosd.navy.mil>
  "We shut-down. Our DB is too big to do transaction log backup"
gibian@stars1.hanscom.af.mil (Marc S. Gibian)
  "There's an add-on but Sun support is bad, also, there's
   a Networker mailing list. Let me know if you want the info."
aub@ide.com
  "Legato's not very good. There is an add-on
dvande01@hac.vlt.eds.com (Doug Vanderlip)
  "You can export DB. We shut-down."
Mike Rembis <ebumfr@ebu.ericsson.se>
  "I think you need add-on from Legato AND Oracle."
Stuart.Little@ssu.stcssl.co.uk
  "I guess you write an ext. ASM. Let me know what you find."
mal@getech.leeds.ac.uk
  "Check out July's issue of Sys Admin on hot backups."
James M Mosley <jmmosley@uncc.edu>
  "We used to have cron copy tablespaces off before Networker
   starts up. There's an add-on which is very expensive.
Garry Robbins <Garry.Robbins@labatt.com>
  "We do hot backups by copying out tablespaces. We don't use
   Networker."
msiple@mfg.wb.xerox.com (Matt Siple)
  "We don't use Networker. We do shut-down the database."
Stevan P. Meck" <DOCDOMAIN.MISOC-PO.MECKSP@state.mi.us
  "Our Sybase software dumps the database and logs to disk
   and we schedule backups of that disk. BTW, Solstice docs
   are horrible."
charles@Grafnetix.COM (Charles Gagnon)
  "We use cron to shut-down DB before Networker backs it up."
Darren King <dking@LEGATO.com>
  "You need Database Module for Oracle (DMO). Ask Sun if you
   can add DMO to Solstice Backup. It's not exactly the same
   as Networker."
iv08480@issc02.mdc.com (Colin Melville)
  "We shut-down, export, bring it back up, do backups as scheduled."

Regards,
dan

-- 
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|   _/ _/_/  _/_/  _/_/  _/_/_/    _/_/_/_/_/ _/_/_/_/|
|     _/    _/  _/  _/ _/      _/ _/         _/       |
|    _/    _/      _/ _/      _/ _/    _/_/ _/_/_/    |
|   _/    _/      _/ _/_/_/_/_/ _/      _/ _/         |
|_/_/_/  _/      _/ _/      _/ _/_/_/_/_/ _/_/_/_/_/  |
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
| Dan Penrod - Unix Network Administrator             |
| Image Technologies - World Color New Media          |
| 12425 28th St. N., St. Petersburg, FL 33716         |
| vox:813/573-5268 fax:573-0431 penrod@wcnewmedia.com |
| penrod@wcnewmedia.com                               |
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

#!/usr/bin/sh # backup # quick'n'dirty script to backup scratchy partitions # 6/14/96 Dan & Sam # # Debug Switch #set -x LEVEL=$1 HOST=`hostname` DEVICE="/dev/rmt/0n" DUMP="/usr/sbin/ufsdump" MT="/bin/mt" LOGFILE=$LEVEL-`date '+%y%m%d%H%M%S'` SWITCHES="bdsf 126 54000 36000" OPTIONS="$LEVEL$SWITCHES" # if test $LEVEL = "" then echo "Missing dump level" echo "USAGE: mybackup <level>" exit 1 fi # echo "Running backup on $HOST at level $LEVEL" echo "Spawning Dump Window..." xterm +t -sb -T "Backup Status" -e tail -f $LOGFILE && # # STEP 1: Backup root partition and all root subdirectories echo "Running dump on root partition of $HOST" $DUMP $OPTIONS $DEVICE / 2> $LOGFILE echo "Dump of $HOST root partition finished." # # STEP 2: Move Data Files Off-Line echo "Backing up oracle datafiles." echo "Setting tablespace offline." su oracle -c svrmgrl << EOF1 connect internal alter tablespace ltdata begin backup; alter tablespace ltindex begin backup; alter tablespace lttemp begin backup; alter tablespace orcl begin backup; alter tablespace rbs begin backup; alter tablespace system begin backup; alter tablespace temp begin backup; alter tablespace tools begin backup; alter tablespace users begin backup; !cp /lt01/oradata/LTDB/*.dbf /lt01/oradata/LTDB/ARCHIVE !rm /lt01/oradata/LTDB/control.bak !echo "Setting tablespace back online." alter tablespace ltdata end backup; alter tablespace ltindex end backup; alter tablespace lttemp end backup; alter tablespace orcl end backup; alter tablespace rbs end backup; alter tablespace system end backup; alter tablespace temp end backup; alter tablespace tools end backup; alter tablespace users end backup; alter system switch logfile; alter database backup controlfile to '/lt01/oradata/LTDB/control.bak'; exit EOF1 echo "Copying datafiles complete!" # # STEP 3: Backup the Lucent partition echo "Running dump on Lucent (lt01) partition of $HOST" $DUMP $OPTIONS $DEVICE /lt01 2>> $LOGFILE echo "Dump of $HOST Lucent partition finished." # # STEP 4: Rewind and eject echo "Rewinding and ejecting tape." $MT -f $DEVICE rewoffl # # STEP 5: Clean Archive Log Files today=`date | awk ' { print $1 } '` if [ "$today" = Fri ]; then echo "Don't forget to label the tape [$LOGFILE $HOST]" echo "Today is Friday. Cleaning Archive Log Files..." su oracle -c exp userid=www_dba/www_dba file=/lt01/oradata/LTDB/ARCHIVE/$LOGFILE.dmp owner=www_dba rm -f /lt01/app/oracle/admin/LTDB/arch/*.ARC else echo "Today is $today. We can ignore Archive Logs today." fi # Clean up ARCHIVE directory now that backup is done rm -f /lt01/oradata/LTDB/ARCHIVE/*.dbf # date echo "DONE!!!"



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:04 CDT