monitoring tape write speeds 2

[emotief]

Hi,

I am a recent convert to Netbackup and a little clueless.

My environment is 4.5 datacentre on Sun Solaris8 platform. I have a StorageTek 9710 timberwolf library with 6 x 9840 drives.

How can I monitor what speed my tapes are writing at?

My backup requirements include a number of large databases and large file systems, so of course I need to get the best performance. This morning, I have over 10 jobs queued, only 1 out of 6 drives writing.

The whole setup was configured by a third party who are a little secretive. So any tips on where to start looking, or what configurations I can change/monitor would be greatly appreciated.

 

[birky]

Hi

If only one drive was writing then I would check the global attributes on your master server to see how many 'jobs per client' is specified...we had the same problem the other day and it was set to value of 1 instead of 99.

Good luck

 

[Mortamer]

In addition to the jobs per client, you will also want to check several other options.

-Check the backup policy to see if multi-streaming is enabled.
-Check the storage unit properties, make sure it is enabled to use multiple drives.
-Check the multiplex setting also in the storage unit properties. You can write a max of 32 jobs to one tape at the same time.

Hope this helps.



Morty

 

[drkung]

just use the iostat -ctx 1 to monitor the disks and tapes' speed

 

[comtec17]

I have a script that will email you every morning with the tape throughput and other stats included. If you would like it, post me your email and I will send.

 

[dbase77]

HI,

Comtec17, do you mind email me the script. my email is dbase77@yahoo.com

Thank you.

regards,
feroz

 

[comtec17]

How did the script work for you guys?

 

[CallwaveEmp]

Comtec17 - can you please email the script to davida@telephia.com?

Am on NT, not unix, but will adapt if I can.
Thanks.

 

[wjflores]

comtec17 can you post the script or send me a copy to try. Thanks.

wjflores94@yahoo.com

 

[johnny99]

I would also like to have a copy of the script.

Email: joe@joe.net

 

[comtec17]

OK all,
Here's the script I use for the e-mail output. This is a Kornshell script for Unix. You can use it on NT/Win2k if you download UWin from

http://www.research.att.com/sw/tools/uwin/

This is KornShell for Windows. Please post your thoughts after you use it.

You need to create the following directories first:

/opt/openv/netbackup/logs/bpbkar
/opt/openv/netbackup/logs/bptm
/opt/openv/netbackup/logs/oldlogs/bpbkar
/opt/openv/netbackup/logs/oldlogs/bptm

#!/usr/bin/ksh
#
# file: veritas.email.sh
#
# Description:
# This script takes portions of two different veritas NetBackup reports
# and forms a comprehensive output to E-Mail after the Backup finishes.
# Then it deletes reports older than 3 days as they can get kinda big.
#
# History:
# 02/26/01 JRO - Created
#

TODAY=`date +%m%d%y`
MAILFILE=/usr/local/bin/veritas.out
WORKDIR="/opt/openv/netbackup/logs"
OLDLOGDIR="${WORKDIR}/oldlogs"
WORKFILE="log.${TODAY}"

date > ${MAILFILE}
echo "" >> ${MAILFILE}
cd ${WORKDIR}/bpbkar

# If the backup creates more than one log
# this will put all the logs into one

onefile() {

>log.temp

for LOGFILES in `ls|grep -v log.temp`
do
cat $LOGFILES >> log.temp
rm $LOGFILES
done

mv log.temp $WORKFILE

}

cd ${WORKDIR}/bptm
onefile

cd ${WORKDIR}/bpbkar
onefile

cd ${WORKDIR}/bpbkar

for BUPID in `grep -w BACKUPID $WORKFILE | awk '{ print $8 }' | cut -c10- | sort`
do
JOBPID=`grep -w $BUPID $WORKFILE | grep -w setenv | awk '{ print $2 }' | cut -c2-`
grep -w $JOBPID $WORKFILE | grep -w BACKUPID | cut -d" " -f1,8- > test.${TODAY}.${JOBPID}
grep -w $JOBPID $WORKFILE | grep -w Processing | cut -d" " -f1,7- >> test.${TODAY}.${JOBPID}
grep -w $BUPID ../bptm/$WORKFILE | grep -w write_backup | grep -w wrote.backup.id.${BUPID} | cut -d" " -f1,5- >> test.${TODAY}.${JOBPID}
grep -w $JOBPID $WORKFILE | grep -w EXIT.STATUS | cut -d" " -f1,7- >> test.${TODAY}.${JOBPID}
sort test.${TODAY}.${JOBPID} >> $MAILFILE
rm test.${TODAY}.${JOBPID}
echo "" >> $MAILFILE
done

echo "" >> $MAILFILE
echo "Statistics:" >> $MAILFILE
grep -w write_backup ${WORKDIR}/bptm/${WORKFILE} | grep -w multiplexed | cut -d" " -f1,5- >> ${MAILFILE}

echo "" >> $MAILFILE
echo "Media Used:" >> $MAILFILE
grep -w select_media ${WORKDIR}/bptm/${WORKFILE} | grep -w selected.media.id | cut -d" " -f1,5- >> $MAILFILE
echo "" >> $MAILFILE

echo "" >> $MAILFILE
/usr/openv/volmgr/bin/tpclean -L >> $MAILFILE
echo "" >> $MAILFILE

# E-mail the report
cat $MAILFILE | /usr/ucb/mail -s "`uname -n`: Veritas Backup - `date|cut -c5-`" root

# Clean up log directories
mv ${WORKDIR}/bpbkar/${WORKFILE} ${OLDLOGDIR}/bpbkar
mv ${WORKDIR}/bptm/${WORKFILE} ${OLDLOGDIR}/bptm
find ${OLDLOGDIR}/bp* -mtime +3 -type f -exec rm -f {} \;