From: Darryl Ousterhout (D.Ousterhout@LABSAFETY.COM)
Date: Tue Jun 10 2003 - 09:04:38 EDT
Not sure if this is what your looking for, but there is a set of command
line tools that are packaged with perfagent.tools. The command is called
'ptxrlog'. The command uses user define metrics to gather system information
over a duration. To look at the possibilities for the metrics, the 'xmpeek
-l | more' command will give you a complete listing.
The syntax for the ptxrlog command is as follows:
ptxrlog -f <input file with defined metric> -o <output file for gathered
stats> -i <interval in seconds> -d <duration in hours>
An example would be:
ptxrlog -f /ptxrec/memory.cfg -o /ptxrec/reports/memory.out -i 150 -d 23.50
What I used to do with this is run it for 23.5 hours and it would
automatically end it's processes. At that time I'd package them up and run
them through Access to produce management appeasing graphs.
Here is the main part of the script:
INTERVAL=150
DURATION=23.50
INPUT=/ptxrec/cfg
OUTPUT=/ptxrec/output
CFGFILE=$(cat /ptxrec/ptx_cpu.cfg)
SEARCH=$(ps -ef | grep -v grep | egrep -c ptx)
CFGCOUNT=$(cat /ptxrec/ptx_cpu.cfg | wc -l)
for FIND in $CFGFILE; do
ptxrlog -f $INPUT/$FIND.cfg -o $OUTPUT/$FIND.txt -i $INTERVAL -e
$DURATION &
done
sleep 100
printf "\n\n $SEARCH out of $CFGCOUNT ptxrlog processes were started
successfully
on $(date) for the $(hostname) node \n" >> /ptxrec/ptx_mon.log
if [ $SEARCH -lt $CFGCOUNT ]; then
echo "Not all of the ptxrlog processess were started on the $(hostname)
node on $(date)" \
| mail -s "ptxrlog error on $(hostname)" root@system.com
fi
----------------------------------------------------------------------------
-----------------------------------------------------------
Here is what the metrics .cfg file looks like for measuring CPU in this
case:
[root@localhost] /ptxrec/cfg-> more ptx_cpu.cfg
CPU/cpu0/user
CPU/cpu0/kern
CPU/cpu0/wait
CPU/cpu1/user
CPU/cpu1/kern
CPU/cpu1/wait
Regards,
Darryl
-----Original Message-----
From: Green, Simon [mailto:Simon.Green@EU.ALTRIA.COM]
Sent: Tuesday, June 10, 2003 4:09 AM
To: aix-l@Princeton.EDU
Subject: Re: Performance utility question
I'm not aware of anything which produces such a simple figure as iostat.
"filemon", which is in perfagent.tools, can be used to get filesystem
performance figures but it's a more complicated beast, not suitable for a
quick look at what's going on, and it doesn't do recurring intervals: just
collect a bunch of data, then analyse it.
Simon Green
Altria ITSC Europe s.a.r.l.
AIX-L Archive at http://marc.theaimsgroup.com/?l=aix-l&r=1&w=2
AIX FAQ at http://www.faqs.org/faqs/aix-faq/
N.B. Unsolicited email from vendors will not be appreciated.
> -----Original Message-----
> From: Rick YEH [mailto:rick@BULL.COM.TW]
> Sent: 11 June 2003 09:00
> To: aix-l@Princeton.EDU
> Subject: Performance utility question
>
>
> Hello,
> Does anyone know of any AIX utility that would allow me
> to get the io
> statistics that iostat give, but at the filesystem level. For
> example, in
> Solaris, we use iostat -xnp 600 144 to track per 10 miniutes
> and for 144
> times.
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:16:54 EDT