Hi,
I have requirement to produce a report on high CPU utilization processes and the processes lying on the CPU for long time (Long running queries). The report should append into the files every 3 minutes. I use prstat to pull top 5 and found the following result.
(USER01)/opt/tuxedo/appsrv_clasuat> prstat -u tuxedo -a -s time -s cpu -n 5 0 1
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP
7241 tuxedo 269M 152M sleep 59 0 1:07.36 1.2% cleard/4
23213 tuxedo 198M 95M sleep 59 0 0:11.15 0.7% cleard/4
18524 tuxedo 274M 158M sleep 59 0 0:31.10 0.7% cleard/4
24827 tuxedo 202M 120M sleep 59 0 0:06.46 0.5% WSH/4
15415 tuxedo 139M 37M sleep 59 0 0:07.45 0.5% cleard/4
NPROC USERNAME SIZE RSS MEMORY TIME CPU
49 tuxedo 7129M 2524M 32% 5:51.53 6.6%
Total: 49 processes, 191 lwps, load averages: 0.46, 0.69, 0.70
Along with this report, I need append the the each tuxedo process (cleard_relcom) for the above process id (7241,23213,18524,24827,24827) in the report as below. Is it possible to produce single report with both of these result?
(USER01)/opt/tuxedo/appsrv_clasuat> ps -ef | grep -v grep | grep 2775
tuxedo 2775 1 3 12:18:57 ? 100:01 cleard_relcom -C dom=clasuat2 -g 5 -i 21 -u clauat02 -U /opt/tuxedo/appsrv_clasu
(USER01)/opt/tuxedo/appsrv_clasuat>
I need to merge both the report and append to a file. Can you please help me?
Thina
Use > to write output to a file
and >> to append to the file
eg:
$ prstat -u tuxedo -a -s time -s cpu -n 5 0 1 > report.txt
$ ps -ef | grep -v grep | grep 2775 >> report.txt
Hi Chubler_XL,
Thanks for your reply. I need output in the below format. For each process id listed in the prstat command, I need fetch the process detail through ps command and append in the same row of prstat command.
It means, additional column of PROCESS_DETAIL (cleard_relcom -C dom=clasuat2 -g 5 -i 21 -u clauat02) added in each row of the respective process id listed in the prstat command.
Expected output.
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP PROCESS_DETAIL
7241 tuxedo 269M 152M sleep 59 0 1:07.36 1.2% cleard/4 cleard_relcom -C dom=clasuat2 -g 5 -i 21 -u clauat02 -U /opt/tuxedo/appsrv_clasu
23213 tuxedo 198M 95M sleep 59 0 0:11.15 0.7% cleard/4 cleard_batch -C dom=clasuat2 -g 5 -i 21 -u clauat02 -U /opt/tuxedo/appsrv_clasu
18524 tuxedo 274M 158M sleep 59 0 0:31.10 0.7% cleard/4 cleard_batch -C dom=clasuat2 -g 5 -i 21 -u clauat02 -U /opt/tuxedo/appsrv_clasu
24827 tuxedo 202M 120M sleep 59 0 0:06.46 0.5% WSH/4 WSH -C dom=clasuat2 -g 5 -i 21 -u clauat02 -U /opt/tuxedo/appsrv_clasu
15415 tuxedo 139M 37M sleep 59 0 0:07.45 0.5% cleard/4 cleard_cb -C dom=clasuat2 -g 5 -i 21 -u clauat02 -U /opt/tuxedo/appsrv_clasu
NPROC USERNAME SIZE RSS MEMORY TIME CPU
49 tuxedo 7129M 2524M 32% 5:51.53 6.6%
Total: 49 processes, 191 lwps, load averages: 0.46, 0.69, 0.70
How about this:
addcmdline ()
{
PID=`echo $1`
REM=${PID#* }
PID=${PID% $REM}
CMD=`ps -p $PID -o cmd | tail -1`
printf "%-72s%s" "$1" "$CMD"
}
ADD_DET=0
prstat -u tuxedo -a -s time -s cpu -n 5 0 1 | while read
case "$REPLY" in
*\ PID\ USER*)
echo "$REPLY PROCESS_DETAIL"
ADD_DET=1 ;;
*\ NPROC\ USER*)
echo "$REPLY"
ADD_DET=0 ;;
*) [ $ADD_DET -eq 1 ] && line=`addcmdline "$REPLY"`
echo "$REPLY" ;;
esac
done
Note this uses ps -p 2775 -o cmd to get the command line of a process (2775 in the example). Your ps may have a different code for the -o option, so if it dosn't work check man ps for the correct code to use instead of "cmd".