Problem with SAR

[johnnyos]

It appears a server has the following problem utilisation of the 4 processors is too much - does any one know how to resolve this and why the processors are getting such a hammering - when we have rebooted the server completely ?

00:00:00 proc %usr %sys %wio %idle %intr
01:00:00 0 20 71 2 5 3

02:00:00 0 20 70 5 2 3

03:00:00 0 19 71 6 1 3

04:00:00 0 19 71 7 0 3

05:00:00 0 18 71 7 0 3

06:00:00 0 19 72 2 4 3

Average 0 19 71 5 2 3


06:03:05 Unix restarts
07:00:01 0 22 68 3 5 3

08:00:00 0 27 59 3 8 3

08:20:00 0 21 65 9 2 3

08:40:01 0 20 67 7 3 3

09:00:01 0 22 64 9 2 3

09:20:03 0 21 66 4 3 5

09:40:00 0 22 67 2 4 5

10:00:00 0 25 64 2 4 5

10:20:00 0 26 63 2 5 5

10:40:00 0 22 67 2 5 5

11:00:01 0 24 65 2 4 5

Average 0 23 65 4 5 4

If you can help in releasing the CPU time instead of it being at 95% usage all the time ?

 

[stanhubble]

what processes are running?
what is the version of SCO?

 

[johnnyos]

root 28211 19409 TS 59 0 12:17:38 pts/19 0:00 ps -ef
UID PID PPID CLS PRI C STIME TTY TIME COMD
root 0 0 SYS 79 0 06:02:11 ? 12:49 sysproc
root 1 0 TS 70 0 06:02:13 ? 18:09 /sbin/init
root 822 821 TS 80 0 06:03:06 ? 0:00 /etc/ifor_pmd
root 794 1 TS 80 0 06:03:04 ? 0:00 /etc/conf/bin/idmknodd -
r /etc/conf
root 1009 1 TS 80 0 06:03:34 ? 0:00 hostmibd
root 1891 1 TS 85 0 06:04:11 ? 0:00 /usr/lib/saf/sac -t 300
root 825 822 TS 80 0 06:03:10 ? 0:00 /etc/ifor_sld
root 871 1 TS 88 0 06:03:14 ? 0:00 /usr/lib/mousemgr
root 824 822 TS 80 0 06:03:10 ? 0:00 /etc/sco_cpd
root 821 1 TS 70 0 06:03:06 ? 0:00 /etc/ifor_pmd
root 886 1 TS 80 0 06:03:15 ? 0:02 /usr/sbin/dlpid /etc/ins
t/nd/dlpidPIPE
nobody 1156 1 TS 70 0 06:03:40 ? 0:00 ./ns.httpd -d /usr/ns-ho
me/httpd-scohelphttp/config
root 12819 1 TS 80 0 06:30:02 ? 0:29 /usr/dlc/bin/_mprosrv /t
madata/tmasims -N TCP -S tmasrv -L 150000 -B 100000 -n
root 984 1 TS 80 0 06:03:30 ? 0:00 /usr/lib/sendmail -q1m -
bd
root 932 1 TS 80 0 06:03:28 ? 4:50 /usr/sbin/syslogd root 28211 19409 TS 59 0 12:17:38 pts/19 0:00 ps -ef
UID PID PPID CLS PRI C STIME TTY TIME COMD
root 0 0 SYS 79 0 06:02:11 ? 12:49 sysproc
root 1 0 TS 70 0 06:02:13 ? 18:09 /sbin/init
root 822 821 TS 80 0 06:03:06 ? 0:00 /etc/ifor_pmd
root 794 1 TS 80 0 06:03:04 ? 0:00 /etc/conf/bin/idmknodd -
r /etc/conf
root 1009 1 TS 80 0 06:03:34 ? 0:00 hostmibd
root 1891 1 TS 85 0 06:04:11 ? 0:00 /usr/lib/saf/sac -t 300
root 825 822 TS 80 0 06:03:10 ? 0:00 /etc/ifor_sld
root 871 1 TS 88 0 06:03:14 ? 0:00 /usr/lib/mousemgr
root 824 822 TS 80 0 06:03:10 ? 0:00 /etc/sco_cpd
root 821 1 TS 70 0 06:03:06 ? 0:00 /etc/ifor_pmd
root 886 1 TS 80 0 06:03:15 ? 0:02 /usr/sbin/dlpid /etc/ins
t/nd/dlpidPIPE
nobody 1156 1 TS 70 0 06:03:40 ? 0:00 ./ns.httpd -d /usr/ns-ho
me/httpd-scohelphttp/config
root 12819 1 TS 80 0 06:30:02 ? 0:29 /usr/dlc/bin/_mprosrv /t
madata/tmasims -N TCP -S tmasrv -L 150000 -B 100000 -n
root 984 1 TS 80 0 06:03:30 ? 0:00 /usr/lib/sendmail -q1m -
bd
root 932 1 TS 80 0 06:03:28 ? 4:50 /usr/sbin/syslogd
There are a load more but none of them look strange when running the 'top' command ?

The version is SCO 7.

regards

 

[aithosn8]

When we have this sort of trouble I run the following:

ps -efo pcpu,pid | awk '{if ($1 >= 10) print $0}'

This first command will give you the pid's of "hungry" processes.

Next I run:

for each_pid in `ps -efo pcpu,pid | awk '{if ($1 >= 10) print $2 }'`
do
ps -fp "$each_pid"
done

This will give you the process info on the "hungry" processes, and should give you enough information to investigate the cause.

aithosn8
aithosn8@lycos.com

"Impossible is a word to be found only in the dictionary of fools." - Napoleon Bonaparte

 

[stanhubble]

ok...couple of things...

if you stop your progress databases (tmasims) are the usr and sys values still pegged?

the init process should not have anywhere near that cpu time for being up only 6 hours or so. Are you running a graphical session on the console? if so, do you need to?
the syslogd is also very high. Are there a lot of entries going into syslog? They could possibly point to where the problem is.

hth
stan

 

[johnnyos]

Thanks to you both for your help - we found the problem as an old printing process which even though I had cold rebooted was still in the Print Manager. Deleting the process has now freed up a load of processor utilisation.

Thanks for the help much appreciated.