cclogd 100% cpu

[Juan Manuel López]

My Hp/Ux machine is always at 100% CPU !!!
This is the top exit:
System: ss03 Wed Aug 22 10:14:36 2001
Load averages: 2.27, 2.30, 2.34
114 processes: 102 sleeping, 12 running
Cpu states:
LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
2.27 17.8% 0.0% 82.2% 0.0% 0.0% 0.0% 0.0% 0.0%

Memory: 131592K (55260K) real, 111520K (49744K) virtual, 6196K free Page# 1/4

TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
? 12200 root 216 10 80K 244K run 5799:06 98.51 98.34 cclogd
? 28 root 152 20 0K 0K run 14:21 1.17 1.17 vxfsd
? 2709 genesys 154 24 536K 472K sleep 84:39 0.36 0.36 lca
? 512 root 154 20 32K 48K sleep 72:34 0.20 0.20 syncer

It is not normal ! Why ?

[melvyn burnard]

This has been discussed many times in this forum.
If you go search for the cclogd you will find many documents.
It is normally a result of a leak in the process, and has been fixed by patches which document this. As youy have not stated your OS verion or your DIAGNOSTIC product version, I cannot supply a patch number, or advise which version of DIags to update to.

[Sridhar Bhaskarla]

Please check if there are problems with the hardware by running xstm or cstm.

Also, it will be very helpful if you can run the sar command

$sar 1 10

Observe the %sys if it is more than 20, it could be due to some H/W issues

If there is %WIO > 20, then you have a disk bottleneck. If you see vxfsd continously on the top of the list, there is a lot of file system activity.

There are other options like sar -d , sar -b to check the disk activity

You can use sar -v to check the kernel parameters.

But I would check sar and check %SYS as there is something suspecious about cclogd that is taking quite some CPU. Run stm and check the particularly the chassis errorlogs. This could be a H/W issue.

-Sri

[eran maor · Accepted solution]

Hi

The server returned to "normal" after removing the core file and
restarting the cclogd daemon as follows:

# /sbin/init.d/diagnostic stop

# /sbin/init.d/diagnostic start

this will solve the problem .

[Juan Manuel López]

My Hp/Ux version is B.11.00
The diag version: OnlineDiag B.11.00.16.09 HPUX 11.0 Support Tools Bundle

Sri, next time I will try to find the answer on old messages.
Thanks in advance.
Juanma

[Praveen Bezawada]

Hi
The best thing to do in these situation is to monitor ths system through Glance. It displays quite a few metrics that can help in pinpointing the reason for this kind of system behaviour.
The metrics to study for high cpu utilisation would be the CPU report and proc list to know if one process is causing this or a whole group of process involved.
The paging statistics, thread states,openfiles list, to begin with can give you a good idea of what's happening.
If you do not have glance installed you can get a trial version of it from HP software downloads.
All the best...

...BPK...