Rotate.pl using 80% CPU

[LtStriker]

Hey Everyone

We are running three Urban Terror servers on different ports (usually only peeps on one of them) from our Unmanaged Vertual Server (Specs seen below) with Ubuntu 10.10 x64

The server has been having lag spikes for the last month or so. During the spikes last night we jumped on to see what was going on and noticed that the rotate.pl script was using 80% CPU when normally our server barely scratches 14%.

We ended the process but expect it to come on again as part of a maintenance cycle? Sorry, a little new to Linux here. The sys log that rotate.pl was working with is really large. Does anyone know why that would be and what we can do to change or set a time for this rotate.pl to run??

Also how large should this sys log get before it is deleted or cleared out?

I have searched for solutions to this error without any success.
Has anyone seen this before?




Specs;
Base bandwidth:

2000 GB

Virtual CPU cores:

2
Memory size:
2048 MB

Standard hard drive size:
200 GB

IP addresses:
1

Thanks,

Striker

[Edge100x]

What script are you referring to when you ask about rotate.pl? This is not one that we run on our systems here. What lines are you seeing in the very large log?

[LtStriker]

Hey

We have not caught the issue arising again to date, but I did find this post that may help anyone else experiencing the same symptoms.


http://blog.jimmyr.com/Rotate_pl_eating_CPU_17_2008.php



Rotate.pl eating CPU Posted on Monday, March 17 2008

Today I had an issue in that webmin's /etc/webmin/bandwidth/rotate.pl was taking 99% of the CPU every hour. I guess it's always been there eating up the processor but I've never really noticed until now since it's a cron job. I deleted the entry calling rotate.pl. Suddenly /var/log/bandwidth started increasing at an insane rate. In a day it racked up 17 gigabytes. Deleting it didn't do anything. Here's the solution:

Edit the Syslog.conf with Vi
vi /etc/syslog.conf
click insert key
Add a "#" in front of
kern.=debug -/var/log/bandwidth
Restart the systemlog service
/sbin/service syslog restart
Delete the massive file
rm/var/log/bandwidth

You can check how much disk space any folder takes by typing du -hs foldernamehere. The commmand "df" will tell you how much disk space you have available. Logs definitely can eat tons of space if left unchecked.

Striker