CPU spikes even with empty ET servers (Linux)

Discussion for Admins of ETPro/BayonET servers.
If you don't run a server, please don't post here...

Moderators: Forum moderators, developers

Post Reply
Naked Pope
Posts: 60
Joined: Wed Aug 13, 2003 10:00 pm

CPU spikes even with empty ET servers (Linux)

Post by Naked Pope »

Hi.

I'm running a Linux server with the following specs:

AMD Athlon(tm) XP 2400+
1 GByte RAM
Debian Linux 3.0 (stable)
Kernel 2.4.22

The normal CPU and memory states look like this:

Code: Select all

CPU states:   0.0% user,   0.0% system,   0.0% nice,  100.0% idle
Mem:   1000464K total,   813876K used,   186588K free,    96000K buffers
Swap:  1028148K total,     2152K used,  1025996K free,   542020K cached
But every 2 or 3 minutes there are CPU spikes up to 20% on each server. It doesn't matter if i'm running etmain or etpro (with or without pb).

The following table shows the CPU states in 5 second steps with only one etmain server running:

Code: Select all

  PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
25995 tussi     10   0 71156  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     10   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     11   0 71160  69M  1888 S     0.1  7.0   0:04 etded.x86
25995 tussi     11   0 71160  69M  1888 S     0.0  7.0   0:04 etded.x86
25995 tussi     13   0 71160  69M  1888 S     0.3  7.0   0:04 etded.x86
25995 tussi     15   0 71160  69M  1888 S    11.4  7.0   0:04 etded.x86
25995 tussi     14   0 71160  69M  1888 S    19.9  7.0   0:05 etded.x86
25995 tussi     15   0 71160  69M  1888 S    16.7  7.0   0:06 etded.x86
25995 tussi     15   0 71160  69M  1888 R     0.0  7.0   0:06 etded.x86
25995 tussi     15   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     15   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.1  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 R     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 R     0.0  7.0   0:06 etded.x86
25995 tussi     16   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     17   0 71160  69M  1888 S     0.1  7.0   0:06 etded.x86
25995 tussi     17   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     17   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     17   0 71160  69M  1888 R     0.0  7.0   0:06 etded.x86
25995 tussi     17   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     18   0 71160  69M  1888 S     0.1  7.0   0:06 etded.x86
25995 tussi     18   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     18   0 71160  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     18   0 71164  69M  1888 R     0.0  7.0   0:06 etded.x86
25995 tussi     19   0 71164  69M  1888 S     0.1  7.0   0:06 etded.x86
25995 tussi     19   0 71168  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     19   0 71168  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     19   0 71168  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     10   0 71168  69M  1888 R     0.0  7.0   0:06 etded.x86
25995 tussi      9   0 71168  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi      9   0 71168  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi      9   0 71168  69M  1888 S     0.0  7.0   0:06 etded.x86
25995 tussi     10   0 71168  69M  1888 S     0.1  7.0   0:06 etded.x86
25995 tussi     13   0 71168  69M  1888 R     0.5  7.0   0:06 etded.x86
25995 tussi     17   0 71168  69M  1888 S    16.1  7.0   0:07 etded.x86
25995 tussi     15   0 71168  69M  1888 S    19.7  7.0   0:08 etded.x86
25995 tussi     15   0 71168  69M  1888 S    11.8  7.0   0:09 etded.x86
25995 tussi     16   0 71168  69M  1888 S     0.1  7.0   0:09 etded.x86
25995 tussi     16   0 71168  69M  1888 R     0.0  7.0   0:09 etded.x86
25995 tussi     16   0 71168  69M  1888 S     0.0  7.0   0:09 etded.x86
.
.
.
And this is some mrtg output i did last night running my 3 servers:

Traffic:
Image
CPU Load:
Image

There's no traffic from 3am - 8am so I assume that evey server was empty. But you can see that the CPU spikes are still there (reaching 74%!). On a full server those spikes have to be added to the normal CPU usage of each running server and seem to cause bad lags sometimes :/ There's nothing unusual in the log files.

I thought it could be related to this problem
Q: "my linux et server is taking 90% cpu even when empty, why?"
A: You have TLS installed, et doesnt like it. Do export LD_ASSUME_KERNEL=2.4.1 before calling etded to fix it.
but TLS isn't installed on this machine and exporting LD_ASSUME_KERNEL doesn't solve this problem.

I'm stuck now and hope for your help :(
BOTA
Posts: 136
Joined: Wed Oct 23, 2002 3:53 am

Post by BOTA »

I run Linux servers. I will monitor to see if I detect this.
Naked Pope
Posts: 60
Joined: Wed Aug 13, 2003 10:00 pm

Post by Naked Pope »

After trying some other 2.4 kernels, the upgrade to 2.6.5 did it :lol: Only a few other people reported almost the same strange phenomen (they were running other software, not ET) in some Linux news groups, so I can't say what's causing this problem :? AFAIK it seems not to be hardware related but appears in the later 2.4 kernel series.
BOTA
Posts: 136
Joined: Wed Oct 23, 2002 3:53 am

Post by BOTA »

I myself build everything from source, that allows me to do so.
Simply because most software is so well documented,
you do not need to be a brain-surgeon to figure out what is best for your server.
You just need time to learn, which I usualy have.

I use a 2.4 kernel.
And have not detected the aforementioned problem of cpu spikes.
Post Reply