Last month I had snmp & mrtg installed to measure bandwidth usage for an IRCD network running off my server.

Since the reboot, I noticed two things happening:

1. processor usage keeps spiking.
2. one of the graphs (localhost_4) stopped working.

Now I thought the two were related - the lass that configured mrtg changed "localhost4" to work against the IP address directly and the graph started again. I suspected that some process was struggling and timing out showing these graphs, so imagined that CPU spikes would return to normal once the mrtg thang was corrected - but no joy.

To see the graphs, look at:

http://preacher.foreverchat.net/mrtg
http://preacher.foreverchat.net/mrtg/pr ... pusum.html

The large CPU spikes begin wednesday - looking at the weekly and monthly usage you can see how the server was barely taxed prior to that.

Note: I installed the same monitoring onto another server, which originally displayed the same usage - and that's not been rebooted yet:

http://iceland.foreverchat.net/mrtg

Any ideas, anyone?