The alert system will supply a warning like this:

[1144514]  protostar.cpu red Sun Oct 11 11:42:39 EDT 2009 up: 166 days, 2 users, 326 procs, load=620
LOAD AVG on protostar is 620
<br><br><a href=/mrtg/cs/cpuload/protostar.1.html><img src=/mrtg/cs/cpuload/protostar.1-day.png border=0></a>
<br><br>WARNING Levels: Yellow = 400, Red = 550



Top 15 Proceses (%CPU)
<br>
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
opensrf  16618 96.0  0.3 302192 65444 ?        R    Oct03 10355:28 /usr/sbin/apache2 -k start
opensrf  16233 70.0  0.2 289876 54936 ?        R    01:51 414:03 /usr/sbin/apache2 -k start
opensrf  16223 59.8  0.2 292380 54656 ?        R    06:25 190:06 /usr/sbin/apache2 -k start
opensrf  25193 56.8  0.2 289688 51708 ?        S    11:40   1:09 /usr/sbin/apache2 -k start
opensrf  25090 38.2  0.2 287880 52872 ?        S    11:39   1:11 /usr/sbin/apache2 -k start
opensrf  24912 26.0  0.2 287992 52932 ?        S    11:38   1:12 /usr/sbin/apache2 -k start
opensrf  24522 23.7  0.2 289688 52624 ?        S    11:37   1:11 /usr/sbin/apache2 -k start

Going on to protostar, and issuing top will let you know if you have processes that seem to be looping:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
16618 opensrf   20   0  295m  63m 6104 R  100  0.3  10895:20 apache2
16223 opensrf   20   0  285m  53m 5120 R  100  0.3 730:36.83 apache2
16233 opensrf   20   0  283m  53m 5720 R   99  0.3 953:57.18 apache2

In this case, we can start by zapping the offending processes:

uowadmin@protostar:~$ sudo kill -9 16618
uowadmin@protostar:~$ sudo kill -9 16223
uowadmin@protostar:~$ sudo kill -9 16233

This should buy some breathing room if nothing else.