High system load: find causer?

[Elmi]

Hi,

I use a very slim Linux system with limited logging capabilities. Now I found that there is one application that runs amok after some time, it seems to cause a high system load so that other applications are not able to run properly.

The bad news here: this happens only after some hours and then only for a limited time so that starting "top" and watching the processes there will not work (I afraid I would fall asleep when using this solution).

So is there an other possibility to find the process that causes the troubles? Does anybody e.g. know a script or an application that watches the system load an writes into a logfile as soon as a process exceeds a given limit?

[Irithori]

Use a monitoring solution like
- Munin - Trac
- yellow : Xymon - Status @ Wed May 11 09:21:07 2011
- Nagios - The Industry Standard in IT Infrastructure Monitoring

[Elmi]

Hm, I afraid that doesn't works in my case. This system is a very small embedded linux that runs from flash and that can't be expanded by dozens of packages easily. So isn't there something more primitive available? A script that checks the load every second and stores process names that exceed a limit or something like that would be enough.

[Irithori]

Maybe this, spiced with a script for your usecase.
WYAE - MoSShE - Lightweight, secure server monitoring