Fedora server consuming all memory and grinding to a halt

[Tom Horsley]

On Mon, 16 Nov 2009 11:53:27 +1100
Langdon Stevenson wrote:

> Given these servers' perfect record in the past I am wondering if anyone
> can suggest what might be going on?  Or how best to try to track down
> the reason.

I had a similar problem once and put a cron job in to run every
few minutes and log the output of a ps command to a file with
an output format that showed PID, VM size, and process start time.

Then I whipped out a perl script to analyze the logs, find PIDs
that were running a long time, and kept using more memory the
longer they run.

Of course, that just found what was leaking, it didn't find out
why it was leaking :-).

(And no, I don't think I have any of the scripts around anymore :-).