Hi,

my webserver is getting permanently down/unreachable.

I know that most of my perl scripts running on the server have room for optimization. But they are diverse and I am not enought a linux pro to know how I can track/log which script exactly is the source of the problem.

I got perl-scripts which are invoked for every pageload

I got scripts which run under fcgi

And I got a few ones which require a connection to a local mysql server.

Is there a log which gives me more info about when the server's lights gone off and where the cause was/is? I am on dedicated but managed hosting and don't have much access to low level system-files (1&1). Any tactics, log-loops I can build into my system or the scripts to gather information about cpu cycles used, system load etc?

Any help appreciated...

Maybe run top in batch mode and direct STDOUT to a file. I do this for quick diags on my own machines; hopefully you can do it on a hosted box.
Later versions of sysstat (with kernel 2.6.20 or later) have pidstat to do similar - handy for the history file.

Most of the others I've looked at concentrate on overall system metrics rather than per-process.