I operate a Debian server with Apache and Raid software an..

Hi Everyone,

I operate a Debian server with Apache and Raid software and LVM.

On the server there are a large number of large websites.
>From time to time the loads are jumping and the server stop responding.

I've tried to locate the problem, but didn't.
I did notice that from time to time there are intensive writing to the
hard disk, of about 5M per sec, for about 5 seconds.

When I stop the apache it doesn't happen, but I can't identify the
source of problem.

I also checked on the server the apache status, and can't find any
problem.
I also checked the process list of the Mysql, and found nothing not
normal.

Does anyone knows or familiar with a tool that gives information about
the apache opened files, and information about their read and write.

Thank you for answering.
David

David from b wrote:

> Does anyone knows or familiar with a tool that gives information about
> the apache opened files, and information about their read and write.

Tracking down these sort of short-lived, intermittent bottle necks is
always more of an art than a science. However you should probably have
a look at the man pages for the following:

top
vmstat
lsof
ps
netstat

They (as a group) will provide you with a good picture of exactly what
is going on with your server. Looking at the right tool when the
problem starts will be the challenge.

Your problem might be excessive paging. This will happen if you're a
little short on RAM and Apache gets a flood of requests. It will spawn
as many children as necessary to meet the requests (subject to the
limits in httpd.conf of course). So you might want to focus on the CPU
time that has been consumed by kernel paging daemon (kswapd IIRC). If
your system is paging like mad, the total CPU time used for that process
might be a bit higher than "normal" - normal will vary from server to
server. You may need to watch it for a while.

Good luck.

Cheers,

James