Message-ID: <374EACA0.2DA10C20@sharp.fm>
Date: Fri, 28 May 1999 16:48:00 +0200
From: Graham Leggett <minfrin@sharp.fm>
MIME-Version: 1.0
To: new-httpd@apache.org
Subject: Re: Apache memory problems
References: <199905281355.JAA16602@devsys.jaguNET.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@apache.org

Jim Jagielski wrote:

> It's possible that the OS itself is having a memory leak. What system
> is this?

Solaris v2.6.

> Usage of child processes will pretty much even out eventually, assuming that no
> process gets a really large "job" to do (this can happen sometimes with PHP,
> esp. if the script is bad and it runs until the limit).

Ok.

> Try setting MaxRequestPerChild to something lower than what you have.
> That will/should stop the overwhelming of the box (but it won't fix
> the leak, however).

The trouble is that the processes getting stuck problem I am
experiencing still eats up processes, so reducing the
maxrequestsperchild doesn't help.

Unfortunately I am not sure if I am trying to find one bug or a
combination of bugs, the problems I have encountered so far are:

- leaks memory over time, eventually swamping the box.
- processes asked to gracefully restart don't, and get "stuck" trying to
read a request that's not there. (Visit
http://www.ericsson.se/server-status for an example, see processes
marked as "unavailable").
- sometimes (not always) Apache realises the processes are stuck, and
tries to TERM then KILL them.
- otherwise Apache just ignores these processes, spawning new ones to
replace them until MaxClients is reached.

Regards,
Graham
-- 
-----------------------------------------
minfrin@sharp.fm		"There's a moon
					over Bourbon Street
						tonight...