Apache 2 - child process memory size growing - MaxMemFree

[Christophe LEITIENNE]

Hi all,

We're using Apache 2 to serve big files generated by php or cgi
shell-script.
Apache 2 is configured to use prefork MPM.
It runs on a Linux RedHat 9.0, P200 with 128MB.

We're facing a problem:
Each child process memory size grows proportionnaly to the amount of data
transferred.
For example, if we transfer a 130MB file, the Apache child process then uses
more than 100MB of memory.
The size of each child process then decreases *very* slowly.
Obviously, after 2 file transfers of >100MB each, the machine starts
swapping, until it crashes.

We tried to use the "MaxMemFree" configuration directive (even if I did not
precisely understood its meaning).
When set to values like 100 or 1000 (kilobytes according to the
documentation), nothing changes.
When set to 1 or 2, then the child process frees its memory at the end of
each transaction, but *keep* growing during the transaction.

Under Apache 1.3.x, the behavior was different: no child process memory size
grew like that.

Did anyone faced the same problem ?
How to set correctly the "MaxMemFree" directive ?
Is there some other directive I missed ?

Thanks for your help,

Chris.

[Joshua Slive]

"Christophe LEITIENNE" <c.lei...@bgpartners.fr> wrote in message news:<bds6dm$5mh$1...@news-reader1.wanadoo.fr>...

> Hi all,
>
> We're using Apache 2 to serve big files generated by php or cgi
> shell-script.
> Apache 2 is configured to use prefork MPM.
> It runs on a Linux RedHat 9.0, P200 with 128MB.

So you are probably using 2.0.40? If so, you should upgrade to the latest
release (2.0.46) and I bet the problem will go away. Earlier releases of
apache 2 would queue up the entire request in order to determine the
content-length.

JOshua.

[jaya prakash]

I had to spend 3 hours to figure out that

application/x-httpd-php php

must be added to /usr/local/apache/conf/mime.types

[Christophe LEITIENNE]

I am already in 2.0.46

"Joshua Slive" <goo...@slive.ca> a écrit dans le message de
news:916ecaf4.03070...@posting.google.com...

[Christophe LEITIENNE]

Well, I found this seems to be a known bug, reopened some time ago on
bugzilla.
cf. http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11427.

Chris.

[Joshua Slive]

"Christophe LEITIENNE" <c.lei...@bgpartners.fr> wrote in message news:<bdum0a$9i0$1...@news-reader1.wanadoo.fr>...

> Well, I found this seems to be a known bug, reopened some time ago on
> bugzilla.
> cf. http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11427.

No. The bug you posted to is almost certainly win32 specific.

As I said, you need to start by upgrading to the most recent apache.

Joshua.

[Christophe LEITIENNE]

"Joshua Slive" <goo...@slive.ca> a écrit dans le message de
news:916ecaf4.03070...@posting.google.com...

> "Christophe LEITIENNE" <c.lei...@bgpartners.fr> wrote in message
news:<bdum0a$9i0$1...@news-reader1.wanadoo.fr>...
> > Well, I found this seems to be a known bug, reopened some time ago on
> > bugzilla.
> > cf. http://nagoya.apache.org/bugzilla/show_bug.cgi?id=11427.
>
> No. The bug you posted to is almost certainly win32 specific.
>

Not sure at all. All posts really sounds like my problem.

> As I said, you need to start by upgrading to the most recent apache.
>

As I said, I already did. I use 2.0.46.

Chris.

[Joshua Slive]

"Christophe LEITIENNE" <c.lei...@bgpartners.fr> wrote in message news:<be0g4p$3ti$1...@news-reader3.wanadoo.fr>...

I missed that part, sorry.

But if you are going to report a memory leak, you need to be extremely
specific about your system. You haven't said what modules you are using,
what changes you have made from the default config, etc.

Joshua.