Memory leaks in RT and FastCGI source

(1)
We noticed memory leaks in relation to Apache on our RT main server. Thus,
we have to do scheduled re-starts of Apache. Does anyone have any similar
experience of this?

(2)
We’re using FastCGI, since it works better for us than mod_perl. But it
concerns me that this source distribution is old:

mod_fastcgi-2.4.2.tar.gz (24-Nov-2003 19:07 96k)

Is someone maintaining it?

Tomas A. P. Olaj, email: tomas.olaj@usit.uio.no, web: folk.uio.no/tomaso
University of Oslo / USIT (Center for Information Technology Services)
System- and Application Management / Applications Management Group

Hi Tomas,

(1)
We noticed memory leaks in relation to Apache on our RT main server. Thus,
we have to do scheduled re-starts of Apache. Does anyone have any similar
experience of this?

Yes, we have the same situation! But ist strange, it does not happen if we
have high load at the system, it only happens if nobody works! It starts
with lost connection from the FCGI Process to the DB.

(2)
We’re using FastCGI, since it works better for us than mod_perl. But it
concerns me that this source distribution is old:

mod_fastcgi-2.4.2.tar.gz (24-Nov-2003 19:07 96k)

No idea, i have the same release installed…

I’m using Apache 2.0.54.

Torsten

Tomas A. P. Olaj, email: tomas.olaj@usit.uio.no, web: folk.uio.no/tomaso
University of Oslo / USIT (Center for Information Technology Services)
System- and Application Management / Applications Management Group
http://lists.bestpractical.com/cgi-bin/mailman/listinfo/rt-users

Be sure to check out the RT Wiki at http://wiki.bestpractical.com

Buy your copy of our new book, RT Essentials, today!

Download a free sample chapter from http://rtbook.bestpractical.com

On the marvelous Tue, 25 Oct 2005, Torsten Brumm wrote kindly to me …

(1)
We noticed memory leaks in relation to Apache on our RT main server. Thus,
we have to do scheduled re-starts of Apache. Does anyone have any similar
experience of this?

Yes, we have the same situation! But ist strange, it does not happen if we
have high load at the system, it only happens if nobody works! It starts
with lost connection from the FCGI Process to the DB.

Thanks, it happends slow on our server, since I guess RT is busy all the
time. BP may have the same problems on their servers?

(2)
We’re using FastCGI, since it works better for us than mod_perl. But it
concerns me that this source distribution is old:

mod_fastcgi-2.4.2.tar.gz (24-Nov-2003 19:07 96k)

No idea, i have the same release installed…

It concerns me a bit if this source is maintained or not, and that BP
recommends FastCGI. It could be that this is the main source for bad
garbage collection.

I’m using Apache 2.0.54.

We are using Apache 2.0.52

Cheers,
Tomas

Tomas A. P. Olaj, email: tomas.olaj@usit.uio.no, web: folk.uio.no/tomaso
University of Oslo / USIT (Center for Information Technology Services)
System- and Application Management / Applications Management Group

(1)
We noticed memory leaks in relation to Apache on our RT main server. Thus,
we have to do scheduled re-starts of Apache. Does anyone have any similar
experience of this?

(2)
We’re using FastCGI, since it works better for us than mod_perl. But it
concerns me that this source distribution is old:

mod_fastcgi-2.4.2.tar.gz (24-Nov-2003 19:07 96k)

Is someone maintaining it?

Are you seeing the leaks in the FastCGI process (mason_handler.fcgi) or
the mod_perl children? I’ve not heard much bad about mod_fastcgi,
though I did notice that someone implemented a mod_fastcgi workalike for
Apache2 that’s all-new.

Jesse

Are you seeing the leaks in the FastCGI process (mason_handler.fcgi) or
the mod_perl children? I’ve not heard much bad about mod_fastcgi,
though I did notice that someone implemented a mod_fastcgi workalike for
Apache2 that’s all-new.

Memory leaks in long-running perl programs should be more or less
expected. I thought both mod_perl and fastcgi had mechanisms to
kill and restart the processes after a certain amount of service.
You shouldn’t have to kill the whole apache process to accomplish
this although I don’t know exactly what does control it in fastcgi.

Les Mikesell
les@futuresource.com

We noticed memory leaks in relation to Apache on our RT main
server. Thus, we have to do scheduled re-starts of Apache. Does
anyone have any similar
experience of this?

are you sure they are memory leaks, not just large memory usage?
remember how memory is allocated to processes and when it is returned
to the system, especially for long-running processes.

Are you seeing the leaks in the FastCGI process (mason_handler.fcgi) or
the mod_perl children? I’ve not heard much bad about mod_fastcgi,
though I did notice that someone implemented a mod_fastcgi workalike for
Apache2 that’s all-new.

Jesse

I also tried to switch to fastcgi+apache2.0.55 - freshly compiled.
Works great and is much faster, however after 5-20minutes processes
start to hang and once that happens it’s all downhill - 500 errors
like crazy.

[notice] Apache/2.0.55 (Unix) mod_ssl/2.0.55 OpenSSL/0.9.7a
mod_fastcgi/2.4.2 configured – resuming normal operations

and then

[Thu Oct 20 12:53:19 2005] [error] [client 1.1.1.1] FastCGI: comm with
server “/opt/rt3/bin/mason_handler.fcgi” aborted: error parsing
headers: malformed header ‘Location=“/Ticket/Display.html?id=18530”’,
referer: http://my.server

a rt-test-dependencies is fine…

I tried playing around with the idle-timeout, processes, killInterval
options but never found a workaround. I’d rather be stable with
mod_perl.

FastCgiServer /opt/rt3/bin/mason_handler.fcgi -idle-timeout 120 -processes 10
FastCgiConfig -minProcesses 10 -killInterval 10