Could you try 4.1.0 and/or 4.2.0-dev see if it's fixed?
http://www.php.net/downloads.php (4.1.0)
http://snaps.php.net/ (4.2.0-dev)

Please report the result and update PHP version.

--
Yasuo Ohgaki

Hi

Have installed 4.1, same problem as highlighted below:
[root@fweb2 /root]# strace -p 6739
flock(4, LOCK_EX


[root@fweb2 /root]# lsof | grep sessions
php        6739 fonetastic    4u   REG      72,10       163    294275 /www/external/sessions/sess_104bb2ffcc67af16af4095cbd92a8f64
php        6949 fonetastic    4u   REG      72,10       163    294275 /www/external/sessions/sess_104bb2ffcc67af16af4095cbd92a8f64

Do you have frames with your php script?
If yes, write session_write_close() to see if it helps.
I don't have time to check mod_files.c now, just trying to find work around ;)

PHP version 4.0.6 => 4.1.0

Hi

No Frames involved.  Thanks 

Chris

I guess your httpd is crashing. Please check your apache log. Do you get core file?

Hi

We are not running apache. We are runnig zeus with php running as a fastcgi responder.  At the moment, I am running a script which checks the state of the fastcgi/php processes and kill sany if locked on opening session file for more than 30 seconds.

Thanks

This may be fixed my patch

This bug has been fixed in CVS.

Since I've changed my patch a little. I'm not sure if it really fixed or not.
If you still have this problem, please reopen. 

does it work wit latest CVS now?

I have tried this using 4.1.2, and the problem is still occuring.  As soon as we are in a position to test 4.2, i will advise.

Leaving this with status feedback then.

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

Hi,

I'm experiencing the same problem using a compiled PHP 5.2.13 over an Apache 2.2.15.

The main problem is to reproduce the scenario that blocks the session file, as I can see seems aleatory; and just affects one of the php webapps hosted in this Apache.

Have you solved this issue? I tried to force a session_write_close to flush all lock processes, but I get the same result.

To not pause the development of this project I implemented a bash script that kills the apache thread that blocks the session, but it's just a temporal patch...

Any feedback would be great!

Thanks in advance.
--
X

I see the very same behaviour, but I'm running nginx with php spawned by spawn-fcgi.

I'm running php 5.3.5, spawn-fcgi 1.6.3 on arch linux.

I only use this machine for development; I start 5 php processes and after some inactivity, the processes hangs and the fastcgi interface is frozen.
Strace shows 2 processes blocked on flock(4, LOCK_EX

I guess that this is connected to the php session handling.