Hi everyone,

First, thanks for the work you've done with peruser.

I am running peruser 0.3 (the last stable version) with Apache 2.2.9
under Debian Lenny, with about 150 different virtual hosts.

Each virtual hosts has his own chroot. So far, it's working quite
well, but I have some questions about all different "timeouts" that
exists in the configuration.

My peruser configuration is :

<IfModule peruser.c>
 # Multiplexer pool
 MinMultiplexers 6
 MaxMultiplexers 80
 Multiplexer www-data www-data
 ProcessorWaitTimeout 2 5

 # Fork limits
 ServerLimit 400
 MaxClients 20
 MaxRequestsPerChild 100

 # Processor defaults
 MinProcessors       0
 MinSpareProcessors  15
 MaxProcessors       100

 # Timeouts
 IdleTimeout 5
 ExpireTimeout 10
</IfModule>

For information, my typical vhost configuration is :

<Processor proc_foo>
        User foo
        Group foo
        Chroot /srv/web/foo
</Processor>

<VirtualHost *:80>
        ServerName foo.domain.com
        DocumentRoot /htdocs

        ErrorLog /var/log/apache2/foo/error.log
        CustomLog /var/log/apache2/foo/access.log combined

        ServerEnvironment proc_foo
        [...]
</VirtualHost>

I made the peruser configuration based on this page :
http://www.peruser.org/trac/projects/peruser/wiki/PeruserDocumentationInstallationDebian.

I try to understand the real meaning of IdleTimeout, ExpireTimeout and
ProcessorWaitTimeout. Indeed, after a request is made to one of the
vhosts, the processes forked remains alive for more than all the
timeouts set (say more than 1 min), and sometimes, in apache2 error's
log, I have something like :

[Fri Oct 23 09:30:18 2009] [warn] (3)No such process: kill SIGTERM

[Fri Oct 23 09:30:18 2009] [warn] (3)No such process: kill SIGTERM

[Fri Oct 23 09:30:18 2009] [warn] (3)No such process: kill SIGTERM

[Fri Oct 23 09:30:18 2009] [warn] (3)No such process: kill SIGTERM

[Fri Oct 23 09:30:18 2009] [warn] (3)No such process: kill SIGTERM

[Fri Oct 23 09:30:18 2009] [warn] (3)No such process: kill SIGTERM

[...]

as if apache was trying to kill old processes/processes with wrong
numbers. It continues until I reboot apache. If I strace apache during
such a thing, it acutally looks like old apache's processes...

By the way, I tried to find some answers by reading this :
http://www.directadmin.com/forum/archive/index.php/t-12794.html, but
this seems quite outdated.

Maybe should I use the current experimental version, that sounds
promising based on its changelog
(http://www.telana.com/pipermail/peruser/2009-September/001143.html) ?

Thanks for any help.

Regards,

-- 
Adrien.
_______________________________________________
Peruser mailing list
[email protected]
http://www.telana.com/mailman/listinfo/peruser

Reply via email to