[Catalyst] Re: Pb with fcgid on CentOS

jul.gil at gmail.com jul.gil at gmail.com
Fri Sep 24 07:45:03 GMT 2010


On Tue, Sep 21, 2010 at 16:23, jul.gil at gmail.com <jul.gil at gmail.com> wrote:
> Hi,
>
> I have a trouble to run Catalyst with Apache2/fcgid/suexec on CentOS 5.5
>
> I have a classical VirtualHost definition in the apache configuration with :
>
>    DocumentRoot /opt/myapp/root
>    Alias /static /opt/myapp/root/static
>
>    SuexecUserGroup appuser appuser
>
>    <Location /static>
>        SetHandler default-handler
>    </Location>
>
>    Alias / /var/www/myapp/myapp_wrapper.fpl/
>
>    <Location />
>        Options ExecCGI
>        Order allow,deny
>        Allow from all
>    </Location>
>
> The myapp_wrapper.fpl contains mainly :
>
> unshift @ARGV, "/opt/myapp/script/myapp_fastcgi.pl";
> exec @ARGV or die...
>
> to run the fastcgi command of Catalyst keeping the arguments. (The fpl
> script must be in /var/www because of suexec configuration).
>
> The wrapper script is called, the fastcgi.pl script also, but ends up with :
>
> STDIN is not a socket; specify a listen location at
> .../Catalyst/Engine/FastCGI.pm line 95.
>
> It seems to me, according fcgid documentation, that a socket is
> created, but it is not send to the myapp_fastcgi.pl
> A Dumper(\@ARGV) and Dumper(\%ENV) in the wrapper reveals that no
> arguments are send to the script, and that the environment variables
> only contains PATH, so error message is correct...
>
> Suggestions welcome...
>

I have found the problem : SELinux was activated by default, and the
fcgi wrapper was running with limited permissions. It is probably
possible to tune SELinux to allow the scripts to access ressources,
but deactivate it solves the problems.

-- 
Julien.



More information about the Catalyst mailing list