FCGI Dispatch processes running amok...

I’ve deployed 5-6 Rails sites using Apache/FCGI before. On one in particular I’m having major problems with there being zombie FCGI processes hanging around. Sometimes I can get up to 20 on one box, before I have to ssh in and kill them manually.

Can anyone see a problem with this config file?

AddHandler fastcgi-script .fcgi

FastCgiIpcDir /var/lib/apache2/fastcgi

FastCgiServer /var/www/edeals/current/public/dispatch.fcgi -initial-env RAILS_ENV=production -processes 4 -idle-timeout 60

    FastCgiConfig -minProcesses 4 -maxProcesses 4 -processSlack 1

Also - I’m deploying via Capistrano. Have anything to do with it?

Nobody? Come on now…someone’s got to know something.

Seth wrote:

Nobody? Come on now....someone's got to know something.

>
> I've deployed 5-6 Rails sites using Apache/FCGI before. On one in
> particular I'm having major problems with there being zombie FCGI processes
> hanging around. Sometimes I can get up to 20 on one box, before I have to
> ssh in and kill them manually.
>
> Can anyone see a problem with this config file?
>
> <IfModule mod_fastcgi.c>
> AddHandler fastcgi-script .fcgi
>
> FastCgiIpcDir /var/lib/apache2/fastcgi
>
> FastCgiServer /var/www/edeals/current/public/dispatch.fcgi -initial-env
> RAILS_ENV=production -processes 4 -idle-timeout 60
> FastCgiConfig -minProcesses 4 -maxProcesses 4 -processSlack 1
> </IfModule>
>
> Also - I'm deploying via Capistrano. Have anything to do with it?
>

Ok I will give it a stab.

I had noticed the zombie processes on my server (Apache 1.3+FCGI) as well.
In my case it was due to logrotate. FCGI processes just didn't go away automatically
after logrotate runs. I had to modify the logrotate script (httpdlog) and added

/bin/killall dispatch.fcgi

to the postrotate section.

Hope this will give you some ideas to troubleshoot with.

Good luck,

Long