Apache, Mongrel clusters and sub domains

Rails
1

There seems to be a ton of documentation out there, but I can’t seem
to figure out how to get this to work. I’ve got two rails apps: app1,
which should be available at http://www.domain.com and app2 which
should be available at http://sub.domain.com. Both apps should be
served up by a mongrel cluster (app1 running on ports 8000-8002 and
app2 running on 9000-9002). App1 is working like a charm, however when
I go to http://sub.domain.com, I get app1 instead of app2. As a side
note, if I replace app2 with a static html page and take rails/mongrel
out of the equation entirely, the subdomain works just like it’s
supposed to.

My gut tells me it’s how I’ve got my virtual host files set up
(specifically the RewriteCond/RewriteRule parts). Could someone please
point me in the right direction? Thanks!

-Brian

Physical location:

app1: /var/www/app1
app2: /var/www/app2

/etc/apache2/sites-available/default:

NameVirtualHost *
<Proxy balancer://mongrel_cluster>
BalancerMember http://localhost:8000
BalancerMember http://localhost:8001
BalancerMember http://localhost:8002

<VirtualHost >
ServerName domain.com
DocumentRoot /var/www/app1/public/
.
.
.
RewriteEngine On
RewriteCond %{DOCUMENT_ROOT}/%{REQUEST_FILENAME} !-f
RewriteRule ^/(.)$ balancer://mongrel_cluster%{REQUEST_URI}
[P,QSA,L]

/etc/apache2/sites-available/app2:

NameVirtualHost *
<Proxy balancer://mongrel_cluster>
BalancerMember http://localhost:9000
BalancerMember http://localhost:9001
BalancerMember http://localhost:9002

<VirtualHost >
ServerName sub.domain.com
DocumentRoot /var/www/app2/public/
.
.
.
RewriteEngine On
RewriteCond %{DOCUMENT_ROOT}/%{REQUEST_FILENAME} !-f
RewriteRule ^/(.)$ balancer://mongrel_cluster%{REQUEST_URI}
[P,QSA,L]

2

I suspect your problem is that your clusters have the same name. You
need something more like:

<Proxy balancer://app1_mongrel_cluster>
BalancerMember http://localhost:8000
BalancerMember http://localhost:8001
BalancerMember http://localhost:8002

<Proxy balancer://app2_mongrel_cluster>
BalancerMember http://localhost:9000
BalancerMember http://localhost:9001
BalancerMember http://localhost:9002

Then of course update your rewrites to reflect that change.

good luck!
Tim

3

Thanks for the quick response Tim! That ended up not being the problem
but, after making the change you recommended, I stumbled across the
problem. As I was re-starting both mongrel clusters (just to be on the
safe side), I received an error that the cluster for app2 couldn’t be
shut down because of missing PID files. Sure enough, there should have
been a mongrel.9000.pid, mongrel.9001.pid and mongrel.9002.pid file in
the app2/tmp/pids directory but they weren’t there. So, even though
running “mongrel_rails cluster::start” seemed to be correctly starting
the cluster, in actuality, nothing was starting.

It turns out that somehow those files got deleted while the cluster
was still running. I needed to manually kill the process on the server
and then re-start the mongrel cluster for app2. Everything’s working
like a charm now.

-Brian