Symfony2 Capifony deployment - Conflicts on a same server - deployment

Question

I want to deploy different branchs of a Git repository on a same server.
I've updated app/config/deploy.rb
set :stage_dir, "app/config/deploy"
require "capistrano/ext/multistage"
set :stages, %w(prod stag stag2 stag3)
set :application, "MyApp"
I've created new files in app/config/deploy/
In app/config/deploy/stag.rb :
set :symfony_env_prod, "stag"
set :domain, "stag.example.com"
set :deploy_to, "/var/www/#{domain}"
# Other config
role :web, domain
role :app, domain
role :db, domain, :primary => true
In app/config/deploy/stag2.rb
set :symfony_env_prod, "stag"
set :domain, "stag2.example.com"
set :deploy_to, "/var/www/#{domain}"
# Other config
role :web, domain
role :app, domain
role :db, domain, :primary => true
I also launched cap stag2 deploy:setup.
When I deploy using cap stag deploy, stag.example.com is fine
Then, when I deploy cap stag2 deploy,stag2.example.com is file but stag.example.com get some changes from stag2. The source files of stag.example.com are correct. I suppose that there is a cache conflict.
Is Capifony able to deploy the same app in a same server properly ?
Edit : sounds similar Capistrano Multistage deploying to wrong directory
I also tried with set :deploy_to, "/var/www/stag2/#{domain}"

Answer 1

You have to use the server parameters instead of :domain.
stag.rb:
server "stag.example.com", :app, :web, :primary => true
stag2.rb:
server "stag2.example.com", :app, :web, :primary => true
Don't use :domain for this stuff.
See the docs.

Answer 2

I tried your solution #Elnur , replacing server by domain, deployment works fine but the problem still occurs...
After more investigation and tests I came up with a solution that seems to work.
I noticed that in app/autoload.php we use APC
$loader = new ApcUniversalClassLoader('xx.');
xx. is the A prefix to create a namespace in APC
http://symfony.com/doc/2.0/components/class_loader.html
I replaced by
$loader = new ApcUniversalClassLoader('xx.stagX');
Then restart Apache and rm -Rf app/cache/*
That seems to solve the problem. I would be surprised if the same APC name space could be used by 2 different websites on the same server.
This is the exception stack-trace that we got before, look how we go
from /var/www/stag5.XXXX.com/
to /var/www/stag6.XXXX.com/
#10 /var/www/stag6.XXXX.com/shared/vendor/cg-library/src/CG/Proxy/MethodInvocation.php(58): JMS\SecurityExtraBundle\Security\Authorization\Interception\MethodSecurityInterceptor->intercept(Object(CG\Proxy\MethodInvocation))
#11 /var/www/stag5.XXXX.com/releases/20121210053804/app/cache/stag/jms_aop/proxies/St-XxBundle-Manager-XxxManager.php(85): CG\Proxy\MethodInvocation->proceed()
#12 /var/www/stag6.XXXX.com/releases/20121210060841/src/Xx/XxxBundle/Controller/XxxController.php(85): EnhancedProxy_16c750f17d8113ffbee7fc3acdc4b1625ca7410b\__CG__\St\CoreBundle\Manager\TagsManager->findXxxxx(Object(Xx\XxBundle\Entity\Project))
#13 [internal function]: Xx\XxxxxBundle\Controller\XxxxController->getXxxAction(Object(Symfony\Component\HttpFoundation\Request))
#14 /var/www/stag5.XXXX.com/releases/20121210053804/app/cache/stag/classes.php(4591): call_user_func_array(Array, Array)
#15 /var/www/stag5.XXXX.com/releases/20121210053804/app/cache/stag/classes.php(4555): Symfony\Component\HttpKernel\HttpKernel->handleRaw(Object(Symfony\Component\HttpFoundation\Request), 1)
#16 /var/www/stag5.XXXX.com/releases/20121210053804/app/cache/stag/classes.php(5537): Symfony\Component\HttpKernel\HttpKernel->handle(Object(Symfony\Component\HttpFoundation\Request), 1, true)
#17 /var/www/stag5.XXXX.com/releases/20121210053804/app/bootstrap.php.cache(564): Symfony\Bundle\FrameworkBundle\HttpKernel->handle(Object(Symfony\Component\HttpFoundation\Request), 1, true)
#18 /var/www/stag5.XXXX.com/releases/20121210053804/web/app_stag.php(13): Symfony\Component\HttpKernel\Kernel->handle(Object(Symfony\Component\HttpFoundation\Request))
#19 {main}