Deploying rails app with realtime-rails gem and redis on Heroku - ruby-on-rails

Question

I have built a rails app following the model presented here:
http://mikeatlas.github.io/realtime-rails/
I am using Rails with the realtime-rails gem and redis. My application is working in my development environment and i want to move it to Heroku. I have already setup a Redis database at Redis-to-go and I now want to make whatever changes are necessary for the realtime-gem and associated socket-io server setup. At a minimum, I will need to modify the production portion of my application_controller:
def realtime_server_url
if Rails.env.development?
return 'HTTPLOCALHOST:5001'
end
if Rails.env.production?
return 'PRODUCION-LOCATION'
end
end
I have deployed the socket-io server (realtime-server) that is installed by instructions from the link above for realtime-server to a separate heroku instance. I then made the PRODUCION-LOCATION that url for the realtime -server dyno with port 5001 and both http and https. No joy.
Following the instructions the realtime-server folder was created at the top level of my project folder, parallel to the app folder. Does this mean i should include it in the main repository and somehow have it run from the same dyno as the app? If so, how do i go about starting it? The instructions say to start it by locally with:
cd realtime-server
foreman start
Not clear that I can do that through the heroku cli will be able to run in the same instance and how is it started.
=============
Update
Found documentation on heroku that made me realize, I need to set the REDISCLOUD_URL for the heroku dyno running the realtime-server using:
heroku config:add REDISCLOUD_URL='redis_cloud_url'
and that in production it wasnt using the 5001 port:
if (process.env.NODE_ENV != 'production') {
port = 5001; // run on a different port when in non-production mode.
}
Also, found form console log that the realtime-server was enforcing HTTPS.
Now, the blocking issue seems to be the request of /socket.io/socket.io.js from the realtime-server which returns:
503 (Service Unavailable)
So far, it seems that separating the realtime-server from the rails app repository was the right move.
Looking through the code for the realtime-server to determine how that is routed...
================
Update
i looked at the logs on your advice, thanks. saw that the realtime-server was crashing because it wasnt liking the port, so I tried to set the PORT variable 443, 3000, 5001 variously to no avail using:
heroku config:add PORT='443'
based on this code from the enviornment.js file of the realtime-server:
var port = process.env.PORT || 5001;
if (process.env.NODE_ENV != 'production') {
port = 5001; // run on a different port when in non-production mode.
}
here is an excerpt of logs:
2015-02-22T19:26:45.512317+00:00 heroku[api]: Set PORT config vars by tmt#breakthroughtek.com
2015-02-22T19:26:45.512317+00:00 heroku[api]: Release v5 created by tmt#breakthroughtek.com
2015-02-22T19:26:45.754327+00:00 heroku[web.1]: State changed from crashed to starting
2015-02-22T19:26:48.318928+00:00 heroku[web.1]: Starting process with command `node forever.js`
2015-02-22T19:26:49.540190+00:00 app[web.1]: Detected 512 MB available memory, 512 MB limit per process (WEB_MEMORY)
2015-02-22T19:26:49.540213+00:00 app[web.1]: Recommending WEB_CONCURRENCY=1
2015-02-22T19:26:49.957808+00:00 app[web.1]: STARTING ON PORT: 5001
2015-02-22T19:27:44.163425+00:00 heroku[api]: Scale to web=1 by tmt#breakthroughtek.com
2015-02-22T19:27:48.811493+00:00 heroku[web.1]: Error R10 (Boot timeout) -> Web process failed to bind to $PORT within 60 seconds of launch
2015-02-22T19:27:48.811551+00:00 heroku[web.1]: Stopping process with SIGKILL
2015-02-22T19:27:49.561043+00:00 heroku[web.1]: State changed from starting to crashed
2015-02-22T19:27:49.561720+00:00 heroku[web.1]: State changed from crashed to starting
2015-02-22T19:27:49.534451+00:00 heroku[web.1]: Process exited with status 137
2015-02-22T19:27:51.936860+00:00 heroku[web.1]: Starting process with command `node forever.js`
2015-02-22T19:27:53.289025+00:00 app[web.1]: Detected 512 MB available memory, 512 MB limit per process (WEB_MEMORY)
2015-02-22T19:27:53.289046+00:00 app[web.1]: Recommending WEB_CONCURRENCY=1
2015-02-22T19:27:53.703573+00:00 app[web.1]: STARTING ON PORT: 5001
2015-02-22T19:28:51.991836+00:00 heroku[web.1]: Stopping process with SIGKILL
2015-02-22T19:28:51.991836+00:00 heroku[web.1]: Error R10 (Boot timeout) -> Web process failed to bind to $PORT within 60 seconds of launch
2015-02-22T19:28:52.758191+00:00 heroku[web.1]: Process exited with status 137
2015-02-22T19:28:52.764783+00:00 heroku[web.1]: State changed from starting to crashed
2015-02-22T19:31:22.240362+00:00 heroku[api]: Set PORT config vars by tmt#breakthroughtek.com
2015-02-22T19:31:22.240362+00:00 heroku[api]: Release v6 created by tmt#breakthroughtek.com
2015-02-22T19:31:22.378770+00:00 heroku[web.1]: State changed from crashed to starting
2015-02-22T19:31:24.766187+00:00 heroku[web.1]: Starting process with command `node forever.js`
2015-02-22T19:31:26.316332+00:00 app[web.1]: Detected 512 MB available memory, 512 MB limit per process (WEB_MEMORY)
2015-02-22T19:31:26.316353+00:00 app[web.1]: Recommending WEB_CONCURRENCY=1
2015-02-22T19:31:26.717561+00:00 app[web.1]: STARTING ON PORT: 5001
===========
Update
looking at the logs and seeing the system was still picking port 5001, I checked the heroku environement via :
heroku config
and saw that the NODE_ENV variable was not set. Did a:
heroku config:add NODE_ENV='production'
and now the js file is loading. YAY!!! .. Thanks, d.danailov :)
Have to resolve some issue with rails on heroku erring with a missing template error on access to the admin area:
2015-02-22T20:04:36.492199+00:00 app[web.1]: Processing by LocationsController#index as HTML
2015-02-22T20:04:36.497052+00:00 app[web.1]: * "/app/app/views"
2015-02-22T20:04:36.497054+00:00 app[web.1]: * "/app/vendor/bundle/ruby/2.0.0/gems/devise-3.4.1/app/views"
2015-02-22T20:04:36.497059+00:00 app[web.1]: app/controllers/locations_controller.rb:8:in `index'
2015-02-22T20:04:36.497050+00:00 app[web.1]: ActionView::MissingTemplate (Missing template locations/index, application/index with {:locale=>[:en], :formats=>[:html], :variants=>[], :handlers=>[:erb, :builder, :raw, :ruby, :coffee, :jbuilder]}. Searched in:
2015-02-22T20:04:36.497062+00:00 app[web.1]:
2015-02-22T20:04:36.497057+00:00 app[web.1]: ):
2015-02-22T20:04:36.497060+00:00 app[web.1]:
2015-02-22T20:04:36.497056+00:00 app[web.1]: * "/app/vendor/bundle/ruby/2.0.0/gems/realtime-0.0.12/app/views"
And of course, the page is working fine in development. So I am going to start with the assumption it has something to do with sprockets/asset pipeline.
But i think this portion may be solved. I will close once I verify I can send realtime msgs.
=============
Update
SOLVED: beware you dont use redis elsewhere in your app.
HEADS UP: i had a sneaky little Redis.new in an initializer that was wiping out my Redis config settings.

Answer 1

most of the solution was setting the NODE_ENV & REDISCLOUD_URL vars for the realitme server running in a separate instance.