#293 Nginx & Unicorn pro

The latest version of Nginx on Ubuntu

add-apt-repository ppa:nginx/stable
apt-get update
apt-get install nginx

If the add-apt-repository ppa is not found install:
apt-get install python-software-properties

Thanks! Great episode. I would really like to know more about the benefits of unicorn over passenger. I've found some info in the comments on the github unicorn site.

Matt Freels says
Unicorn can restart your app without any lost requests or stalls.
Passenger drops your app and its workers, and then lets the next request bring the new version up.

I would say the biggest difference is that where Passenger is opinionated and fairly all-or-nothing, Unicorn gives you more transparency and control over its configuration (while still nicely managing workers).

Question is where is passenger ok to use? It seems easier to set up? (or is it?)

Hey Ryan B, read on #railscasts that you had the flu. Thanks for your Railscasts despite your being sick!

These deployment related screencasts are great! Keep them coming!

Nice railscast, I can't wait to see it with capistrano.

It would be interesting to show us how to setup and run multiple apps with different version of ruby and rails with this.

Thank you!

My deployment recipe

Great timing, I was just investigating the topic as I start witching to ruby 1.9 and wanted to keep that as painless as possible.

Can't wait for the deployment screencast now!

You've inspired me to give Unicorn a try. Quick question: Don't you need to run the following after you've created the /etc/init.d/unicorn symlink to unicorn_init.sh so that Unicorn automatically starts when the server is rebooted?

sudo /usr/sbin/update-rc.d -f unicorn defaults

I found this to be the case for a production server I'm running on Ubuntu. Can't wait to see more episodes on deployment. Thanks!

is it also possible to run multiple ruby versions at the same time? (e.g. 1.8.7 and 1.9.2)

Worth noting that you can use bundle install --binstubs --shebang ruby-local-exec instead of replacing with sed.

I can't find documentation for the --shebang flag anywhere, but it has the desired effect.

Excellent episode, looking forward to chef/capistrano in the future!

Edit: The --shebang option is defined here in the source.

nginx chef cookbook using the nginx app template described in this episode.

I always let the apps define the rack server. Some are fans or Thin while others go with Unicorn (or Rainbows). Capistrano+Foreman are a good combo for this. This is how I go about deploys.

Hey, I'm having trouble with redirecting. I have the same nginx.conf as in the tutorial, and when I do redirect_to posts_path in my controller, I get redirected to http://unicorn/posts, instead of http://example.com/posts. What could've I missed?

Even if I do just redirect_to '/', I get redirected to http://unicorn/.

i am getting this error when try to run sudo service unicorn restart

/usr/bin/env ruby-local-exec no such file or direcotry

i am using rbenv and have done 'rbenv local 1.9.2-p290' in my project

what can be the issue?

I'm using rvm and have managed to deploy 1 app with a certain degree of success. I'm now struggling to deploy another where my gems are defined in a gemset. Unicorn seems to be expecting them in the global space. Any idea what I can stuff in my unicorn.rb to tell it to use my gemset in this instance?

Anyone have experience setting up progressive uploads with an nginx/Unicorn setup? If so, can you post a gist of your setup?

I think I followed every step of this screencast, but I could not start unicorn. It seems that the pid files are busy (?). Any idea on how to solve this problem?

I'm using unicorn 4.2.0, which is the newest version of the gem.

When I run " sudo service unicorn restart" I get:

kill: 71: No such process

Couldn't reload, starting '/vagrant/bin/unicorn -D -c /vagrant/config/unicorn.rb -E development' instead
master failed to start, check stderr log for details

And here is my sderr log:

I, [2012-02-02T16:29:54.313595 #2007]  INFO -- : unlinking existing socket=/tmp/unicorn.todo.sock
I, [2012-02-02T16:29:54.316860 #2007]  INFO -- : listening on addr=/tmp/unicorn.todo.sock fd=3
/home/vagrant/.rbenv/versions/1.9.2-p290/lib/ruby/gems/1.9.1/gems/unicorn-4.2.0/lib/unicorn/http_server.rb:202:in `rename': Text file busy - (/vagrant/tmp/pids/0.933261511233375.2007, /vagrant/tmp/pids/unicorn.pid) (Errno::ETXTBSY)
        from /home/vagrant/.rbenv/versions/1.9.2-p290/lib/ruby/gems/1.9.1/gems/unicorn-4.2.0/lib/unicorn/http_server.rb:202:in `pid='
        from /home/vagrant/.rbenv/versions/1.9.2-p290/lib/ruby/gems/1.9.1/gems/unicorn-4.2.0/lib/unicorn/http_server.rb:133:in `start'
        from /home/vagrant/.rbenv/versions/1.9.2-p290/lib/ruby/gems/1.9.1/gems/unicorn-4.2.0/bin/unicorn:121:in `<top (required)>'
        from /vagrant/bin/unicorn:16:in `load'
        from /vagrant/bin/unicorn:16:in `<main>'

Thanks so much Ryan.

I has been beating my head against the wall trying to deploy on a clean unix box for days. This was straight forward and super easy to follow. I also love using vagrant to test my setup!

Keep em coming, best $9 a month I can imagine sending.

I'm anxious to see your capistrano chapter to this series. Any ideas when you're going to do it? Would be a great final installment.

Would anyone know the equivalent fix for rvm (i.e ruby-local-exec) below:

rbenv local 1.9.2-p290
sed -i 's/env ruby/env ruby-local-exec/' bin/*

Hey Ryan, great post!
I'm just having a har time to figure out, how unicorn.rb should be with RVM ? Could you give us an example?

How different would the process be if the target was a non-vagrant Ubuntu server?

How would I ago about dealing with the absence of the vagrantfile?

I am trying to build a deployment that has the fewest possible components, and it seems to me that while nginx and unicorn are necessary, vagrant is not.

Thanks!

Hello, great screencast..

I don't think it's a good manner to get the unicorn started by symlinked script in /etc/init.d, because the script which can be changed by deployer user is ran by root user at boot so it makes huge security vulnerability when the deployer user account is hacked throught the webapp..

Probably better would be to use some gem like whenever and get the init script started by the cron "every :reboot"

Problem in Ubuntu Server

If I run $ /etc/init.d/unicorn start, it runs just fine.
If I run $ service unicorn start, I get this error:

/usr/local/rvm/rubies/ruby-1.9.3-p194/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:36:in 'require': cannot load such file -- bundler/setup (LoadError)
from /usr/local/rvm/rubies/ruby-1.9.3-p194/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:36:in 'require'
from /var/rails/mbiri/shs/bin/unicorn:14:in ''

I cannot figure out why /etc/init.d would work, but service would not. It's driving me crazy trying to find the answer. I also think this is the reason why unicorn won't start on boot, because I'v already ran update-rc.d

So I followed the episode exactly and I'm getting 404 errors for anything in my public folder? Any ideas why...

Stack
nginx 0.7.65
unicorn
rails 3.2.3
Ubuntu 10.04
rvm latest
ruby 1.9.3-p194

I know about the rails config option to serve_static_assets, but wouldn't I want nginx to do that?

I'd like to share Unicorn+Nginx configuration with preload_app and optimized deployment scripts: https://gist.github.com/3052776

Nice episode, one question... I am trying to integrate this with facebook login using omniauth, but the return url using this configuration will be "unicorn" what doesn't match the url on facebook app. How can I make facebook accept this configuration? Thanks for any help.

I keep getting this error when trying to run the /etc/init.d/unicorn start script:

/usr/share/nginx/www/rails_apps/symbolfy/bin/unicorn:13:in `require': no such file to load -- rubygems (LoadError)
from /usr/share/nginx/www/rails_apps/symbolfy/bin/unicorn:13

But I don't have any trouble running:

bundle exec unicorn -c config/unicorn.rb -D

Can anyone help point me in the right direction?

I found a great unicorn init.d script working with rvm instead of rbenv.

RVM init.d script

Cheers everybody,

aaaaand it's broken

vagrant@lucid32:/vagrant$ bundle exec unicorn -c config/unicorn.rb -D -d
{:unicorn_options=>{:listeners=>[], :config_file=>"config/unicorn.rb"},
 :app=>
  #<Proc:0xa15f748@/home/vagrant/.rbenv/versions/1.9.3-p392/lib/ruby/gems/1.9.1/gems/unicorn-4.6.2/lib/unicorn.rb:39 (lambda)>,
 :daemonize=>true}
Exception `EOFError' at /home/vagrant/.rbenv/versions/1.9.3-p392/lib/ruby/gems/1.9.1/gems/unicorn-4.6.2/lib/unicorn/launcher.rb:46 - end of file reached
master failed to start, check stderr log for details

Does anyone have any ideas?

Hello, It works for me, but I can't make it run on production :( please help. I have no errors, and I have done exactly as the tutorial says. I'm on ubuntu, digital ocean.
I put <%= Rails.env %> on the third line
http://rubyonrailsperu.com
I have restarted everything a lot of times. and I'm pretty sure that the "- E production" flag it's on the right place.

Hi, Ryan, is this a typo ?

The init.d file means that nginx will will have been automatically restarted too.

Hey there,

You have pointed to unicorn.todo.sock. What is in that file? Where can I find information?

Thanks!
M.

I keep getting this error. There is no unicorn.sock file in that dir so I'm guessing that it's just not getting created?

 "connect() to unix:/tmp/unicorn.sock failed (2: No such file or directory) while connecting to upstream"

The answer for me was to add

listen "/tmp/unicorn.sock"

to my unicorn.rb file.