getaddrinfo: nodename nor servname provided, or not known

Question

I have a Ruby on Rails application that I am deploying on a computer running Mac OS X 10.6. The code where the problem arises is run by a delayed_job. The problem only occur

Answer 1

I ran into a similar situation today - deploying an app to a mac os x server, and receiving the 'getaddrinfo' message when I tried to access an external api. It turns out that the error occurs when the ssh session that originally launched the app is no longer active. That's why everything works perfectly if you ssh into your server and run commands manually (or launch the server manually) - as long as you keep your ssh session alive, this error won't occur.

Whether this is a bug or a quirk in OS X, I'm not sure. Here's the page that led me to the solution - http://lists.apple.com/archives/unix-porting/2010/Jul/msg00001.html

All I had to do was update my capistrano task to launch the app using 'nohup'. So changing

run "cd #{current_path} && RAILS_ENV=production unicorn_rails -c config/unicorn.rb -D"

to

run "cd #{current_path} && RAILS_ENV=production nohup unicorn_rails -c config/unicorn.rb -D"

did the trick for me.

Hope this helps somebody - it was quite a pain to figure out!