Rails middleware to provide information about errors during requests

application.rb

module MyApp
  class Application < Rails::Application
    if Rails.env == 'test'
      require 'diagnostic'
      config.middleware.use(MyApp::DiagnosticMiddleware)
    end
  end
end

diagnostic.rb

module MyApp
  class DiagnosticMiddleware
    FILENAME = 'diagnostic.txt'

    def initialize(app)
      @app = app
    end

    def call(env)
      return @app.call(env)
    rescue StandardError => e
      trace = e.backtrace.select{ |l|l.start_with?(Rails.root.to_s) }.join("\n    ")
      msg = "#{e.class}\n#{e.message}\n#{trace}\n"
      File.open(FILENAME, 'a') { |f| f.write msg }
      raise e
    end
  end
end

This is great, really helpful when testing Capybara request specs that use JS drivers. Thanks man!

My god thank you for this, i was going crazy trying to get the exception into the log.

BTW, you can just stick the lines in test.rb instead of application.rb with checking the environment, thats what test.rb is there for.

also, putting the lines straight into log/test.log with some delimiters for visual ease help out a lot as well

module MyApp
  class DiagnosticMiddleware
    FILENAME = File.join(Rails.root, 'log', 'test.log')

    def initialize(app)
      @app = app
    end

    def call(env)
      return @app.call(env)
    rescue StandardError => e
      trace = e.backtrace.select{ |l|l.start_with?(Rails.root.to_s) }.join("\n    ")
      msg = "#{e.class}\n#{e.message}\n#{trace}\n"
      File.open(FILENAME, 'a') { |f| 
        f.write "\n"
        f.write "\n"
        f.write " ---   BACKTRACE   --- \n"
        f.write "\n"

        f.write msg 

        f.write "\n"
        f.write " ---   BACKTRACE   --- \n"
        f.write "\n"
        f.write "\n"
      }
      raise e
    end
  end
end

Works like a charm. Thanks Steve!

The only way I found, to track down Internal server errors that come from ajax calls. You made my day! Thanks! (also very convenient additions by amalagaura and scottklein!)

I placed the diagnostic.rb file inside the config directory but needed to replace the require 'diagnostic line in application.rb with the following:

require File.expand_path("../diagnostic.rb", __FILE__)

Awesome. Thank you!

That is absolutely awesome! Thank you!
@scootklein's modifications make it even more powerful by seeing the backtrace along with the request. This makes debugging integration tests very similar to debugging in development env.

Excellent - thanks @steve9001 and @scootklein. I put the 'middleware.use' line into config/environments/test.rb to keep my application.rb file clean.

Thanks, very useful. I just print to stdout:

module MyApp
  class DiagnosticMiddleware

    def initialize(app)
      @app = app
    end

    def call(env)
      return @app.call(env)
    rescue StandardError => error
      puts error.message
      puts error.backtrace
      raise error
    end
  end
end

Proved helpful in a case of "Internal server error" while running a spec which did not provide much detail either on the page or in the test log file.

Thanks a bunch !

the same here, i would also like to have it as a spec, so it does not work for me.

It didn't work for me until I changed it to rescue Exception instead of StandardError.

application.rb should be:

if Rails.env == 'test'
  require File.expand_path('../diagnostic', __FILE__)
  config.middleware.use(MyApp::DiagnosticMiddleware)
end

I happened to handle this a little bit differently -- by detecting RoutingErrors on image paths: https://gist.github.com/mcmire/68cd9c74ba765a2d5dfb14abf58409aa