Rackstash - Sane Logs for Rack and Rails

A gem which tames the Rack and Rails (2.3.x and 3.x) logs and generates JSON log lines in the native Logstash JSON Event format.

It is thus similar to the excellent Lograge by Mathias Meyer. The main difference between Rackstash and Lograge is that Lograge attempts to completely remove the existing logging and to replaces it with its own log line. Rackstash instead retains the existing logs and just enhances them with structured fields which can then be used in a Logstash environment. By default, Rackstash collects the very same data that Lograge collects plus the original full request log.

Given that Rackstash deals with potentially large amounts of log data per request, it might be difficult to use with syslog. You would have to set the supported message size rather high and have to make sure that all syslog servers can handle the large messages. Rackstash is known to work with a syslog_logger as the underlying logger when its shipping to a sufficiently configured rsyslog.

In any case is probably much easier to setup Logstash directly on the application server to read the logs from the default log file location and to eventually forward them to their final destination.

Installation

Add this line to your application's Gemfile:

gem 'rackstash'

And then execute:

$ bundle

Or install it yourself as:

$ gem install rackstash

Usage

Rails 3, Rails 4

Just add Rackstash to your Gemfile as described above. Then, in the environment you want to enable Rackstash output, add a simple

# config/environments/production.rb
MyApp::Application.configure do
  config.rackstash.enabled = true
end

Additionally, you can configure Rackstash by setting one or more of the settings described in the configuration section below in the respective environment file.

Rails 2

When using bundler (if not, you really should start using it), you can just add Rackstash to your Gemfile as described above. Then, in the environment you want to enable Rackstash output, add a simple

require 'rackstash'
config.rackstash.enabled = true

If you use Bundler.require during your Rails initialization, you can skip the first line of the above step.

Note though that is is not sufficient to require Rackstash in an initializer (i.e. one of the files in config/initializers) as these files are evaluated too late during Rails initialization for Rackstash to take over all of the Rails logging. You have to require it in either config/environment.rb or one or more of config/environments/<environment name>.rb.

Additionally, you can configure Rackstash by setting one or more of the settings described in the configuration section in the respective environment file.

Configuration

You have to set the enabled attribute to true to convert the logs to JSON using Rackstash:

config.rackstash.enabled = true

Then you can configure a multitude of options and additional fields

# The source attribute of all Logstash events
# By default: "unknown"
config.rackstash.source = "http://rails.example.com"

# An array of strings with which all emited log events are tagged.
# By default empty.
config.rackstash.tags = ['ruby', 'rails2']

# Additional fields which are included into each log event that
# originates from a captured request.
# Can either be a Hash or an object which responds to to_proc which
# subsequently returns a Hash. If it is the latter, the proc will be exceuted
# similar to an after filter in every request of the controller and thus has
# access to the controller state after the request was handled.
config.rackstash.request_fields = lambda do |controller|
  {
    :host => request.host,
    :source_ip => request.remote_ip,
    :user_agent => request.user_agent
  }
end

# Additional fields that are to be included into every emitted log, both
# buffered and not. You can use this to add global state information to the
# log, e.g. from the current thread or from the current environment.
# Similar to the request_fields, this can be either a static Hash or an
# object which responds to to_proc and returns a Hash there.
#
# Note that the proc is not executed in a controller instance and thus doesn't
# directly have access to the controller state.
config.rackstash.fields = lambda do
  {
    :thread_id => Thread.current.object_id,
    :app_server => Socket.gethostname
  }
end

# Buffered logs events are emitted with this log level. If the logger is
# not buffering, it just passes the original log level through.
# Note that the underlying logger should log events with at least this log
# level
# By default: :info
config.rackstash.log_level = :info

Caveats

Few tests
No plain Rack support yet

Contributing

Fork it
Create your feature branch (git checkout -b my-new-feature)
Commit your changes (git commit -am 'Add some feature')
Push to the branch (git push origin my-new-feature)
Create new Pull Request