SimplerWorkflow

A wrapper around Amazon's Simple Workflow Service meant to simplify declaring and using activities and workflow. Provides some sane defaults and work around some idiosyncracies of the platform.

Installation

Add this line to your application's Gemfile:

gem 'simpler_workflow'

And then execute:

$ bundle

Or install it yourself as:

$ gem install simpler_workflow

Usage

Configuring AWS Gem

We are using the aws-sdk gem to communicate with Amazon's service. You can configure the service by putting an aws.yml file in your config directory. The file should contain the following information:

development:
  access_key_id: <Amazon Acess Key ID>
  secret_access_key: <Amazon's secret access key>

This will authenticate your application or script against AWS and give you access to your SWF domain, workflows and activity.

Access a domain

You will need to get a handle on a SWF domain before you can do anything else. Domains are accessed using the SimplerWorkflow::Domain class. This declares a domain that does not retain workflow execution data:

domain = SimplerWorkflow::Domain["my-domain"]

An other option is to use the domains method to get a handle on a domain and pass a block. This allows you to write the following code:

domain = SimplerWorkflow::Domain.domains("my_domain") do
  # Register activities
  register_activity(:an_activity, '1.0.0') do
    # See details below...
  end
  # Register workflow(s)
  register_workflow(:a_workflow, '1.1.0') do
    # See details below...
  end
end

You can also get a handle on a domain that retains information about workflow execution for 10 days with the following code:

domain = SimplerWorkflow::Domain.new("my-domain", 10)

Domains are scoped by AWS accounts. The name of the domain must be unique within the account. You do not need to create the domain on AWS since it is created the first time it is accessed.

Creating an activity

Activities perform the work attached to the workflow and report back to SWF when the activity completes or it fails.

SimplerWorkflow makes it easier to register an activity with your domain.

Activities must provide the following:

A name
A version
Some code to run when it is invoked

You can also optionaly declare when what to do when the activity fails or succeeds.

my_activity = domain.register_activity :my_activity, "1.0.0" do
  perform_activity do |task|
    input = task.input
    puts task
  end
end

my_activity.start_activity_loop

The activity manages a loop that waits for messages from SWF. It runs under it's own sub-process to make it easier to manage and isolate each of the activities.

Activities are passed a task parameter. This parameter is provided to the activity by SWF and provides a lot of information about the task at end. One item passed is the input attribute.

The block attached to the perform_activity method is called when the activity is invoked. This block contains the actions that an activity will perform. The SimplerWorkflow::Activity class will automatically report that the activity completed successfully when the block returns unless a response has been provided in the block. It will automatically report that an activity failed when an unhandled exception is thrown within the block.

The activity can influence what happens when the activity succeeds or fail. You can specify the activity's failure response through the SimplerWorkflow::Activity#on_fail method. By default, the activity will ask the workflow to abort itself on failure. You can also ask the workflow to repeat the activity by passing :retry to the method:

my_activity = domain.register_activity :my_activity, "1.0.0" do
  on_fail :retry

  perform_activity do |task|
    # ...
  end
end

The activity can also tell a workflow what activity to trigger next on the workflow. This only works when using the default decision loop (described later). This is done by declaring what is the next activity should be:

my_activity = domain.register_activity :my_activity, "1.0.0" do
  on_success :my_next_activity, "1.0.0"

  perform_activity do |task|
    # ...
  end
end

Workflow and Decision Loops

The next key concept in SimplerWorkflow is the workflow. The workflow decides what activities to invoke, what to do when they complete and what to do when they fail. The SimplerWorkflow::Workflow object manages the decision loop.

By default, the workflow is setup to allow for a linear set of activities until the list runs out. This is convenient for simple workflows. There are also hooks to override what happens with each decision point to accomodate more complex workflows.

Workflows are declared and registered through the SimplerWorkflow::Domain#register_workflow method. This will register a workflow and configure it to start a linear workflow with version 1.0.0 of the :my_activity activity:

my_workflow = domain.register_workflow :my_workflow, '1.0.0' do
  initial_activity :my_activity, '1.0.0'
end

The next step is to start the decision loop:

my_workflow.decision_loop

The decision loop will be launched in it's own sub-process and will proceed to poll SWF for decision tasks.

Customizing the workflow

There are hooks for different section of the decision loop. You can specify what happens when the workflow is started with the on_start_execution method:

my_workflow = domain.register_workflow :my_workflow, '1.0.0' do
  on_start_execution do |task, event|
    puts "Mary had a little lamb"
    task.schedule_activity_task my_activity.to_activity_type, :input => { :my_param => 'value'}
  end
end

The task and event parameters are received from SWF. Unfortunately, you must still work within the constraints of the AWS SWF SDK. The SimplerWorkflow::Activity#to_activity_type generates the proper incantation used by SWF to identify and locate an activity.

You can also define similar hooks for events using the following methods:

on_activity_completed is called when an activity completes and SWF reports back to the decision loop.
on_activity_failed is called when an activity reports a failure to SWF.

Triggering Workflows

You can trigger the worklow with the following command:

SimplerWorkflow::Domain["my-test-domain"].start_workflow("hello-world", "1.0.1", "AWS")

The parameters are the name of the workflow, the version as well as a string representing the input to the workflow.

Workflow and Activity Versioning

All workflow types and activity types are versioned under SWF. This allows an organization to release new versions of a workflow or activity as they are updated and allow workflows to complete under the old version number if need be.

Here are a few recommendations on when to change the version of a workflow or activity when using SimplerWorkflow:

You need to bump the version when you are changing the defaults associated with an activity or workflow. These are set using the default_ methods. The defaults are commnicated to AWS when the workflow or activity is registered. It does not get updated when they are changed within the code.
You may want to bump a version if you have work in progress under an existing workflow and you need to introduce changes for new work. You will need to keep the older activity and or workflow around while it completes.
You do not need to bump the version when you change the work performed by the activity or the decision loop itself. This is code that is directly managed by SimplerWorkflow and isn't communicated to AWS. This only works if you do not want previous workflows to finish using the previous version of the code though.

Running Workflows as Daemons

Often you'll need to expose several SWF components and make sure they run at all time on a server as daemons. For this, you may use ParentProcess in companion with a demonizer like the daemons gem. In the parent process, you define what children it must monitor and how many versions of each children must run.

Here's a sample use of parent process.

class SimpleWorkflowAPI
  
  extend SimplerWorkflow::ParentProcess
 
  log_level Logger::DEBUG
 
  workers 4
 
  domain = SimplerWorkflow::Domain.domains "my_ops"
 
  # define your activities
  test_activity = domain.register_activity :test, "1.0.0" do
    perform_activity do |task|
      task.complete! :result => "approved"
    end
  end
 
  # let the parent process start the loop
  # this boot will be run as many times as workers you indicate
  # e.g. here you'll have 4 instances of the test_activity running in parallel.
  on_boot do
    test_activity.start_activity_loop
    # more activities initializations go here
  end
end

This is only the class starting the processes, if you need to daemonize this an expose as a unix service, for example in /etc/init.d/simpler_workflow, you can create a daemon script and link it to the services folder and then configure it to start on boot.

Assuming the last example is called boot.rb this can be a sample daemon.rb:

#!/usr/bin/env/ ruby
require 'rubygems'
require 'bundler/setup' #load all gems in your Gemfile
require 'daemons' # require the daemons gem

dirmode = :normal
# Check if we're in a deployment machine
# to store logs in /var/run or in the app folder
if ENV['ENV']
        log_dir = "/var/log/swf"
        pid_dir = "/var/run/swf"
else
        log_dir = File.expand_path '../log/', __FILE__
        pid_dir = File.expand_path '../log/pid', __FILE__
end
script_path = File.expand_path '../boot.rb', __FILE__
Daemons.run script_path, {
        :app_name   => "my_daemon",
        :dir_mode   => dirmode,
        :log_dir        => log_dir,
        :dir        => pid_dir,
        :multiple   => false,
        :monitor    => true,
        :log_output => true,
        # backtrace causes errors detecting as uncatched
        # some correctly handled exceptions. Keep disabled.
        :backtrace  => false
  }

For more details on daemons you can look in http://daemons.rubyforge.org/Daemons.html You can still use any other daemonization script, or even handle it yourself as taught in http://www.jstorimer.com/products/working-with-unix-processes

There is a new Rake task called simpler_workflow:work that will look for workflows located under the lib/workflow directory. This makes it much easier to put together a few workflow and manage them.

Another addition in 0.2.0 is the swf command. This command provides a script that starts and stops the workflows and provide other monitoring tools.

Reporting Errors

We now log exceptions and errors that occur while running activities or handling decision tasks. The default strategy is to log the errors, but it is easy to plug in a custom strategy. This is how you can plug in a strategy in a Rails initializer:

module SimplerWorkflow
  unless Rails.env.development?
    exception_reporter do |e, context|
      Exceptional.context(context)
      Exceptional.handle(e)
    end
  end
end

The block passed to the exception_reporter method will receive the exception and the context that has been computed from the activity or decition task.

This has been extremely helpful in tracking down and fixing errors within a workflow that was failing quietly.

Contributing

We welcome all kinds of contributions. This include code, fixes, issues, documentation, tests... Here's how you can contribute:

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

License missing from gemspec

RubyGems.org doesn't report a license for your gem. This is because it is not specified in the gemspec of your last release.

via e.g.

  spec.license = 'MIT'
  # or
  spec.licenses = ['MIT', 'GPL-2']

Including a license in your gemspec is an easy way for rubygems.org and other tools to check how your gem is licensed. As you can imagine, scanning your repository for a LICENSE file or parsing the README, and then attempting to identify the license or licenses is much more difficult and more error prone. So, even for projects that already specify a license, including a license in your gemspec is a good practice. See, for example, how rubygems.org uses the gemspec to display the rails gem license.

There is even a License Finder gem to help companies/individuals ensure all gems they use meet their licensing needs. This tool depends on license information being available in the gemspec. This is an important enough issue that even Bundler now generates gems with a default 'MIT' license.

I hope you'll consider specifying a license in your gemspec. If not, please just close the issue with a nice message. In either case, I'll follow up. Thanks for your time!

Appendix:

If you need help choosing a license (sorry, I haven't checked your readme or looked for a license file), GitHub has created a license picker tool. Code without a license specified defaults to 'All rights reserved'-- denying others all rights to use of the code.
Here's a list of the license names I've found and their frequencies

p.s. In case you're wondering how I found you and why I made this issue, it's because I'm collecting stats on gems (I was originally looking for download data) and decided to collect license metadata,too, and make issues for gemspecs not specifying a license as a public service :). See the previous link or my blog post about this project for more information.

fredjean / simpler_workflow Goto Github PK