• Stars
    star
    153
  • Rank 243,368 (Top 5 %)
  • Language
    Ruby
  • License
    MIT License
  • Created over 12 years ago
  • Updated about 3 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Extension to Sidekiq to pass job execution metadata such as status and result back to the client

SidekiqStatus

Build Status Dependency Status Test coverage

Sidekiq extension to track job execution statuses and returning job results back to the client in a convenient manner

Installation

Add this line to your application's Gemfile:

gem 'sidekiq_status'

And then execute:

$ bundle

Usage

Basic

Create a status-friendly worker by include SidekiqStatus::Worker module having #perform method with Sidekiq worker-compatible signature:

class MyWorker
   include SidekiqStatus::Worker

   def perform(arg1, arg2)
      # do something
   end
end

Now you can enqueue some jobs for this worker

jid = MyWorker.perform_async('val_for_arg1', 'val_for_arg2')

If a job is rejected by some Client middleware, #perform_async returns false (as it does with ordinary Sidekiq worker).

Now, you can easily track the status of the job execution:

status_container = SidekiqStatus::Container.load(jid)
status_container.status # => 'waiting'

When a jobs is scheduled its status is waiting. As soon sidekiq worker begins job execution its status is changed to working. If the job successfully finishes (i.e. doesn't raise an unhandled exception) its status is complete. Otherwise its status is failed.

Communication from Worker to Client

SidekiqStatus::Container has some attributes and SidekiqStatus::Worker module extends your Worker class with a few methods which allow Worker to leave some info for the subsequent fetch by a Client. For example you can notify client of the worker progress via at and total= methods

class MyWorker
   include SidekiqStatus::Worker

   def perform(arg1, arg2)
      objects = Array.new(200) { 'some_object_to_process' }
      self.total = objects.count
      objects.each_with_index do |object, index|
        at(index, "Processing object #{object}")
        object.process!
      end
   end
end

Lets presume a client refreshes container at the middle of job execution (when it's processing the object number 50):

container = SidekiqStatus::Container.load(jid) # or container.reload

container.status       # => 'working'
container.at           # => 50
container.total        # => 200
container.pct_complete # => 25
container.message      # => 'Processing object #{50}'

Also, a job can leave for the client any custom payload. The only requirement is json-serializeability

class MyWorker
   include SidekiqStatus::Worker

   def perform(arg1, arg2)
      objects = Array.new(5) { |i| i }
      self.total = objects.count
      result = objects.inject([]) do |accum, object|
        accum << "result #{object}"
        accum
      end

      self.payload= result
   end
end

Then a client can fetch the result payload

container = SidekiqStatus::Container.load(jid)
container.status # => 'complete'
container.payload # => ["result 0", "result 1", "result 2", "result 3", "result 4"]

SidekiqStatus stores all container attributes in a separate redis key until it's explicitly deleted via container.delete method or until redis key expires (see SidekiqStatus::Container.ttl class_attribute).

Job kill

Any job which is waiting or working can be killed. A working job is killed at the moment of container access.

container = SidekiqStatus::Container.load(jid)
container.status # => 'working'
container.killable? # => true
container.should_kill # => false

container.request_kill

container.status # => 'working'
container.killable? # => false
container.should_kill # => true

sleep(1)

container.reload
container.status # => 'killed'

Sidekiq web integration

SidekiqStatus also provides an extension to Sidekiq web interface with /statuses page where you can track and kill jobs and clean status containers.

  1. Setup Sidekiq web interface according to Sidekiq documentation
  2. Add "require 'sidekiq_status/web'" beneath "require 'sidekiq/web'"

Changelog

1.2.0

  • Support for sidekiq 4.2, integration with the new non-sinatra-based Sidekiq::Web (ncuesta)

1.1.0

  • Support for sidekiq 4.1, 4.0, 3.5, 3.4
  • No more replacement of original job arguments with generated unique jid. This is not needed anymore as Sidekiq started to do it. This change should make integration with other middlewares easier.
  • Dropped support for sidekiq versions older than 3.3
  • Dropped support for ruby 1.9.x, 2.0.x
  • Experimental support for Rubinius

1.0.7

  • Sidekiq 2.16 support

1.0.6

  • Sidekiq 2.15 support

1.0.5

  • Sidekiq 2.14 support
  • Do not create (and display in sidekiq_status/web) status containers for the jobs scheduled to run in the future by the means of perform_at/perform_in (mhfs)
  • Sidekiq web templates converted from .slim to .erb
  • Allow specifying worker name as a String (gumayunov)
  • Added ruby 2.0 to travis build matrix
  • Don't be too smart in extending Sinatra template search path (springbok)
  • Show worker names and adjust sidekiq-web template tags to conform to Sidekiq conventions (mhfs)

1.0.4

  • Sidekiq 2.10 and 2.11 support

1.0.3

  • Include SidekiqStatus::Web app into Sidekiq::Web app unobtrusively (pdf)
  • sidekiq 2.8.0 and 2.9.0 support

1.0.2

  • sidekiq 2.7.0 support
  • sidekiq integration tests
  • Display progress bar and last message in sidekiq-web tab (leandrocg)

1.0.1

  • sidekiq 2.6.x support

1.0.0

  • First release

Roadmap

  • Add some sidekiq-web specs
  • Support running inline with sidekiq/testing

Contributing

  1. Fork it
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Don't forget to write specs. Make sure rake spec passes
  4. Commit your changes (git commit -am 'Added some feature')
  5. Push to the branch (git push origin my-new-feature)
  6. Create new Pull Request

Copyright

SidekiqStatus © 2012-2013 by Artem Ignatyev. SidekiqStatus is licensed under the MIT license