Deployment

Overview

To safely shut down Sidekiq, you need to send it the TSTP signal as early as possible in your deploy process and the TERM signal as late as possible. TSTP tells Sidekiq to stop pulling new work and finish all current work. TERM tells Sidekiq to exit within N seconds, where N is set by the -t timeout option and defaults to 8. Using TSTP+TERM in your deploy process gives your jobs the maximum amount of time to finish before exiting.

If any jobs are still running when the timeout is up, Sidekiq will push those jobs back to Redis so they can be rerun later.

Network Architecture

I recommend running 1 or more Sidekiqs per app server in your production cluster. At a previous employer, I ran two Sidekiq processes on three app servers, for six total processes. With default concurrency of 25, this gave us 150 worker threads. All six processes talked to the same master Redis server and we used high, default and low queues to keep queue management and job priority as simple as possible.

Running separate worker machines for Sidekiq or using many different queues adds complexity where it was not needed for us.

Heroku

Using sidekiq with Heroku is simple, add a Procfile to your Rails app to start a sidekiq worker process:

web: bundle exec puma ...
worker: bundle exec sidekiq -t 25 ...

To connect to your Redis addon, you'll need to tell Sidekiq which environment variable name to use by setting REDIS_PROVIDER to the name of that variable, e.g. heroku config:set REDIS_PROVIDER=REDISTOGO_URL. Make sure you're using discrete Redis instances for any other purposes such as a Rails cache store.

Keep in mind that Heroku puts a hard limit of 30 seconds on a process restart, the -t 25 tells Sidekiq to give the jobs 25 seconds to finish before starting the "force shutdown" procedure. You cannot send the TSTP signal but you can optionally use Sidekiq::ProcessSet.new.each(&:quiet!) in a pre-deploy hook to give your jobs even more time to finish. After that 25 second limit, Sidekiq will push any unfinished jobs back to Redis; make sure your jobs are idempotent so it can restart them when the process starts back up.

See this page for more details about Procfiles, Foreman and Heroku.

Setting this environment variable will greatly reduce Sidekiq's memory usage and is highly recommended:

heroku config:set MALLOC_ARENA_MAX=2

Running your own Process

If you want to run Sidekiq on your own server, use Upstart or Systemd to start Sidekiq as a system service. This will ensure the process is restarted if Sidekiq crashes.

Here's an example Upstart config used to manage Sidekiq instances. These files go in /etc/init and will automatically startup Sidekiq when the machine is booted. The Sidekiq process group can be very simply managed with [start | stop | restart] workers.

Here's an example systemd unit file. The systemd documentation has sections on the .service file and executing processes. Every developer who uses Linux should read these pages many times.

Sidekiq Enterprise's sidekiqswarm binary makes it trivial to start N Sidekiq processes with a single Upstart/Systemd service. Take a look at Foreman for a useful tool to automate your application processes in development and production.

Capistrano

Use the capistrano-sidekiq gem (github). Integrated support has been removed. Warning: Capistrano uses daemonization by default so if the Sidekiq process crashes, it will not restart automatically.

Events

Sidekiq fires process lifecycle events when starting up and shutting down:

Sidekiq.configure_server do |config|
  # runs after your app has finished initializing but before any jobs are dispatched.
  config.on(:startup) do
    make_some_singleton
  end
  config.on(:quiet) do
    puts "Got TSTP, stopping further job processing..."
  end
  config.on(:shutdown) do
    puts "Got TERM, shutting down process..."
    stop_the_world
  end
end

This can be extremely useful if you want to start/stop your own threads/actors or signal.

Sidekiq Pro users can use :quiet to signal long-running jobs that they need to prepare for an upcoming restart (this requires super_fetch to be enabled).

# config/initializers/sidekiq.rb
$shutdown_pending = false
Sidekiq.configure_server do |config|
  config.on(:quiet) do
    $shutdown_pending = true
  end
end

# some_job.rb
def perform
  # This job might process 1000s of items and take an hour.
  # Have each iteration check for shutdown. big_list_of_items
  # should only return unprocessed items so the loop will continue
  # where it left off.
  big_list_of_items.find_each do |item|
    process(item)
    # Sidekiq Pro will restart job immediately on process restart
    raise Sidekiq::Shutdown if $shutdown_pending
  end
end

Previous: Delayed Extensions Next: Monitoring