Skip to content
Notes

IDEN (current language)

Retry waits in retryable-async do not block other Async tasks

2 min readRisqi Romadhoni
  • Ruby
  • Async

I wrote retryable-async, a small Ruby gem that retries a block of code when it fails. You can use it inside an Async task.

What the gem adds

By default, Retryable::Async.run tries the block up to three times. The wait between attempts grows linearly: 1 second, then 2 seconds. If the third attempt also fails, run raises the error again.

You can change this behavior with these options:

  • max_attempts sets the number of attempts.
  • on selects the exceptions to retry. By default, it covers every StandardError.
  • base_delay and backoff set the delay. backoff is :linear or :exponential.
  • jitter multiplies the delay by a random factor between 0.5 and 1.5.
  • before_retry receives the number of the failed attempt and its exception, before each retry.

Example from the README

The gem does not depend on async, so you require async yourself.

require 'async'
require 'retryable-async'

Async do
  Retryable::Async.run do
    puts "Doing something asynchronously..."
    raise "fail"
  end
end

The block in this example always fails, so there are two waits: 1 second before the second attempt and 2 seconds before the third. After that, Retryable::Async.run raises the error again.

The retry wait does not block other tasks

This section applies to async 2.46.0, which requires Ruby 3.3 or later.

Async runs each task on a fiber. Each retry wait calls a plain sleep. Inside an Async task, Ruby passes that sleep to the Async scheduler.

The scheduler only pauses the fiber that calls sleep. Other Async tasks can keep running during the retry wait.

Spotted a mistake? Let me know