Retry

The Retry module provides a wrapper around Ruby's own retry keyword to reduce the amount of boilerplate required to rerun a block of code if an exception occurs.

Without Retry, attempting a block thrice might look like this:

def fetch_data
 attempts = 0

 begin
   attempts += 1
   Net::HTTP.get('example.com', '/index.html')
 rescue
   retry if attempts < 3
 end
end

fetch_data

The simplest usage of Retry reduces the code to this:

Retry.attempt { Net::HTTP.get('example.com', '/index.html') }

Usage as a Module Function

Retry can be used directly as a module function. Its arguments are:

1. A list of exceptions to recover from. If none are provided [StandardError] is the default list.

2. The number of times to reattempt the block. The default is 2. (If the number of retry attempts is 2, the block is attempted thrice: initially once and then up to two more times if the block causes an exception.) Use the keyword reattempts to specify the number of reattempts.

3. An optional time in seconds to delay between attempts. The default is 0, meaning no delay occurs after an attempt. Use the keyword wait to specify the delay.

The following shows uses all the arguments to quickly customize retry behavior.

##
# Recover from Net::HTTP::ConnectionError or SocketError exceptions. Retry twice.
##
Retry.attempt(Net::HTTP::ConnectionError, SocketError, delay: 1, reattempts: 4) do
  Net::HTTP.get('example.com', '/index.html')
end

The example makes 4 reattempts if either the Net::HTTP::ConnectionError or SocketError exception occurs during the GET request. A delay of 1 second occurs between each attempt.

If an exception occurs that is not in the list of named exceptions, the code exception is raised per normal.

Usage as a Concern

Since Retry is a module, it be readily mixed into any Ruby class and called as an instance method.

class WebPage
  include Retryable

  def initialize(url)
    @url = url
  end

  def fetch
    attempt(Net::HTTP::ConnectionError, SocketError, delay: 1, reattempts: 4) do
      Net::HTTP.get(@url, '/index.html')
    end
  end
end

Exceptions

If all attempts fail, attempt raises Retry::RetryExhaustedError.

An instance of Retry::RetryExhaustedError contais the history of all exceptions encountered during all attempts via the exception_history method.

begin
   attempt(ZeroDivisionError, reattempts: 2) do
     10/0
   end
 rescue Retryable::RetryExhaustedError => e
   # e.exception_history # => [ZeroDivisionError, ZeroDivisionError, ZeroDivisionError]
end