Retry cycle is performed one attempt less than expected

[ironaimer]

Environment

• Elixir & Erlang versions (elixir --version): Erlang/OTP 23, Interactive Elixir (1.10.4)

• ExAws version mix deps |grep ex_aws2.0.2

• HTTP client version. IE for hackney do mix deps | grep hackney 1.16

Current behavior

In case of SQS unavailability retry cycle is executed one iteration less than expected.

It seems strict ">" should be used in attempt_again? in request.ex as "attempt" starts from 1.

  def attempt_again?(attempt, reason, config) do
    if attempt >= config[:retries][:max_attempts] do

So retry cycle is executed (max_attempts - 1) times.

Expected behavior

Retry cycle should be executed max_attempts times.

[Bernard Duggan]

I may be wrong, but that's not how I read it:

• attempt is initially set to 1 (request.ex:20).
• An attempt is made.
• If it fails, attempt_again? is called. Now in this circumstance:
  • If max_attempts is 1, the >= condition will match and we'll fail.
  • If max_attempts is 2, the >= condition will not match, and we'll go to the else block
• At this point, we add 1 to the attempt count, and loop back and try again. So we'll get 2 attempts, as per the config option.