aws cli: retry on RequestLimitExceeded error

aws_limit_exceeded_retry.sh

#!/usr/bin/bash

function retry() {
  tries=0
  max_tries=3
  sleep_sec=1
  exit_code=256
  error=''

  until { error=$(${@} 2>&1 1>&${stdout}); } {stdout}>&1; do
    exit_code=$?

    tries=$(( ${tries} + 1 ))
    if [[ ${tries} -gt ${max_tries} ]]; then
      exit ${exit_code}
    fi

    if [[ ${exit_code} == 255 ]] && (echo "${error}" | grep -q 'RequestLimitExceeded'); then
      if [[ ${tries} != 1 ]]; then
        sleep ${sleep_sec}
      fi
      sleep_sec=$((${sleep_sec} * 2))
      echo "${error}" >&2
      echo 'Being throttled. Retrying..' >&2
    else
      echo "${error}" >&2
      exit ${exit_code}
    fi
  done
}

retry $@

./aws_limit_exceeded_retry.sh aws ec2 describe-instances

A server error (RequestLimitExceeded) occurred when calling the DescribeInstances operation: Request limit exceeded.
Being throttled. Retrying..

A server error (RequestLimitExceeded) occurred when calling the DescribeInstances operation: Request limit exceeded.
Being throttled. Retrying..

Nice, a couple of potential improvements though:

1. quote the ${@} so that arguments with spaces in (e.g. aws ec2 create-tags --tags Key=type,Value="my type" ... ) will work
2. when calling from a script, might not want to exit that script (i.e. let the caller deal with the aftermath of failure) - putting a subshell in here (i.e. wrap in parentheses) would help