Please use AWS Ruby SDK instead of the AWS CLI

[dhs-rec]

I wonder about the usage of the AWS CLI here. This is all Ruby code, so why not use the Ruby SDK instead of pulling in a Python application? This would also handle the retries in case of rate limits (see commit 57b3cd6) automatically and with exponential backoff, if the client was initialized accordingly, for example

ec2       = Aws::EC2::Client.new(
  credentials: Aws::SharedCredentials.new(profile_name: some_profile),
  region:      region,
  retry_limit: 20
)

I didn't run into API rate limits anymore since I added the retry_limit: 20 part some years ago...

[dhs-rec]

Here's a complete Ruby-only example for getting the tags of the instance the script is running on:

#!/usr/bin/ruby

require 'net/http'
require 'aws-sdk-ec2'

begin
  # No need for URI here
  http = Net::HTTP.new('169.254.169.254', 80)

  # Get instance ID
  instance_id = http.get('/latest/meta-data/instance-id').body

  # Get region (via availability zone)
  region = http.get('/latest/meta-data/placement/availability-zone').body[0..-2]
rescue
  puts 'Not an AWS instance'
end

# Initialize EC2 client (uses exponential backoff by default for retries)
ec2 = Aws::EC2::Client.new(
  region: region,
  retry_limit: 20,
)

# Get the tags
ec2.describe_tags({filters: [{ name: 'resource-id', values: [instance_id]}]}).tags.each do |tag|
  puts "\t" + tag[:key] + ' = ' + tag[:value]
end

[jimsheldon]

We have a desire to use this module, so we did some work on this. I have opened #28 which adds this feature.

[dhs-rec]

BTW, two things can be changed in the example code above:

1. The region can now be obtained directly (via .../placement/region)
2. While still available, exponential back-off might not be the default retry strategy anymore (see SDK documentation for details).