Console glows blue in a dim apartment. Another Rails deploy gone sideways—jobs stacking like unpaid bills.
Configuring AWS SQS in Ruby on Rails? It’s not brain surgery. But don’t kid yourself: if Redis and Sidekiq treat you fine, why chase Amazon’s queue? Still, for those multi-app setups or pure cloud plays, here’s the drill. Skeptical as ever.
First off, producer mode. Dead simple. Toss in the gem.
# Gemfile
gem 'aws-sdk-sqs'
bundle install. Set your envs: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION=us-east-1, AWS_SQS_QUEUE_NAME, AWS_QUEUE_URL. Boom. Docs warn of variations—check ‘em, or regret it.
result = Aws::SQS::Client.new.send_message(
queue_url: ENV['AWS_QUEUE_URL'],
message_body: { order_id: "123" }.to_json
)
That’s it. Message flies. No fanfare. But workers? Ah, there’s the rub.
Why Shoryuken? Sidekiq’s AWS Cousin
Shoryuken apes Sidekiq. Redis? Nah, SQS. Add the gem.
# Gemfile
gem 'shoryuken'
Config file next: config/shoryuken.yml. Concurrency: 5. Require your workers. Queues list.
concurrency: 5 require: ./app/workers queues: - [ <%= ENV.fetch(‘AWS_SQS_QUEUE_NAME’) %>, 1]
Rails might snub your workers folder. Hack it: eager_load_paths tweaks. Delete the default, unshift app/. Rails versions meddle—autoload quirks for non-standard dirs like jobs or services.
Worker class. Include Shoryuken::Worker. Options: queue from ENV, auto_delete: true.
class ExampleWorker
include Shoryuken::Worker
shoryuken_options queue: ENV.fetch('AWS_SQS_QUEUE_NAME'), auto_delete: true
def perform(sqs_msg, body)
payload = JSON.parse(body)
::ExampleService.new(payload).call
end
end
Fire it up: bundle exec shoryuken -R -C config/shoryuken.yml. -R loads Rails env. -C slurps config. Sans that, you’d CLI-spam queues. Lazy? Fine by me.
ActiveJob Magic: SQS or Bust?
Want Rails-native jobs? Flip the adapter.
# config/application.rb
config.active_job.queue_adapter = :shoryuken
In-memory? Gone. SQS rules. Gen a job: rails g job SayHello.
class SayHelloJob < ActiveJob::Base
queue_as ENV.fetch('AWS_SQS_QUEUE_NAME', :default)
def perform(name)
sleep(2)
puts "Hello #{name}"
end
end
Console test: SayHelloJob.perform_later("jackson"). Peek SQS—job payload’s a JSON beast:
{ “job_class”: “SayHelloJob”, “arguments”:[ “jackson” ], “enqueued_at”: “2026-04-03T12:57:41.629214813Z” }
perform_now? Sync trap. Stays local. Sneaky.
But here’s my hot take—the one the original skips: Shoryuken’s a 2015 relic. GitHub’s quiet. AWS pushes Lambda, EventBridge. Why glue Rails to SQS workers when serverless eats this alive? Historical parallel: remember Resque? Redis queues ruled pre-Sidekiq. SQS feels like that pivot—cloud or die. Prediction? Rails 8 bakes async queues sans gems. Shoryuken’s bridge fuel.
Corporate spin? AWS docs gloss lock-in. “Scalable!” Sure. But bills stack too. Free tier? Laughable for prod.
Pitfalls galore. Env typos? Silent fail. Region mismatch? Messages vanish. Shoryuken concurrency? Tune wrong, backlog city. And Rails eager_load? Dev-prod drift kills.
Scaling? SQS shines—millions messages, no sweat. But monitor CloudWatch. Dead-letter queues? Set ‘em, or retries loop forever.
Is AWS SQS in Rails Actually Better Than Sidekiq?
Short answer: rarely. Sidekiq’s dashboard? Chef’s kiss. Shoryuken’s CLI logs—yawn. Redis local? Lightning tests. SQS latency? Network tax.
Use cases scream yes: cross-account queues, hybrid apps, serverless frontends. Otherwise? Hype.
Tweak shoryuken.yml for priorities. Multiple queues? List ‘em. But concurrency per queue? Manual math.
Error handling. perform squawks? Message redelivers. Visibility timeout—default 30s. Bump for long jobs.
shoryuken_options queue: 'high-priority', auto_delete: true, visibility_timeout: 60
Prod tip: Dockerize workers. Separate container. Scale horizontally. Kubernetes? HorizontalPodAutoscaler on queue depth.
Why Does This Matter for Rails Developers?
Multi-region? SQS FIFO. Compliance? Audits baked. But vendor lock? Ironclad. Exit plan? Pain.
Benchmark it. Sidekiq: 10k jobs/min local. SQS: half, network blames. Cost? Pennies first, dollars later.
Original nods docs. Good. But misses: IAM roles over keys. Secrets Manager. Vault. Env files? Dev only.
Upgrade path. Rails 7? CableReady async. SQS fits, but Solid Queue (Postgres) nips heels—free, SQL.
Dry humor: Amazon’s queue feels like dating a control freak. Reliable. Expensive. Always watching.
Test it. LocalStack mocks SQS. bundle exec rails c, fire jobs. Verify payloads. Chaos: delete queue mid-test.
Security. Encrypt queues. KMS keys. Policies tight—least priv.
And workers folder? Generators ignore. mkdir app/workers. Autoload wins.
Worth it? If you’re all-in AWS, yes. Else, Redis laughs last.
🧬 Related Insights
- Read more: Solo Dev’s Epic: Full ERP from Scratch in Months
- Read more: Python Dev’s Laravel IAM Gambit: Fixing RBAC or Just Another Package?
Frequently Asked Questions
What does AWS SQS do in Ruby on Rails?
Handles job queues and messages scalably, via Shoryuken or ActiveJob—no Redis needed.
How to set up Shoryuken workers in Rails?
Gem install, yml config, include Worker module, run with -R -C flags. Tweak eager_load for Rails recog.
AWS SQS vs Sidekiq: which is better?
Sidekiq for speed/simplicity; SQS for massive scale/multi-app. Test your load.
