Dockerizing Ruby apps and moving off Engine Yard
summary
Company moves ~7–8 services off Engine Yard onto AWS using Docker images in ECR, Terraform for infra-as-code, and auto scaling groups with CodeDeploy. Engine Yard pain points include no VPC peering, manual Elastic IP moves and reboot delays during Spectre; cost grows from $450 to ~ $1,000 as services multiply. Alpine images require LibreSSL fixes and the_rubyracer is replaced by NodeJS for JS tests. Secrets go into KMS + SSM parameters, monitoring uses Prometheus + cAdvisor, and Packer rotates AMIs weekly; Kubernetes is deferred.
AI-written summary · about
chapters (12)
- 0:00Introduction and setup
- 3:42Why move off Engine Yard
- 9:50Why Docker for Ruby apps
- 11:15Building the Docker image
- 15:14Toolbox pitfalls and lessons
- 17:26Running tests in containers
- 21:40Security and secrets management
- 27:38Deploying containers at scale
- 31:11Current infrastructure overview
- 34:44Future plans and monitoring
- 38:35Key takeaways and next steps
- 42:27Q&A and closing
related talks
43mServer Infrastructure for Rails in 2016
27mScaling Ruby Applications with Mesos and Docker
20mRails Conf 2013 Engine Yard Cloud by Edward Chiu & PJ Hagerty
29mPackaging and Shipping Rails Applications in Docker
21mMigrating Beanstalk Ruby/Js Applications to ECS Using Docker
04mA message from Engineyard
39mReal World Docker for the Rubyist
40mDeveloping Ruby Applications with Docker