Amazon EC2, Auto-Scaling, AMIs, is there an easier way?

Question

I have a production Rails app on my Amazon EC2 instance, and set up a load balancer and auto-scaling (I did this all manually, not using Elastic Beanstalk. I also used Dokku to make deployment easier.). I'm confused though. When new instances are needed and built with Auto-scaling, it appears as though I need to build an AMI, and then set the new instances (built via auto-scaling) to use this AMI.

But there must be an easier way. Am I missing something? With Heroku, you just add a dyno and BAM, you're done. Do I really need to copy my AMI and set it is as my auto-scaling image every time I deploy to production? Is there a command, workflow, or post-deployment script I should be using?

Answer 1

You need to specify a AMI which contains the basic set up of the your production node. The usually best practice is to config the auto scale with a user data script. So when the new AMI boots up during the auto scaling it reads the user data (cloud init/upstart). The user data script can pull the code from the git or what ever source control and run the necessary pre-deployment commands. You can't be copying new AMI config to your auto scaling config every time you deploy the code.

I hope it makes sense to have a master AMI and so some kind of automation like user data script which run specific commands as soon as it boots the production node and then attaches to ELB.

Answer 2

There is a blog post covering dokku-based deployments using Opsworks here . One of the other Dokku maintainers (Michael Hobbs) uses this setup in production.

Note: I am one of the Dokku Maintainers, josegonzalez .