Fork me on GitHub

@yogidevbear late reply, but we were trying to use AWS CodeDeploy to deploy Clojure code for a while and we basically hated it and ditched it. It seems to be kind of a weird forgotten stepchild in the AWS infrastructure, which for some reason has like 15 different solutions to deploy code, each with weird light variations.


What we do currently is build plain-vanilla uberjars from our CI env (CircleCI) and then load them into Elastic Beanstalk instances running the "Java SE" stack, which works fine and is not super complex. We were also experimenting with building Docker containers with the jars in them for a while, but getting compose to work locally wound up being too annoying for us because of ~@reasons.


I filed this ticket about CodeDeploy back in June 2016, it's indicative of the general care that AWS seems to have put into CodeDeploy:


Thanks for the feedback @timgilbert. Really appreciate it. I must admit that CodeDeploy does seem like it's not quite there in terms of doing what it says on the tin.


We've been reasonably happy with Elastic Beanstalk, it has its own little command-line tool eb which is handy for prototyping / debugging deployment stuff. There's still a large configuration/management overhead for it which might be easier to deal with via terraform or a similar project