Docker in production

I’ve just finished a list at work. A list of questions that I personally want answered before I throw anything docker in production. That list consists of nearly 200 questions, ranging from “engineering grooming” to “networking technicalities”. Despite me already having some answers to maybe half of the questions, it is always nice to see on writing.

For my personal projects, I’ve pretty much settled on having nginx serve as transparent proxies, just sending traffic through to Rancher — which has docker containers running, including some load balancers. The guys behind Rancher has created a container that updates Route53 DNS automatically, which works remarkably clever. Doing things this way, means that SSL termination can be handled anywhere in the chain and opens up for a lot of flexibility for doing a lot of crazy experimentation.

For work use, I’m not entirely sure if I want to use Rancher for the current set up. Down the road, sure — but there’s still quite a lot of work, before engineers are ready to adopt it.

I should clean up that list and post it here, for future reference at least.