busbud / devops-challenge-apps Goto Github PK

This is something that you should pose as a question in a personal interview. The engineer is likely going to want to ask a few probing questions in order to deliver the answer he feels is best, and this format does not allow the engineer to do that.

Problem

As it's currently designed, part 4 (monitoring), would be a more appropriate interview question. (similar reasons as #2 ).

I'd also like to add that as an engineer, I wouldn't want to answer these sorts of interview questions in a github repo, where other applicants following me would be able to see, learn, and improve upon my answers.

A selection is meant to be competitive, and I'd like to think that my own experiences have given me a competitive advantage. For any item you're asking the engineer to deliver in the current PR, ask yourself "is he giving away his competitive advantage"? If so, this should be an interview question, not part of a code-challenge.

Alternative

However, monitoring itself, is a great topic to put into a coding challenge. You might reword this part of the challenge to ask the engineer to create another application tier that handles monitoring of the other tiers.

It's clear you're looking to do a docker based infrastructure. If that's so, then for Part 2: for you to have all kinds of "extra" substance to do a canary release, then you're starting from the wrong place.

A good docker deployment will have the ability to deploy itself depending on the env it's deployed within, meaning you would have the same infrastructure for canary, and you'd just need to engineer some environment settings/configs allowing for discovery of the proper source build and deploy targets from your build pipeline.

It amounts to Step 2 is not really answerable, it's a trick question.

There isn't any reason to have multiple compose files. The whole point of the compose file is to describe multiple services, and doing so in a single file gives you the added benefit of using docker networking, orchestration, and security in your apps entire ecosystem.

I believe this is the biggest reason your #devopschallenge has received the feedback it has. It's not a matter of time, it's become a matter of now having to refactor code just to start the challenge.

If you're going to challenge an engineer to set up an infrastructure, application design is just as important as the build, delivery, and hosting models.