How to experiment with CD in a GitFlow environment

May 9, 2022
Can we experiment with continuous deployment without completely migrating away from GitFlow first? Here are a couple approaches.

A reader recently asked me if it would be possible to experiment with Continuous Deployment on a project that’s effectively using GitFlow (That is: two long-lived branches, typically called main and develop, wherein after some verification, code in develop is later merged to main for deployment). I really love this question, because as you may know, I’m a big advocate of experimentation.

So how can we, or even can we experiment with continuous deployment for one team, or a subset of stories, without completely migrating away from GitFlow first?

Here’s my two-part answer:

  1. You could write an automatic deployment script that only applies to certain pull requests. Maybe it looks for branch names that begin with cd-, or some special token in the PR description such as #CD.

  2. However, I wouldn’t actually suggest doing this. Instead, I would set up automated deployments to run on merge. Which could be thought of as the deployment half of CD, or what I call “Lean CD”. Then it’s up to you to decide when (and if) to merge directly into main, bypassing the develop branch. For some tasks, you could do trunk-based development, and merge frequently (possibly using feature flags). And for other tasks, you could take whatever traditional approach you use (possibly involving manual QA, etc).

If you’re on a team using GitFlow, or a similar branching strategy, and you or someone else on your team is unsure about the impact of implementing Continuous Deployment, this could be a great strategy to dip your toes in the water before making the plunge.

And in the worst case, if you decide not to go forward with all-out Continuous Deployment, you’ll at least have a faster and more reliable automated deployment in place at the end!

Related Content

A GitFlow story

A team I recently started working with has been using GitFlow. Until last week.

We can't afford automation right now

Avoid a big up-front investment in automation by building it piece by piece, as needed.

Adventures In DevOps 093: Deconstructing the Minimum Viable Continuous Deployment

The panel jumps in and discusses the Minimum Viable Continuous Deployment from the manifesto on the web and discuss where they agree and disagree with the requirements listed there and clarify some of the ideas in the document.