CI/CD pipeline creates Helm chart on the fly - anyone done this?
Hi DevOpsers,
looking for a little bit of feedback on this idea:
let's say we have a bunch of containerized apps. Every repo already has a docker-compose.yml file which allows developers to quickly spin up the app with all its dependencies locally.
As a next step, we want to add automation to our CI/CD pipeline which spins up a test instance of an app on a Kubernetes cluster whenever a new PR is created ("PR Environments"). This will be useful for a range of things, such as automated testing, user acceptance testing, customer demos etc.
My thinking here is that the docker-compose.yml files already contain most of what we need to achieve this (service definitions etc) and I'd like to avoid duplication. I.e. I wouldn't wanna maintain a separate set of K8S config files or Helm charts or similar.
The approach I'm trying out right now is to set up a CI/CD step which creates a Helm chart on the fly using kompose.io, using alternative conversion. I.e. the Helm charts never get checked into source control, they always get created as part of the CI/CD build automation. Whatever additional config is needed on top of Docker Compose gets injected via --set, e.g.
helm install --wait --set service.type=LoadBalancer Any feedback on this approach, is this sensible? Anyone done something similar?
[link] [comments]
