kubeshop / monokle-admission-controller

Kubernetes Admission Controller for enforcing policies in your clusters
https://monokle.io
MIT License
9 stars 1 forks source link

Add deployment scripts #8

Closed f1ames closed 1 year ago

f1ames commented 1 year ago

This PR introduces helm chart for admission controller for easier and more flexible deployment.

It focuses on Helm chart only. I would like to cover release related pipeline in a follow-up PR (so building and publishing docker images, packaging and releasing Helm chart, etc).

Changes

Fixes

Testing

See CONTRIBUTING.md. This still can be only tested with locally build docker images and helm.

Checklist

f1ames commented 1 year ago

LGTM but I'd also like Stefan's eyes on the deployment part.

Yes, me too 👍 Waiting for Stefan input then before merging.

f1ames commented 1 year ago

@murarustefaan would you take a look?

f1ames commented 1 year ago

I'm wondering whether we should test agains multiple Kubernetes version. As this is not requesting a specific one, it will vary between minikube releases.

We can run the test in parallel for like the last 3-5 kubernetes version and see the results, as most Kubernetes add-ons in the CNCF space do.

@murarustefaan Could be useful 👍 Do you have any such projects at hand which does that so I can take a look? Do you suggest it could be like `[8 latest major K8s versions x few latest major helm version]? (like K8s 1.20.x - 1.28.x? not sure for Helm 🤔)

EDIT

Run an experiment with such matrix here - https://github.com/kubeshop/monokle-admission-controller/actions/runs/6574236171.

The number of combinations could get high quickly. So either we go with [k8s x helm] (as job above) or multiple k8s version + one helm version and then one k8s version + multiple helm versions. Not exactly sure how one effects the other 🤔

@WitoDelnat @murarustefaan any additional thoughts on this?

murarustefaan commented 1 year ago

I'd say 3 helm versions * last 3-4 kube versions works for now, we also don't want to use a lot of Github Actions time as that will cost us in the longer run.