search

giantswarm / roadmap

Giant Swarm Product Roadmap
https://github.com/orgs/giantswarm/projects/273
Apache License 2.0
3 stars 0 forks source link

KaaS Staged Deployments #620

Open alex-dabija opened 2 years ago

alex-dabija commented 2 years ago

User Stories

- As a developer, I want to have a staged approach to deployments in order simplify development and testing of new features.

- As a developer, I want to use the same tools in development and testing for deployments that I use for rolling out features to production in order to not use error prone hacks and workarounds.

- As a customer, I want to be sure that during a freeze period no deployments are rolled out to specific management clusters.

Details, Background

Currently, KaaS releases are made available to all installations as soon as the PR is merged. Testing them requires developers to manually make them available on the test management cluster(s). This situation has led to ArgoCD overwriting the manual changes. Also, it's a manual procedure which I think should be avoided.

Going forward, CAPI controllers are going to be continuously deployed to management clusters. Using the same tools in development and testing should ease the developer experience,lead to fewer errors and allow us to gain experience with this practice. The goal is to figure out as soon as possible if this is a good idea or not.

Some of our customers have deployment freezes during certain periods of time throughout the year and they like us to guarantee that no unintended changes or deployments make it to their environments. It's true we have some freedom now with deployments to management clusters but this is going to change when CAPI controllers are deployed continuously.

Blocked by / depends on

None

weatherhog commented 2 months ago

@alex-dabija this one is from 2021. Can you please check if this request is still valid?

alex-dabija commented 2 months ago

@alex-dabija this one is from 2021. Can you please check if this request is still valid?

Yes, the request is still valid and it's coming from KaaS in order to have an easier time managing our clusters (management and workload).

Recently, we wanted to focus on this for management clusters with Bigmac. Those plans need to be reevaluated.