Transparency Regarding Cluster State

cornelius-keller commented 4 years ago

User story

These are exmples:

As a cluster-admin I like to see how many nodes of my new cluster are already available.
As a cluster-admin I like to know how many nodes are actually running within a cluster, what dimensions (RAM, CPU) and what versions they have.

Rationale

We need to distinguish between desired and actual state. This is especially important for:

cluster creation
cluster scaling
upgrades
cluster deletion

Background

At the moment our UIs reflect a desired state for a cluster. For a better understhttps://github.com/giantswarm/roadmap/issues/57anding of the cluster we would also need to show our users the current state. Knowing the state would also enable us to add some more features.

Open questions

How deep do we wanna go regarding the state of the cluster? Where do we draw the line between actual state within our API/UIs as opposed to prometheus monitoring and Grafana.

Possible side effects

Dependencies/prerequisites

We display cluster state in our Backstage UI (e. g. https://devportal.giantswarm.io/clusters) and both Ready and Deleting definitely appear there already
Not sure whether we also show anything regarding upgrading or creating.
I don't think we should show anything regarding scaling in that context, as it's just a normal thing for a K8s cluster, separate from Readiness

puja108 commented 3 months ago

I agree that scaling is not needed. I'd say we should check if other CAPI states like creating or upgrading (is that even a CAPI state?) also show up and if we have that we can close this (we should also document where people can see this, also in the CRs, but that might rather be a topic for @giantswarm/team-turtles).

giantswarm / roadmap