Questions about Cluster/Projets/Namespaces/Nodes organisation


I’ve used Rancher 1.6 with cattle orchestration for years with a lot of joy, so thanks for all your work!

But today I’m moving to Rancher 2 with kubenetes clusters hosted with on premise bare metal hardware.

I’ve successfully installed Rancher on a 3 nodes k8s with rke and helm, so now I can create my clusters.
…And here comes troubles in my head.

I’m working alone and I’m not very experienced with k8s, so I’m little afraid to make architecture mistakes.

If I understand, I need 3 nodes for masters (etcd, controlplane, kubelet, etc) and another three nodes for workers, which means 6 nodes for each clusters, am I right ?

I also have staging and production workloads, so I read in k8s documentations that is better to physically separate staging and production clusters, is it right ? 12 nodes in 2 clusters for staging + production ?

And I’ve different clients which means tenants. So I need 12 nodes by client ?

Or is it a good practice to have one cluster (3 masters, 3+ workers) for a client, and use “projects” object to separate, staging and production ? (and k8s namespaces to separateur apps stacks)

How do you usually organize that ? I’m a little lost…
Tell me if my questions aren’t very clear… I can try to reformulate.


Best regards,