This message was deleted.

Are you asking about cluster/node auto scaling, or about scaling of a workload? Both are possible, just want to give you an accurate answer.

Cluster/node autoscaling

Ok. To do that on vSphere, you can leverage the upstream K8s auto scaler. More info can be found here: https://github.com/rancher/barn/tree/main/Integrations/RKE2ClusterAutoscaler

Perfect, I will check it out. Thanks!

What we do is have an Nginx Plus LB instance in front of our K8s cluster upstreams. In each K8s cluster, we always have at least 2 workers (we have a split datacenter so one in each) that have a consistent FQDN in our DNS service Infoblox - and tell Nginx Plus those for routing. Once inside the cluster, the K8s ingress controller (Nginx open source) routes to the desired service/pods.

Super solid enterprise solution, @modern-television-79263!

We’ve been happy with it! We have a dedicated /24 in Infoblox that incorporates DDNS and DHCP for the K8s nodes.

Thanks for sharing @modern-television-79263. That's an impressive solution.

We do need to incorporate more use of health checks on upstream nodes on the Nginx Plus side of things. So little time, so much to do. ;)