Why do you want this feature?
When I'm using kubectl, I don't see an obvious way to tell which EKS cluster I'm using.
It'd be nice to be able to easily switch between kubectl and eksctl.
What feature/behavior/change do you want?
For example, something like eksctl scale nodegroup [--use-kubeconfig-currnet-context|--use-kubeconfig-context=<context_name>] would be a good start.
And, in the opposite direction, we could probably do a few things, perhaps eksctl utils kubectl --name=<cluster_name> could be used with a temporary kubeconfig file where authenticator binary is not required, but that seems like a different use-case all together (we might consider adding eksctl utils write-kubeconfig --temp for that). Perhaps we could start with eksctl utils print-kubectl-command --name=<cluster_name> or something along those lines? Let's discuss!
@errordeveloper I would like tow ork on this issue, I will start by setting up my local environment today.
@cored thank you! Please join slack also, so we can help you with anything :)
Sure thing, I will probably start with this tomorrow.
When I'm using kubectl, I don't see an obvious way to tell which EKS cluster I'm using.
This part seems easy enough right now - kubectl --cluster=xxx.region.eksctl.io.
I came here to say that the cluster name added to kubeconfig is not the same as the name returned by eksctl get cluster - the former has region.eksctl.io added.
Hi @errordeveloper, I opened #1732 to work on this issue.
It adds the eksctl scale nodegroup [--use-kubeconfig-context=<context_name>] flag, which would read the cluster name from the context, the other info it could use from the context would be for IAM.
For the opposite direction going from using eksctl to kubectl, would eksctl utils write-kubeconfig --temp create a kubeconfig without the user authentication part, so that a user adds it later?
Most helpful comment
@errordeveloper I would like tow ork on this issue, I will start by setting up my local environment today.