Hi guys, Need some suggestion on following things. I'm new to rancher and need your help on installing latest rancher 1. I'm trying to install latest version of rancher in a multi-node k3s cluster. I was able to get the nodes running but when I'm trying to use kubectl, getting this error

couldn't get current server API group list: Get "LOAD_BALANCER_DNS:6443/api?timeout=32s": dial tcp INSTANCE_IP:6443: connectex: No connection could be made because the target machine actively refused it.

I'm sure all the necessary ports are opened. 2. Is it necessary to set up kubectl at the local? Can I directly go ahead and install rancher on the nodes? 3. Also I'm not fully understanding how to set up certificates while installing rancher. I tried

--set tls=external

since I'm using ALB but not sure which is the right option to use 4. Should I be using a NGINX ingress in this setup? We have a rancher 2.6.x version running already and I don't see any NGINX ingress installed there Any suggestions would be extreme helpful. We are using an AWS EKS version 1.21 so we are getting charged for extended support. I really need to get the rancher working for upgrading the EKS cluster

I'm afraid i can't be much help on your core issue, but k8s 1.21 has not been supported since Rancher 2.6 which was EOL last month. We would recommend using the latest stable release, which i think is 2.8.4 and supports k8s 1.25 - 1.28

Yeah.. I'm trying to install rancher 2.8.4 right now but got stuck with the above mentioned issues

What happens when you try with k8s 1.28?

I'm first trying to install rancher then try to import the cluster into it. During the installation itself I'm facing the #1 issue. When I tried to install rancher directly inside the node, I got

node had disc pressure

error and rancher pods kept on getting evicted. The node type is AWS EC2 t3.medium instance with 2vCPU and 4gb ram

ok, so you're trying to install Rancher 2.8.4 into EC2 and then import an EKS 1.21 cluster into it to update it's kube version?

I'm trying to create new EKS cluster with version 1.28 and rebuild our infrastructure there and import this cluster into latest rancher. I thought create a new cluster is a better approach instead of upgrading the old one.

In terms of upgrade we would recommend follow this guide. Think somewhere in there it mentions upgrading local/downstream clusters k8s version in line with the max supported version for the rancher version, then upgrading rancher, repeat that cycle until on the desired version of both. I don't think importing an unsupported cluster and then upgrading it will work (there are also further limitations on imported clusters, most types won't allow lifecycle management actions like upgrading k8s)

sure... i'll consider.