This message was deleted.
# vsphere
a
This message was deleted.
m
Howdy, since you are on vSphere and deploying K8s clusters there - I would recommend having an external (to your K8s clusters) load balancer like Nginx Plus or HA Proxy. You will also need to rely on whatever DNS provider you are using for your vSphere environment for the service URLs (ours is Infoblox at my institution).
In my environment, we have the service URL in Infoblox as a Shared CNAME of our Nginx Plus VM. Then we use Nginx Plus to point our listeners to the appropriate K8s cluster’s worker nodes via FQDN, as upstreams. From there, we rely on the cluster’s Nginx Load Balancing config to route to the proper workloads/services running in our K8s clusters.
b
@modern-television-79263 Thanks for your response and time. As per your answer, traffic follow this path External loadbalancer(Ex: HA proxy out side the k8s cliuster on another virtual machine) ==> k8s nodes in node pool in rancher ==> Now traffic is at inside the k8s cluster, from here we can use nginx class to direct the different cluster ip services as per ingress rules . Please correct me if my understanding is wrong. is there any way we can add k8s nodes in node pool in rancher to HAproxy dynamically?
@modern-television-79263 Please let me know your response that will help me a lot. Thanks
m
Howdy @brash-jewelry-40212 that sounds correct. I’ve not found a way to auto-gen the additional K8s nodes into HA Proxy or Nginx config - without using something like a Git/terraform/Jenkins pipeline that also controls the state of the cluster.
b
@modern-television-79263 Ok.
@modern-television-79263 I did setup metalLB instead of HAproxy. It worked for me to get the external private ip.