Load Balancer IP Address

I'm working with my DKU platform (running in Azure) to configure my kubernetes cluster with the goal of ensuring my load balancer node runs with the same IP address every time - e.g. 10.1.1.100. My AKS plugin configuration has a node pool named API-LoadBalancer. It has one node and does not autoscale. The admin panel's clusters tab has a single cluster named dev-clstr1 with four node pools defined (three compute and one system). To get the API-LoadBalancer node pool running, I have to deploy the load balancer yaml after the DSS instance and its cluster are running. By the time I get to deploying the yaml, the IP address I want to use is already being consumed by one of the other node pools. How do I prevent the other node pools from using the IP address I want? I want to do this inside the DKU UI and not have to script something in the Azure CLI if possible.
FYI: all node pools are running in the same subnet - It'd be easier if I don't have to go back to infrastructure to change the design.
Operating system used: Almalinux
Answers
-
Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 2,580 Neuron
Dataiku does not need a load balancer to run work loads in Kubernetes. What exactly are you trying to achieve? What's the point of the load balancer?
-
The load balancer handles incoming requests and then forward them to the appropriate node pools for execution. I want the load balancer to have the same IP address all the time so I can use a name like https://api.dev.acme.com/dku/model1 without having to update the DNS server every time the ip address changes.
I should add that the load balancer is to handle the api call load of thousands (10-12) of daily calls instead of the automation node. This is the reasoning behind using the load balancer.
Does this sound like an odd situation to you? Maybe I am missing some key concept? -
Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 2,580 Neuron
What kind of work loads are you looking to run? Flow / Recipe / Scenarios won’t use a load balancer. Only custom API services deployed to a API node infrastructure running in Kubernetes would use a load balancer.
-
Hi Turribeach,
It is the latter - custom APIs using an ingress-nginx node running in Kubernetes.