https://rancher.com/ logo
Title
m

most-sunset-36476

09/28/2022, 2:08 PM
Hi all do we have less visibility over the downstream clusters using AKS ? (Nodes Provisioned by Hosted Kubernetes Providers) than with RKE downstream clusters (Nodes Hosted by an Infrastructure Provider) ? Though the Rancher UI can we see same amount of info for both clusters ?
q

quick-sandwich-76600

09/28/2022, 5:35 PM
In terms of the info you can get through the Kubernetes API, you'll see the same info from both clusters. From an operational point of view, the details you'll see for each cluster are different, and so are the type of changes that you can make to those clusters. Is that what you wanted to know?
1
m

most-sunset-36476

09/28/2022, 5:45 PM
My problem is that i use the Terraform rancher2 provider to create downstream RKE Rancher clusters on Azure. And when I create the downstream RKE cluster with a user-addon (Which is a manifest that creates a service type LoadBalancer pointing to the ingress controller and is ran as a user-addon job at cluster creation time) that resource (cloud Load Balancer) has not been provisioned by Terraform and prevents me from running
terraform destroy
successfully because the Load Balancer stays in the resource group and uses the subnet created by Terraform.
Ah sorry wrong thread
I wanted to know if we have same visibility over the downstream clusters created by Rancher for AKS or RKE downstream clusters. From the documentation it appears that both have full visibility over cluster resources, but i wanted to have confirmation. Can we see as much information about cluster resources for both cluster types ? • https://docs.ranchermanager.rancher.io/pages-for-subheaders/cluster-configurationhttps://docs.ranchermanager.rancher.io/how-to-guides/advanced-user-guides/manage-clusters/nodes-and-node-pools
q

quick-sandwich-76600

10/01/2022, 4:29 PM
For RKE you have full control as you are the provisioner of the RKE Kubernetes cluster itself. When using AKS the number of cluster settings you can see is limited, for example, the control layer is something you can't modify in any way as it's provided by Azure as part of their managed service, and you can't even scale it up or down (you can only scale up/down the workers).
m

most-sunset-36476

10/02/2022, 7:09 PM
I wasn't able to scale up/down the workers for AKS, are you sure it is possible to scale up the worker nodes though the Rancher UI ?
By contorl plane in AKS, you mean the default node pool ?
q

quick-sandwich-76600

10/04/2022, 9:06 PM
No, I mean the control nodes (where API Server, etcd, ... are placed), that's hidden from users in AKS.
Just change the node count
1
m

most-sunset-36476

10/04/2022, 9:10 PM
I already tested creating an AKS downstream cluster with 1 default system pool. The scaling of nodes through the UI was not possible. The Rancher docs also state that scaling of nodes is only possible for RKE clusters created using node templates
And thank you for your help and taking the time to answer me btw! :)
👍 1
I was finally able to fix my problem using the kubectl provider
q

quick-sandwich-76600

10/07/2022, 1:44 PM
It was my pleasure to help, and I'm glad you made it work.
I had planned to update the thread a couple of days ago, but I didn't have time. Today I found a slot to record a small video showing what I tried to explain above. In the video, I scale an AKS cluster from 2 worker nodes to 4 worker nodes just updating the node count in the config screen:

https://youtu.be/8u2CbyynQIU

It just shows the process with no cut, it takes like 6 minutes but in the end, you have your cluster with 4 nodes ready.
m

most-sunset-36476

10/07/2022, 1:59 PM
perfect, thank you very much for your help :)