https://rancher.com/ logo
#rke2
Title
# rke2
a

adamant-kite-43734

05/05/2022, 3:21 PM
This message was deleted.
b

bland-account-99790

05/05/2022, 4:06 PM
why not? What's your concern?
g

gentle-church-53565

05/05/2022, 4:19 PM
Hello Manuel, The issue faced atm the moment is that we have an RKE2 Server Running Rancher Server. We have used that Rancher Server to deploy another RKE2 but in an Air-Gapped Setup. We have then installed Rancher Server onto that RKE2. The conflict happens at the agent level. It seems that, when the server is setup by rancher it installs and rancher agent + fleet. The issue we are facing is when you we install Rancher Server on the Second RKE2 it already exists an agent and it trys to over write it. Then we get this conflict:
Copy code
2022/05/05 16:12:30 [FATAL] feature flag [multi-cluster-management] value has changed, rancher must be restarted
We also see attempts in the background of the second rancher server trying it clone data:
Copy code
unable to access '<https://git.rancher.io/rke2-charts/>
We have used Helm v3 Rancher 2.6.4 to get to this stage of the conflict.
a

acceptable-gigabyte-29847

05/05/2022, 5:26 PM
@bland-account-99790 any ideas?
@creamy-pencil-82913 any ideas?
w

wonderful-helicopter-16401

05/06/2022, 1:06 AM
Watching this. I am going to be attempting similar next week.
g

gentle-church-53565

05/06/2022, 8:08 AM
Hey Troy, we wish you the best of luck on this journey. We did solve one of the issues related to the charts being pulled over the internet. We enabled the offline backage option in the values files of helm and told the deployment the directory to find locally for the asc package. Though this conflict with the agents over writing is a major issue. I wonder, is there a way to install rancher without importing the local cluster its sitting on? this would solve agent conflict. Since the Hierarchy Rancher is managing the cluster we dont need 2 Rancher servers managing the same cluster.
a

acceptable-gigabyte-29847

05/06/2022, 4:49 PM
Turns out its not supported 🤦
If you have a Docker installation of Rancher, the node running the Rancher server should be separate from your downstream clusters.
If Rancher is intended to manage downstream Kubernetes clusters, the Kubernetes cluster that the Rancher server runs on should also be separate from the downstream user clusters.
https://rancher.com/docs/rancher/v2.5/en/overview/architecture-recommendations/
w

wonderful-helicopter-16401

05/07/2022, 10:06 PM
I found this out a couple weeks ago. I'm primarily interested in Rancher/Harvester for edge use cases and have to deploy a cluster just for Rancher to manage that Harvester cluster. It definitely limits the potential of some situations that I originally thought would be game changers, but there are enough opportunities that I've decided to move forward despite my reduced excitement of possibilities. As far as running a cluster on a cluster goes, though I don't condone this, I wonder if you'd be able to run a cluster via VMs in Harvester HCI. I know you can use other hypervisors to use the same hardware. I've just done it in Proxmox as a lab.