Deployment of Nephio on Physical Servers for CI Testing

[arora-sagar]

Anuket lab of UNH can provide Nephio resources to create the same topology of R2 and topology of R1.

Here is what I think we can start with:

• In the beginning, we can start by reserving four servers in Anuket Lab, one for the Nephio MGMT cluster, three for the workload cluster, and reserving 10 VLANs.
• Pre-installed OS (Ubuntu or Fedora) on Nephio MGMT cluster
• Pre-installed K8s cluster (Any distribution of K8s) for Nephio mgmt cluster
• Option 1: Pre-installed OS on workload cluster and create a cluster using Nephio
• Option 2: OS installation and cluster creation via Nephio (We can take the help of Sylva)
• Networking will be manual unless we find an alternative to container lab
• After every CI run we will have to un-install all the Nephio packages and the Nephio workload cluster
• Test on physical servers can run weekly rather than triggering on every MR.

The idea is to discuss and define the CI testbed.

Feel free to join the thread.

[henderiw]

I believe it is fine to have static VLANs assigned. We need to register these to the resource backend or use kuid.

[electrocucaracha]

What about to control the K8s installation of the workload clusters with Kubean?