Verification¶

Run these against the target cluster before installing. All six must pass.

1. Version consistency¶

kubectl get nodes -o custom-columns="NAME:.metadata.name,VERSION:.status.nodeInfo.kubeletVersion"

Expect: every node on the same version, 1.30 or later.

2. Admin access¶

kubectl auth can-i '*' '*'

Expect: yes.

3. Node readiness¶

kubectl get nodes
kubectl get nodes --show-labels

Expect: all target nodes Ready, with workload placement labels applied.

4. GPU availability¶

kubectl describe node <gpu-node> | grep -A2 "nvidia.com/gpu"

Expect: non-zero nvidia.com/gpu under both Capacity and Allocatable. Zero or absent means the driver stack is not working.

5. Default StorageClass¶

kubectl get storageclass

Expect: one class marked (default).

6. LoadBalancer provisioning¶

kubectl create svc loadbalancer lb-test --tcp=80:80
kubectl get svc lb-test -w
kubectl delete svc lb-test

Expect: an EXTERNAL-IP is assigned. A persistent <pending> means the cluster has no LoadBalancer provider - install MetalLB, or another LoadBalancer implementation, before installing shaide.

Next¶

Continue to Prerequisites for what the installer itself requires.