Cluster Validation

The cluster validation script checks that your OpenShift cluster is properly configured and ready to run KubeVirt performance tests.

Validation Checks

The script validates:

kubectl access and cluster connectivity
OpenShift Virtualization installation and health
KubeVirt resource status (Deployed phase)
Critical deployments: virt-api, virt-controller, virt-operator
virt-handler daemonset on all nodes
Storage class availability
Worker node readiness
DataSource availability (optional — only needed for datasource-clone and chaos tests)
User permissions
Node resource utilization

Running Validation

# Basic validation
virtbench validate-cluster --storage-class YOUR-STORAGE-CLASS

# Quick validation (skip slower checks)
virtbench validate-cluster --storage-class YOUR-STORAGE-CLASS --quick

# Use a specific kubeconfig
virtbench --kubeconfig /path/to/kubeconfig validate-cluster \
  --storage-class YOUR-STORAGE-CLASS

# Comprehensive validation
virtbench validate-cluster --all --storage-class YOUR-STORAGE-CLASS

# With custom DataSource
virtbench validate-cluster \
  --storage-class YOUR-STORAGE-CLASS \
  --datasource fedora \
  --datasource-namespace openshift-virtualization-os-images

# Require minimum worker nodes
virtbench validate-cluster \
  --storage-class YOUR-STORAGE-CLASS \
  --min-worker-nodes 5

Validation Options

Option	Description	Default
`--storage-class NAME`	Storage class name to validate	(required)
`--quick`	Skip DataSource, SSH pod, and node resource checks	false
`--datasource NAME`	DataSource name to validate	rhel9
`--datasource-namespace NS`	DataSource namespace	openshift-virtualization-os-images
`--ssh-pod NAME`	SSH pod name to validate	ssh-test-pod
`--ssh-pod-namespace NS`	SSH pod namespace	default
`--min-worker-nodes NUM`	Minimum worker nodes required	1
`--all`	Run all validation checks	false
`--kubeconfig PATH`	Global `virtbench` option for kubeconfig path	`KUBECONFIG` environment variable or kubectl default

Exit Codes

0 - All checks passed, cluster is ready
1 - One or more checks failed, cluster not ready

Understanding Validation Output

Successful Validation

✓ kubectl access and cluster connectivity
✓ OpenShift Virtualization installed and healthy
✓ Storage class available
✓ Worker nodes ready (5 nodes)
✓ DataSource available
✓ User permissions verified

Cluster validation passed! Ready to run benchmarks.

Failed Validation

✓ kubectl access and cluster connectivity
✗ OpenShift Virtualization not found or not healthy
  - KubeVirt resource not in Deployed phase
✗ Storage class 'YOUR-STORAGE-CLASS' not found
✓ Worker nodes ready (3 nodes)
⚠ DataSource 'rhel9' exists but is not ready - WARNING (only needed for datasource-clone/chaos tests)

Cluster validation failed. Please fix the issues above.

A missing or not-ready DataSource is reported as a warning, not a failure — it only blocks the datasource-clone and chaos tests. Validation fails only on the hard checks above (connectivity, virtualization, storage class, worker nodes, permissions).

Troubleshooting Validation Failures

OpenShift Virtualization Not Found

Check if KubeVirt resource exists:

kubectl get kubevirt -A
# Expected: NAMESPACE openshift-cnv, PHASE Deployed

Solution: Install OpenShift Virtualization operator or KubeVirt

Components Not Ready

Check deployment status:

# Check deployment status
kubectl get deployment -n openshift-cnv | grep -E "virt-api|virt-controller|virt-operator"

# Check pod logs for errors
kubectl logs -n openshift-cnv deployment/virt-api

Solution: Review operator logs and ensure all components are running

Storage Class Not Found

List all storage classes:

kubectl get storageclass

Solution: Create a storage class appropriate for your storage backend. Refer to your storage provider's documentation.

DataSource Not Found

Check available DataSources:

kubectl get datasource -n openshift-virtualization-os-images

Solution: - Verify the DataSource name and namespace - Create DataSource if missing - Use --datasource flag to specify a different DataSource

Insufficient Worker Nodes

Check worker node count:

kubectl get nodes -l node-role.kubernetes.io/worker

Solution: - Add more worker nodes to the cluster - Adjust --min-worker-nodes if fewer nodes are acceptable

Permission Denied

Check user permissions:

# Test namespace creation
kubectl auth can-i create namespaces

# Test VM creation
kubectl auth can-i create virtualmachines -n default

# Test pod exec
kubectl auth can-i create pods/exec -n default

Solution: Request cluster-admin or appropriate RBAC permissions from cluster administrator

Pre-Flight Checklist

Before running benchmarks, ensure:

[ ] Cluster validation passes
[ ] Storage class supports dynamic provisioning
[ ] Storage class is compatible with KubeVirt DataVolumes
[ ] SSH test pod is running (for network tests)
[ ] Sufficient cluster resources available
[ ] DataSource exists and is ready (only for datasource-clone / chaos tests)