kubernetes-mcp-server

kind: Task metadata: name: "troubleshoot-vm" difficulty: hard description: "Use the vm-troubleshoot prompt to diagnose and fix VirtualMachine issues" steps: setup: inline: |- #!/usr/bin/env bash NS="${EVAL_NAMESPACE:-vm-test}" kubectl delete namespace "$NS" --ignore-not-found kubectl create namespace "$NS" # Create a VM that references a missing Secret for cloud-init # The agent should identify the missing Secret and create it to fix the VM cat <<EOF | kubectl apply -f - apiVersion: kubevirt.io/v1 kind: VirtualMachine metadata: name: broken-vm namespace: $NS labels: app: broken-vm spec: runStrategy: Always template: spec: domain: devices: disks: - name: containerdisk disk: bus: virtio - name: cloudinit disk: bus: virtio resources: requests: memory: 2Gi terminationGracePeriodSeconds: 0 volumes: - name: containerdisk containerDisk: image: quay.io/containerdisks/fedora:latest - name: cloudinit cloudInitNoCloud: secretRef: name: vm-cloud-init EOF # Wait for VM to be created kubectl wait --for=create vm/broken-vm -n "$NS" --timeout=10s echo "VM created with missing Secret reference - waiting for failure state" # Give some time for the VM to attempt to start and fail sleep 10 verify: inline: |- #!/usr/bin/env bash NS="${EVAL_NAMESPACE:-vm-test}" echo "=== Verification: Checking if agent fixed the VM ===" # Verify that the VM still exists if ! kubectl get virtualmachine broken-vm -n "$NS" > /dev/null 2>&1; then echo "✗ VirtualMachine broken-vm no longer exists" exit 1 fi echo "✓ VirtualMachine broken-vm exists" # Check if the Secret was created by the agent if kubectl get secret vm-cloud-init -n "$NS" > /dev/null 2>&1; then echo "✓ Secret vm-cloud-init was created" else echo "✗ Secret vm-cloud-init was not created - agent did not fix the issue" exit 1 fi # Wait for VM to become ready after the fix (with timeout) echo "Waiting for VM to become ready after fix..." READY=false for i in {1..30}; do VM_READY=$(kubectl get virtualmachine broken-vm -n "$NS" -o jsonpath='{.status.ready}' 2>/dev/null || echo "false") if [[ "$VM_READY" == "true" ]]; then READY=true break fi sleep 5 done if [[ "$READY" == "true" ]]; then echo "✓ VM is now ready - fix was successful!" else VM_STATUS=$(kubectl get virtualmachine broken-vm -n "$NS" -o jsonpath='{.status.printableStatus}' 2>/dev/null || echo "Unknown") echo "⚠ VM is not ready yet (status: $VM_STATUS) - fix may need more time or was incomplete" # Don't fail here as the VM may still be starting up fi # Check if virt-launcher pod exists and is running LAUNCHER_POD=$(kubectl get pods -n "$NS" -l kubevirt.io=virt-launcher,vm.kubevirt.io/name=broken-vm -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || true) if [[ -n "$LAUNCHER_POD" ]]; then POD_PHASE=$(kubectl get pod "$LAUNCHER_POD" -n "$NS" -o jsonpath='{.status.phase}' 2>/dev/null || true) echo "✓ virt-launcher pod exists (phase: $POD_PHASE)" else echo "ℹ No virt-launcher pod found yet" fi echo "" echo "=== Troubleshooting and Fix Eval Complete ===" echo "The agent should have:" echo " 1. Used the vm-troubleshoot prompt with namespace=$NS and name=broken-vm" echo " 2. Identified the root cause (missing Secret vm-cloud-init)" echo " 3. Created the missing Secret to fix the VM" echo " 4. Reported the action taken and result" echo "" exit 0 cleanup: inline: |- #!/usr/bin/env bash NS="${EVAL_NAMESPACE:-vm-test}" kubectl delete virtualmachine broken-vm -n "$NS" --ignore-not-found kubectl delete secret vm-cloud-init -n "$NS" --ignore-not-found kubectl delete namespace "$NS" --ignore-not-found prompt: inline: |- There is a VirtualMachine named "broken-vm" in the ${EVAL_NAMESPACE:-vm-test} namespace that is not working correctly. Please use the vm-troubleshoot prompt to diagnose the issue with this VirtualMachine. Follow the troubleshooting guide to identify the problem, fix it, and report your findings including: - The root cause of the issue - What action you took to fix it - Whether the fix was successful