Operate And Monitor

Sequence

1. Confirm scale — See resource-types.
2. Establish blast radius — Read-only: cluster, racks, bare metal machines, storage appliances, fabric.
3. Check control plane — Cluster Manager and NFC provisioning state before blaming workloads.
4. Use documented monitoring — Fetch multi-rack-monitor-overview and cluster metrics guidance via docs-map.
5. Narrow by layer — Workload, then logical network, then fabric, then hardware. Do not skip layers.
6. Targeted references — Arc VM troubleshooting for VM-level faults; storage appliance error messages for SAN faults; serial console for machine-level access.
7. Propose remediation — Present the change, its blast radius, and its reversibility. Get confirmation before acting.

Triage routing

Guardrails

• Never restart, reimage, or replace a machine without explicit confirmation.
• Never treat missing Resource Health data as a healthy fabric or SAN.
• Do not recommend redeployment or rack removal as an early step.
• Report partial evidence as partial; do not present absence of data as a clean result.
• Follow safety-rules.

Evidence to collect

• Cluster, Cluster Manager, NFC IDs and provisioning state.
• Affected rack, machine, and VM identifiers.
• Fabric and isolation domain state.
• Storage appliance status and error messages.
• Metrics and logs covering the incident window.
• Recent changes: updates, fabric edits, workload deployments.

Related references

• Docs map
• CLI and prerequisites
• Resource types
• Safety rules