Model-specific installation checklist · exact AFF/FAS platform manual required
Example: a new two-controller AFF HA pair with one external shelf and two management switches. This is a planning example, not a port map; the port map must come from the exact system model guide.
About the command and output examples
Names, addresses and results are illustrative. Commands have not been executed against a live storage system or SAN switch for this publication. Confirm syntax and compatibility for your release, then rehearse in an approved lab. Existing-data recovery and destructive changes need an environment-specific plan. Read the editorial standards.
01Prepare a model-specific worksheet
- Record the controller model, serial numbers, shelf and module type, number of shelves, ONTAP release, rail kit, and switch models. Download the exact platform installation guide and Hardware Universe entry.
- Reserve rack units, verify rack depth, weight, cooling, grounding, redundant PDUs, lifting assistance, power connectors, and console access. Label intended rack positions before equipment arrives.
- Create a port-to-port matrix for controller A and B: cluster/HA, management, NAS/SAN host, storage shelf, and switch endpoint. Plan separate power feeds. Confirm transceivers and cable lengths.
Acceptance evidence: approved rack elevation, port map, IP plan, power plan, and inventory.
02Rack and cable
- Install shelves and controllers with the supplied rails and lifting procedure. Secure chassis and cable-management arms; keep ventilation and service clearance.
- Cable cluster/HA and management using the platform-specific switched or switchless diagram. Cable host networks and shelves with the specified module ports and multipath pattern. Label both ends and have another person trace every cable against the worksheet.
- Do not infer port numbers from a different AFF or FAS model. Connector orientation and shelf module port names can vary.
Confirm both controllers reach their planned switches and each required shelf path is present in the approved diagram.
03Power on and accept
- Follow the platform guide for shelf IDs and power-on sequence. Assign unique shelf IDs where required, allow any required power cycle, then bring up controllers with a console connected.
- Record boot messages and hardware faults. Complete cluster setup only after the hardware is discovered as expected.
- Verify nodes, HA, ports, shelves, and system health; record serials, shelf IDs, ONTAP and firmware versions, management addresses, and any exceptions in the handover.
cluster show
storage failover show
network port show
system health alert showBoth nodes healthy and eligible; HA ready; expected ports and shelves present; no unresolved hardware alert.
04Installation worksheet: prove each independent path
The practical deliverable of an installation is an as-built map that another administrator can trace. A neat rack and green front-panel indicators are not enough to demonstrate redundant storage, cluster and host connectivity. Record the system-specific cabling diagram revision and map each physical connection to it.
| Connection group | Record at both ends | Acceptance evidence |
|---|---|---|
| Power feed A/B | Chassis/PSU → PDU circuit | Approved separate power sources and PSU state |
| Cluster / HA | Controller port → peer or cluster switch port | Exact platform topology and operational health |
| Management | Node port → management switch | Console and management reachability for each node |
| Shelf paths | Controller → shelf module/port → next endpoint | Every required path in the platform diagram is traced |
| Client fabrics | Data port → switch/VLAN or FC fabric | Both intended failure domains are visible |
- Before powering on, have a second installer independently compare labels to the matrix. Record a mismatch and fix the worksheet or cable deliberately; do not guess the intended port.
- After boot, compare discovered inventory with the shipping inventory. Reconcile serial number, shelf/module count and faulted components before client provisioning.
- Take a baseline of health and the final map. A later replacement can then be compared with a known accepted state instead of a remembered layout.
05Example discrepancy: one shelf is visible through one path
Suppose a shelf appears in inventory but the expected second controller path is absent. Discovery has proved the shelf exists, not that it meets the model’s supported multipath design. Keep client commissioning on hold. Compare the missing path against the exact platform cable diagram, module type, port label and link status. Use the model service procedure for any cable manipulation and retain console/health evidence for support.
Handover includes serial inventory, model diagram revision, as-built cabling and resolved exceptions, not just a screenshot of cluster membership.
Use the section links to revisit a step, or follow a related guide below.
Official references and further reading
Choose the matching release in the vendor documentation before using a command or design. References open on the publisher’s website; linked material remains with its publisher.