Scenario
Example: a host sees only one of four expected paths after switch maintenance. Preserve the surviving path while diagnosing the missing fabric.
Before you change production
Commands and screens can differ by release and platform. Replace example names and documentation IP addresses. Check prerequisites, impact, current health and rollback with your change owner.
1. Confirm impact and identity
- Record host, IQN/WWPN, LUN serial, LUN ID, SVM, protocol, affected fabric, and time. Check whether I/O is still running and whether multipath is protecting the workload.
- On ONTAP, confirm LUN online and mapped to the intended igroup. Compare host initiator ID exactly.
Read-only checks
lun show
lun mapping show
network interface show
storage failover show2. Walk the path
- For iSCSI: check host NIC/VLAN/MTU and route, TCP portal reachability, target service, discovery/session login, igroup IQN, then multipath.
- For FC: check HBA and switch port link, active zone set, target WWPN login, FC LIF state, igroup WWPN, host rescan, then multipath.
- Inspect one failed path at a time. Do not change a healthy fabric during an active outage unless the plan requires it.
Verify
The missing path is identified at a specific layer, such as switch zoning or initiator mapping.
3. Restore and validate
- Apply the narrow fix through change control. Rescan the host and verify expected path count, path states, device serial, and I/O latency.
- If any device appears duplicated, stop before formatting or mounting. Verify serial and multipath aggregation.
Verify
All planned paths restored and stable; host logs show no new I/O errors.
If validation fails
- When the host reports an I/O error, preserve surviving multipath connectivity and correlate host time with ONTAP EMS and switch logs.
- When the LUN appears as a new disk after maintenance, compare serial numbers and multipath WWID before mounting, importing or formatting.
- When paths oscillate, inspect link errors, SFPs, fabric events and target LIF placement rather than remapping the LUN repeatedly.
Verify
Re-run the original validation and record the observed result, exact error, time, and corrective action.