ESXi Paths Remain "Dead" Following Storage Switch Recovery

Products

• VMware vSphere ESXi

Issue/Introduction

Following a storage switch restart or hardware failure, ESXi hosts may fail to automatically restore storage paths, even in a redundant multipath configuration. While some paths may recover, others can remain in a "Dead" or "Offline" state. Standard HBA rescans may remove the dead paths from the UI but fail to bring the underlying storage targets back online, eventually necessitating a full host reboot to restore connectivity.

Symptoms

• Storage connectivity is lost or degraded across multiple ESXi hosts.

• The ESXi client shows several paths in a Dead state.

• Running a storage rescan removes the "Dead" entries, but the paths do not transition to "Active/On." 

• Initial IO Aborts and Host Status H:0x5

2026-02-12T12:24:13.944Z In(182) vmkernel: ... lpfc_handle_status:5631: vmhba3 3271: FCP cmd x2a failed ... Abort Requested Host Abort Req
2026-02-12T12:24:13.944Z In(182) vmkernel: ... NMP: nmp_ThrottleLogForDevice:3898: H:0x5 D:0x0 P:0x0 . Act:EVAL.
2026-02-12T12:24:13.944Z Wa(180) vmkwarning: ... WARNING: NMP: nmp_DeviceRequestFastDeviceProbe:235: NMP device "naa.624..." state in doubt; requested fast path state update...

• Partial Recovery of Some Paths

2026-02-12T13:02:15.002Z In(14) vobd[2098535]: [vob.scsi.scsipath.pathstate.on] scsiPath vmhba3:C0:T3:L2## changed state from dead
2026-02-12T13:02:15.005Z In(14) vobd[2098535]: [esx.clear.storage.redundancy.restored] Path redundancy to storage device ... restored.

• Prolonged IO Failures and Path Removal

2026-02-12T12:25:19.301Z In(182) vmkernel: ... VMW_SATP_ALUA: satp_alua_issueCommandOnPath:1005: Path "vmhba3:C0:T5:L2##" (UP) command 0x12 failed with status Timeout. H:0x5 D:0x0 P:0x0 .
2026-02-12T15:32:38.291Z Wa(180) vmkwarning: ... Could not get page 83 INQUIRY data for path "vmhba3:C0:T5:L242" - No connection (195887168)
2026-02-12T15:36:14.038Z Wa(180) vmkwarning: ... WARNING: ScsiPath: 8283: Remove path: vmhba3:C0:T5:L242

Environment

 

VMware ESXi 7.x 
VMware ESXi 8.x
VMware ESX 9.x 

 

Cause

The issue is caused by a Link state discrepancy following an abrupt switch crash or restart.

• In a normal scenario when a SCSI command fails with Host Status H:0x5, the system issues a TEST_UNIT_READY (0x0) command. If this also fails, the path is marked Dead.
  
  
• While the physical link may return to an "Up" state, the HBA driver can become stuck in a state where it fails to initiate the necessary FLOGI/PLOGI  sequence.
  
  
• Following an abrupt switch crash or restart, the HBA can enter an 'awake but blind' state. While the physical link is active, the software driver fails to re-register with the fabric, preventing the discovery of any storage targets.

Resolution

To avoid a disruptive host reboot, follow these steps to force the HBA to re-initialize:

Step 1: Bounce the Switch Port

• A "ghost" FLOGI session may persist on the switch. Log into the SAN Switch CLI, identify the port connected to the ESXi host, and toggle the port (shutdown/no shutdown).
  
  
• This forces the switch to clear the stale session and triggers a new login request from the ESXi host.

Step 2: Force an HBA-level Reset (Software Cold Boot)

• If a standard rescan fails, force the ESXi kernel to reset the HBA driver. This simulates a hardware reset for that specific PCIe card without affecting the rest of the host.
  
  
• Run the following command via SSH (replace vmhbaX with your specific adapter):

esxcli storage san fc reset -A vmhba2

• This forces the HBA to drop the laser link, bring it back up, and issue a fresh FLOGI.

Step 3: Host Reboot

• If these steps fail to restore the paths to an "Active" state, a full reboot of the ESXi host is required.