Storage paths to Pure Storage FlashArray appearing as Dead or Down on ESXi

Products

• VMware vSphere ESXi

Issue/Introduction

When inspecting storage paths on a VMware ESXi host, a single physical storage device (identified by its unique NAA ID) is presented with differing LUN IDs depending on the host bus adapter (HBA) or target port used.

naa.6000##################
-> vmhba3:C0:T0:L246
-> vmhba3:C0:T1:L246
-> vmhba3:C0:T2:L237
-> vmhba3:C0:T3:L237
-> vmhba4:C0:T0:L237
-> vmhba4:C0:T1:L237
-> vmhba4:C0:T2:L246
-> vmhba4:C0:T3:L246

Environment

ESXi 7.x and above

Cause

This issue is caused by a Storage Array Presentation Error.

 

VMware ESXi strictly requires that a single physical volume (NAA ID) is presented with the exact same LUN ID across all available paths. The storage array (in this case, a Pure Storage FlashArray) is currently configured with overlapping, misaligned, or manual mappings.

 

 

Resolution

To resolve this issue, the storage presentation must be corrected directly on the storage array. This process should be performed during a maintenance window, as unmapping and remapping active storage can disrupt I/O.

 

Step 1: Correct the Storage Array Configuration (Pure Storage Example)

• Log into the Pure Storage FlashArray management interface.
• Navigate to the Storage > Hosts (or Host Groups) section.
• Verify the initiator WWNs for the affected ESXi host. Ensure that all WWNs belonging to this host are consolidated into a single Host object.
• If this ESXi host is part of a vSphere Cluster, ensure the Host object is placed within a Host Group.
• Disconnect the disjointed volume mappings.
• Remap the volumes directly to the Host Group (or the unified Host object). Mapping at the group level allows the array to automatically enforce a consistent LUN ID across all paths.

 

Step 2: Rescan ESXi Storage

• Once the storage presentation is uniform on the array:
• Log into VMware vCenter or the direct ESXi Host Client.
• Right-click the affected ESXi host(s) and select Storage > Rescan Storage.
• Allow the rescan to complete so ESXi can clear the stale, mismatched paths from its PSA tables.

 

Step 3: Verification

Verify the resolution by checking the storage paths via the ESXi CLI:

#esxcli storage core path list -d naa.6000##################

 

Additional Information

This typically occurs when:

 

1. Fibre Channel WWNs (initiators) for a single ESXi host are mapped as individual objects on the storage array, rather than being grouped into a unified Host Group.

 

2. A volume was manually mapped to specific initiator ports with a forced LUN ID, while the storage array automatically assigned a different "next available" LUN ID to other ports.

 

 

The impact on the  Host

If the host is using the Round Robin Path Selection Policy (`VMW_PSP_RR`), ESXi needs all paths to be uniform to balance I/O correctly.

Because the LUN ID mismatch, ESXi is taking protective measures:

Discarded Paths: If you look at the  `Working Paths` for the first device, ESXi is only actively using the paths associated with **L237**. It has intentionally ignored the paths presenting as L246 to prevent data corruption. 

ESXi HBA	Target Port	Presented LUN ID	Resulting Status
vmhba3	C0:T0	L246	❌ Ignored by Host
vmhba3	C0:T1	L246	❌ Ignored by Host
vmhba3	C0:T2	L237	✅ Active Working Path
vmhba3	C0:T3	L237	✅ Active Working Path
vmhba4	C0:T0	L237	✅ Active Working Path
vmhba4	C0:T1	L237	✅ Active Working Path
vmhba4	C0:T2	L246	❌ Ignored by Host
vmhba4	C0:T3	L246	❌ Ignored by Host