ESXi storage adapter alias (vmhba) changed after replacing NVMe or PCIe storage devices
Article ID: 453121
Updated On:
Products
Issue/Introduction
When replacing a failed NVMe SSD or PCIe storage adapter on an ESXi host (such as a vSAN ReadyNode):
The newly installed device receives a different storage adapter alias (e.g., changing from vmhba6 to vmhba9).
The replacement drive does not automatically revert to its original vmhba number.
Running "esxcli nvme device list" or "esxcli storage core adapter list" displays the new vmhba identifier.
Environment
VMware vSphere ESXi
Cause
This behavior is by design and complies with the VMware ESXi device naming architecture described in Broadcom KB 324534.
Non-Persistent Runtime Aliases:
vmhba names are dynamic runtime aliases assigned by ESXi drivers during device enumeration. They are not permanently fixed to a physical PCIe slot.
Device Addition Logic (Hot-Plug Handling):
When a hardware component is swapped while the system is online or during hot-plug events, ESXi treats the replacement as a removal event followed by an addition event ("a removal followed by a new addition").
Next Available Alias Allocation:
To prevent runtime logical conflicts, the native driver (nvme or nvme-pcie) does not immediately recycle the newly released alias. Instead, it assigns the next available unassigned alias in the sequence (e.g., assigning vmhba9 if vmhba8 was the highest active identifier).
Resolution
No administrative action is strictly required for the vmhba numbering change. However, ensure the newly replaced drive is properly claimed in vSAN:
vSAN OSA (Original Storage Architecture): Ensure the drive is claimed into the target Disk Group as Cache or Capacity tier via vCenter.
vSAN ESA (Express Storage Architecture): Ensure the drive is added back to the vSAN Storage Pool.
Feedback
