Unable to discover NVME storage on one vmhba device. Host reports error: An error occurred during host configuration: . Unable to connect to the specified transport target using the provided connection specification.
VMkernel.log:
2024-12-16T17:06:16.383Z Al(177) vmkalert: cpu44:8749658)ALERT: nvmerdma: 465: No tagged vmknic interface found. Please tag relevant vmknic(s) for steering NVMe/RDMA traffic correctly.
2024-12-16T17:06:16.383Z In(182) vmkernel: cpu44:8749658)nvmerdma:476 [ctlr 329, queue 0] Failed to get source ip address for uplink vmnic10: Not found
2024-12-16T00:21:33.834Z Wa(180) vmkwarning: cpu9:8697676)WARNING: NVMFDEV:3554 vmhba66 adapter find in the global connection list failed: Not found
VMware vSphere ESXi 7.x
VMware vSphere ESXi 8.x
Since one vmhba is working and the other isn't, it suggests that the issue might be specific to the non-working VMHBA.
This could be due to a configuration problem on the virtual switch, missing VLAN tags, or a configuration problem on the fabric at the host level.
Check VMkernel NIC Tagging:
Ensure that the VMkernel NIC is tagged correctly for NVMe/RDMA traffic. You can use the following command to tag the NIC:
esxcli network ip interface tag add --interface-name vmkN --tagname NVMeTCP
Replace vmkN with the appropriate VMkernel NIC ID.
Verify RDMA Configuration:
Ensure that the RDMA configuration is correct and that the network adapter supports RDMA. You can check this with:
esxcli network nic list
Look for the RDMA capability in the output.
Try to perform a rescan operation to see if it gets detected:
esxcli storage core adapter rescan
esxcli storage core adapter list
Then try to reset the HBA from command line:
esxcli storage san fc reset -A vmhba66
Verify Fabric Configuration on Host:
Check the configuration of the uplinks on the fabric at the host level.
In this example, we found that the uplink for Fabric B on the host was not correctly configured and was missing its required VLAN.