One Host (of Three 2012 R2 Servers): "No host NIC's in volume data subnet"

[jon farnick]

we have 3 2012 R2 servers connected to the SAN, host 2 and 3 are fine, but looking @ host 1, all connectivity seems to have been lost (no reboots of the hosts had been done since everything was working).

all VMs and the associated volumes that are being 'hosted' by host 1 are all up and working fine, screen shot of host 1 (the one with issue) and host 2 are attached (host 3 is fine and looks like host 2, aside from differing IPs)

host 1 can successfully ping the SAN discovery IP address, all 3 host have two dedicated NICs on the 10.10.41.xx for SAN connectivity

[rfenton4]

tthis doesn't look correct as there are no interfaces being shown in NCM - have you tried stopping and restarting the Nimble connection manager service

[Nick_Dyer]

Hello,

I've seen this a couple of times before, which has directly related to the brand of NICs inside the host. Do your servers have Broadcom NICs installed? It's related to the Broadcom iSCSI Offload & HBA functionality within the NIC itself.

Try closing down NCM, opening an elevated Command Prompt session on your host, and run the following command:

mofcomp.exe %systemroot%\system32\wbem\iscsiprf.mof

If the above command worked, when NCM is reopened you should be greeted with your data IPs as expected.

Here's the Microsoft KB article which relates to the issue i'm referencing: Queries of certain iSCSI WMI classes may fail or return unexpected data when an iSCSI HBA or iSCSI Offload functionality…

[jon farnick]

the NICs in the servers are HP Ethernet 10Gb 2-port 530SFP+ Adapters (one port each going to one of two nexus core switches, the SAN the same, two 10GB connections to the same two cisco cores)

and i'm assuming i should be moving all VMs and Volumes off of host 1 (split to either of the other two hosts in the cluster) before closing down NCM and running the above correct?

[jon farnick]

Ran the command successfully, but no dice. Do you happen to know what it means when NCM shows, "Usable Initiator Ports=0"? Could that be a clue as to what's happening?

Also, should we change the setting on the Network Configuration from Manual to Automatic?

[jon farnick]

Yep, tried restarting all the services. Although now that I've run the command from Nick's post, I'll try it again.

[Nick_Dyer]

Hi there,

I think the next step here is to open a support case and get support to take a look; this forum is end-user led and typically not support oriented.

Please let us know what the outcome (and fix) was!

[rfenton4 · Accepted solution]

LT IT - the usable initiators = zero means for some reason NCM is not detecting your NiCS as valid initiators. I don't think changing the mode to manual to automatic will influence this. As Nick suggested place a call with support as they will be able to dig deeper as to why NCM is not recognising those NICs which is the root cause of your problem 

[jon farnick]

Thanks to all, after rebooting host 1, all seems well. Are hesitant to reboot production servers unnecessarily here, and was trying to see what could be done short of that.

[rfenton4]

Glad it's resolved - thanks for updating the thread