VMware Cloud Community
Intex
Contributor
Contributor
Jump to solution

Adding NICs Causes Disconnection

I've recently added a couple of additional NICs to a 3.0.2 ESX server. Everything went as planned and the server detected the new hardware and re-configured itself and performed a auto-restart of the OS.

When the server came online, all looked fine and I could ping the host. However virtual infrastructure shows the host as disconnected. If I attempt to connect to the web console or directly with the VC, I can't get in.

This would seem to suggest that the additional NICs have some how re-configured the setup. If I remove both adapters and perform a reboot, it's all OK again. Is there anything that you need to do to ensure that the configuration remains the same for the initial ethernet card or is there simply a way to determine what NIC is the preferred card and re-configure the settings for it to use that?

I'm doing all of this in an ESX server that's running in a VM on VMW6, so I'm speculating this may have something to do with the way in which additional cards are added to the VM.

Any help appreciated.

Reply
0 Kudos
1 Solution

Accepted Solutions
ctfoster
Expert
Expert
Jump to solution

When you add a new pNIC this can mess up the enumeration of the PCI bus - this can reorder the vmnics. The first job it to get your console back online.

Put the ethernet cable into the port you want to run the VI console on.

from the console run

esxcfg-nics -l

Make a note of the vmnic(x) of the live port. Lets call it vmnic2

List all the vswitches on your host - find the one with the service console - lets call it vswitch0

esxcfg-vswitch -l

Remove any pNIcs from that vSwitch if any are assigned.

esxcfg-vswitch -U vmnic0 vSwitch0

Once its freed up add vmnic2 back to into vswitch0

esxcfg-vswitch -L vmnic2 vSwitch0

Get back onto the console and tidy up from there.

View solution in original post

Reply
0 Kudos
7 Replies
ctfoster
Expert
Expert
Jump to solution

When you add a new pNIC this can mess up the enumeration of the PCI bus - this can reorder the vmnics. The first job it to get your console back online.

Put the ethernet cable into the port you want to run the VI console on.

from the console run

esxcfg-nics -l

Make a note of the vmnic(x) of the live port. Lets call it vmnic2

List all the vswitches on your host - find the one with the service console - lets call it vswitch0

esxcfg-vswitch -l

Remove any pNIcs from that vSwitch if any are assigned.

esxcfg-vswitch -U vmnic0 vSwitch0

Once its freed up add vmnic2 back to into vswitch0

esxcfg-vswitch -L vmnic2 vSwitch0

Get back onto the console and tidy up from there.

Reply
0 Kudos
Intex
Contributor
Contributor
Jump to solution

Great, thanks.

Reply
0 Kudos
Intex
Contributor
Contributor
Jump to solution

OK, this doesn't appear to work.

My pNIC was changed (from VMNIC0 to VMNIC2) after adding two additional NIC's to the server. vSwitch0 was therefore not correct as this now pointed to the wrong physical NIC. I therefore remove that NIC (VMNIC0) from vswitch0 and re-assigned VMNIC2 to vSwitch0. Everything then appeared to look OK with the correct bindings.

I can ping other ESX servers, but the host still doesn't re-connect in VC. I can't connect to the server using the web console, nor can I connect using the VI cilent directly.

You mentioned about going into the VI client and tidying up from there, but as I can't connect, I can't do any tidying up. What tidying remains?

Any ideas?

Reply
0 Kudos
Intex
Contributor
Contributor
Jump to solution

Problem solved.

The original answer would have worked if the ESX setup was all in the physical world. However, as the environment was contained within VMW6, the issue actually related to the fact that the additional NIC's were presented to the ESX server config as the wrong NIC type.

I remembered that the original installation of ESX required a configuration change to the .vmx file, specifically the LAN card had to be identified as "E1000", otherwise VC wouldn't connect. After adding the two NIC's and modifiying vmx file, everything worked.

Reply
0 Kudos
ctfoster
Expert
Expert
Jump to solution

First can you get a ping from the VC client to the ESX host address and you routing between the two or are they on the same subnet. I assume you've tried rebooting the server or restarting the mangement services from the CLI.

service mgmt-vmware start

service vmware-vpxa restart

When I said tidy up I just meant going into Configuration / Networking and making sure all the vSwitches were bound to the right pNICs.

ctfoster
Expert
Expert
Jump to solution

You got there before me - thanks for posting up the fix. Glad it worked out.

Reply
0 Kudos
Intex
Contributor
Contributor
Jump to solution

An easy one to forget! Thanks for your help.

Reply
0 Kudos