VMs randomly lose NIC?

[HWMatt]

Hello -

I have 2 ESXi 4.0+ servers, running XP VMs (32 and 64 bit). We have only had these servers a couple months. On 3 separate occasions now, a random VM will lose it's NIC (VMXNET3 or otherwise) from it's settings configuration in ESXi. I don't see anything in the logs, users panic because they can't reach their VM, but it's generally quite simple to restore the NIC and reboot the VM and all is well. There is no smoking gun to this for me, I have read other similar posts and don't feel like the same conditions apply (e.g. - I haven't renamed any vswitch0 or anything). All the VM's will be up and running and perform great for weeks and week, then suddenly 1 will lose it's NIC, not disable, or lose connectivity, the actual NIC adapter will disappear?

Does anyone have any ideas of why this might happen?

Thanks!

[Duncan Epping]

First, check consistency of portgroup names, do these drops occur after a VMotion for instance?

I would secondly recommend to contact support. might just be a known issue with a driver for instance.

Duncan

VMware Communities User Moderator | VCP | VCDX

-

---

Now available: vSphere 4.0 Quick Start Guide: Shortcuts down the path of Virtualization

Blogging: | Twitter:

[HWMatt]

Esxi, no vmotion is there. No support.

Sent from my Windows Mobile phone

[danm66]

When you go into the Edit Settings for the VM is there no network adapter there or, is it that Windows doesn't see the hardware?

If it's the former, look in the vmware.log for information regarding the network adapter. If you have shutdown the VM, a new vmware.log will be created and the previous one rolled to vmware.log.1 or 2, etc...

If you want to attach it for the community to help you review, that may be quicker. Post the approximate date/time of the event, too.

If it's the latter case, I would look deeper into the OS side.

What are you using anything to back up these VM's and do you see it on only one host or both?

[HWMatt]

It's in the edit settings for the VM, not the OS, I haven't checked the logs yet. We're not using anything to back up the VM's, we're not concerned with data changes, we can just poor man clone a new VM as necessary, these are test boxes for developers.

[danm66]

Ok, just trying to think of any app that might be messing with VM files on the datastore, but seems strange that it would be specific to removing the network adapter...

Only kb article I could find that was similar was in relation to upgrading virtual hardware and it could cause the adapter to be seen in a different virtual PCI slot, forcing windows to see it as a new device.

There should be something in the vmware.log files if the configuration is getting changed. Post them if you can.

[jbown]

Hi There,

We are seeing this issue at one particular site. Single ESXi 4 + U1 (Build 219382) Host running on HP Proliant ML350-G5. The host has no fancy configuration, just a single NIC / vSwitch0 setup into a standard switch. The physical vmnic0 is using the bnx2 driver.

Three guests are running on the box, but it's always the same guest that looses it's NIC. Guest in question is Windows Server 2008 Std Edition (running as a Terminal Server). VMWare Tools are up to date on the box. Interestingly another Windows Server 2008 guest has no issues, and neither does a Windows 2003 guest.

We can "fix" the problem by simply hot adding a NIC back into the server without even shutting it down and it comes back on the network. Suspect it's some kind of bug as the other guests are not having issues. Any suggestions would be appreciated.

Regards,

Jeff

[gabe0819]

I'm having the same issue. Only one VM loses it's NIC randomly. We'll go months between incidents. I've attached the vmware.log

[HWMatt]

nice..the same thing is happening to me still... makes me feel better to know someone else has the same issue

[a_p_]

Take a look at the system tray's "Safely remove hardware" icon. Check whether or not your NIC is displayed as a removable device..

If somebody with the appropriate permissions decides to remove the NIC.....BINGO!

André

[jbown]

I can confirm this is exactly what was causing the problem in our case. We had a terminal server that requried administrative permissions for a few users in order to run a legacy application. Turns out one of these users was removing the NIC via the "Safely Remove Hardware" button in the system tray (why they would do that I am yet to work out).

I believe there is a setting in the VMX file for the VM that allows you to specify the NIC as non removable which makes this problem go away.

--Jeff