Restored 2008 VM loses network adapter, resets IP to DHCP

[chris.trotter]

Has anyone had the following symptoms when restoring Windows Server 2008 R2 VMs?

- Restore is 100% successful.
- Change the vCPU to 1, RAM to 1GB, network label to our 'sandbox'. (did the same for the 2003/2008 VMs)
- When VM boots up, the 'computer did not shut down properly' screen comes up (i.e. choose between normal or safe mood boot).
- Network adapter name is showing as 'Local area connection 2'.
- Network settings are set to DHCP, all previous static entries gone (because it's a new adapter).
- Setting the new adapter to the old settings brings up a message: Another old adapter has these settings, clear them? (not verbatim)

I have restored multiple 2008 R2 VMs now, and every single one has this issue. When I restore Server 2003 or 2008, the adapter does not disappear, and the old settings are still present.

Any suggestions would be appreciated!

[Vitaliy S.]

Hi Chris,

Dirty power off message is expected, you may find more info on that in the following topic: restore image question

As for the new network connection, could this issue be connected to a missing VM network at the ESX(i) host where you're restoring to? If you're restoring to a different VM network, I believe this could be an expected behavior.

I've just done the smoke test following your scenario above, and restored Windows 2008 R2 with the same network settings as it previously were.

Hope it helps!

[chris.trotter]

Vitaliy, could this be caused by a different network label? I am restoring to a network that is completely cut-off from our production network and is on a different vswitch/network label. These are just test restores, I cannot really test restoring to a box that has the same production settings without breaking things.

[TMassa]

Every restore we've done (about 30 in the last week and a half) has been fine where the VM NICs are concerned, we just edit the VM settings to place the original NIC on our sandbox VLAN and start up the recovered VM. The original NIC is was still there with the static IP settings that were configured during the backup. Does the OS actually detect and install new hardware when you startup? We're all just 2008, no R2

[chris.trotter]

Yes, it detects new hardware on startup - it sees it as a new piece of hardware. The 2008 boxes had no issues at all, this is ONLY for 2008 R2.

[Vitaliy S.]

Chris,

Yes, that's exactly what I've thought of. Additionally, if you're using completely different physical NIC on the ESX server you're restoring to, I believe that might also cause the creation of new network connection in Windows 2008 R2. I would try to backup a test VM with Windows 2008 R2 installed and then restore to it's original location, to see if this behavior confirms or not.

[chris.trotter]

Gotcha, will do. I'll report back later.

[chris.trotter]

Yes, confirmed this - if you change the vswitch and network label, that's what destroys the adapter. This is NOT a Veeam problem. Case closed!

[Vitaliy S.]

Chris, thank you for updating the topic! Glad that we've figured everything out.

[RumataRus]

If I restore a VM from a replica or by "Instant Recovery" I lose original IP-address in the recovered VM. Why? How to keep it?

[Vitaliy S.]

I guess you've been using VMXNET3 network adapter on your Windows 2008R2 VMs, am I right?

If this is the case, then please have a look at this VMware KB article: http://kb.vmware.com/kb/1020078

Please note that this VMware issue relates to Windows 2008 R2 or Windows 7 VMs only, other VMs shouldn't be affected by this behaviour.

[RumataRus]

I guess you've been using VMXNET3 network adapter on your Windows 2008R2 VMs, am I right?

Yes! You are right as usual.

Please note that this VMware issue relates to Windows 2008 R2 or Windows 7 VMs only, other VMs shouldn't be affected by this behaviour.

I use Windows 2008 R2 and I will try to apply the Microsoft hotfix for this issue.
Thank You!

[Gostev]

VMware KB article about this issue:
http://kb.vmware.com/kb/1020078

[cag]

I am see a problem testing a replica at our DR site. This may not be a Veeam issue, but how are people testing DR replica's.
What I did is use the vshpere client to start the VM. We are not using any re-ip etc, just starting the VM.
The VM is a 2008R2 Server.
This is a problem I am seeing. When the VM starts up it see's the network cards as new hardware and is using DHCP for startup.
I thought it would be static as the orginal source server. Is this normal ? for 2008 my windows XP workstations, linux servers etc work as expected.
This seems to be a Windows Problem, but would like to know how people solve this for testing.

[samuk]

Hi,

Unsure if this is a VMware or a Veeam issue..

We have a Exchange 2010 server running on Windows 2008 R2 using VMXnet3 Nic.

We are testing the backup \ restores, and when restoring the server we have found that when booting up the Nic is disabled, servers hung.

If i edit setting disconnect and reconnect the nic, it works however all network settings are lost.

On another test - i found if i left the server for say 10mins - i wouldl see a popup saying installing vmxnet3 driver - nice would come online however all ip info is lost...(this is set static in the original server)

The server is part of a Exchange Cluster.


Any one got any idea's


Thanks

[AndyGDIT]

Looking to see if anyone else is having issues with restoring Full VMs (VMWare) to their enviroment. I have done two restores over the past couple of days where the IP configuration does not come with the restore and it defaults to DHCP. This is an issue becasue we also use RSA on our servers and that messes that up. I have had to use Safe Mode and other tools to get into get into the server and get the right IP entered on the NIC.

I havent opened a case on this yet, but I would like to see if anyone else is having this issue.

I am on the latest version of 6.1 (6.1.0.204)

[cgrebing]

Hi,

we had this problem too. There is an issue with the vmxnet3 driver within W2K8 R2 as Vitaly always mentioned. You should install the MS Patch before you backup the VM and then the NIC does not appear as new.
It is the same problem if you clone a VM within vCenter.

[mchang]

I'm wondering if anyone else has seen this issue or would be willing to verify/confirm what I'm seeing.

Up until a week ago, I was running a stable Veeam 6.5 + vSphere 4.1 environment for a couple of years. Backups/replication/restores of all kinds were working great. After my upgrade to 7.0 last week (in-place upgrade), I've been able to validate all functionality except for restoring an entire VM (to same VM and storage environment, but different name/network/datastore/etc). The restore job completes successfully and the machine boots to desktop. At first glance everything looks to be working, but the machine cannot communicate on the network. Closer examination of the IP settings on the machine shows that something is very wrong with the network adapter settings.

In this particular example (I have reproduced this on 3 different VM's), my machine is:
Windows Server 2008 R2 Std machine with an Intel E1000 adapter and a static IP of x.x.x.47. On the restored version of the machine, the IP configuration shows it has IP of "x.x.x.47 (duplicate)". Duplicate right? Strange...on top of that it also has the Windows Autoconfiguration IPv4 address of "169.254.149.139 (preferred)". In my experience, the only time you should see the autoconfiguration IP in Windows is when you are set for DHCP, but there is no DHCP server available. In this case my server has a static IP so it shouldn't being using the autofiguration in the first place and why is the static IP suddenly seen as being a duplicate? Needless to say, my restored machine cannot communicate with anything on the network in this state. Obviously something has munged the IP stack...

Things I've tried to no avail:
- Remove the IP address, set to DHCP, re-add IP - no effect
- Try a different IP from the same subnet - no matter which IP I try Windows says there is an IP conflict
- Run "netsh int ip reset c:\resetlog.txt" to reset the IP settings - no effect
- Try an IP from a different subnet - this works!
- Also, a Windows 2003 machine seems to be unaffected by this issue

I haven't tried removing the adapter and re-adding yet, but at this point that would be an unacceptable workaround. In the event of some sort of disaster, it's obviously not feasible to manually touch every machine to get it to work.

I've got a ticket open with support (Case # 00529676), but thought I'd run this by you guys. Some of you may be having this problem, but just don't realize it yet so it would be great if you could test it and report your results. Thanks!

[Gostev]

As I like to say, 99.999% of "upgrading to version X broke feature Y" type of cases have nothing to deal with the upgrade

This is very old and well known Microsoft/VMware interop issue affecting Windows 2008 and Windows 7 virtual machines with VMXNET3 vNIC.

Here is the VMware KB article about this issue:
http://kb.vmware.com/kb/1020078