4

I have a troublesome Windows Server 2008 R2 VM in a vSphere 5.5 environment which is starting up with an APIPA address (169.254.0.0/16) on its interface instead of the statically preconfigured IPv4 address. Disabling and re-enabling the interface (either through devmgmt.msc or ncpa.cpl) helps with restoring connectivity, just as does disconnecting / reconnecting the virtual adapter in vSphere:

connect / disconnect

History so far:

  • Guest has been running on vSphere 5.0 / 5.1 in the last years
  • vSphere environment has been upgraded to 5.5 2 months ago
  • Guest has been migrated to another cluster last week
  • VMWare tools have been upgraded this week

It is not restarted all that often, so it is difficult to tell when the issue actually started showing, but we noticed it this week after the tools upgrade. The first restart after Tools installation had no connectivity, subsequent restarts did not change anything in this regard. Other guests with a similar history indeed did have trouble with connectivity on the first restart after the tools upgrade, but subsequent restarts were okay.

The virtual machine is v7 and has a single E1000-type network interface, the guest's drivers are Windows' stock PRO/1000MT (E1G6032E 8.4.1.0). Tools version is 9.4.5 build-1734305.

The Windows event log shows nothing unusual or driver-related on startup apart from entries of starting services indicating that the host has no IP connectivity. The VM's "Tasks & Events" tab of the vSphere console has no suspicious entries either.

Is this really related to the Tools upgrade? How would I fix it?

Edit: I am aware of vmWare KB 2012646, it basically suggests connecting / disconnecting the adapter and checking the registry for APIPA configurations on interfaces. I have done both (no APIPA-configured entries are present in the registry on the machines in question).

the-wabbit
  • 40,737
  • 13
  • 111
  • 174
  • Just wondering since you have vmtools installed and upgraded - would adding a new VMXNET3 nic in place of the E1000 (with same IP reapplied of course) give the same issues? – afx May 07 '14 at 09:45
  • @afx it might resolve or work around the problem, the trouble however is that the network adapter has an automatically assigned MAC address and the server is running a piece of software bound to the MAC address. While we could simply ask for having the VMX file edited to have this very MAC address be set as "Manual", it would break some of our procedures, so we'd prefer not to if another resolution is available. – the-wabbit May 07 '14 at 09:56
  • ah righto - thought as much. keeping head on the grindstone... – afx May 07 '14 at 10:05
  • Please elaborate a bit on the network setup - how are your virtual switches set up, do you use VLANS? If so, how are the set up on your virtual switches? have you checked each port of the virtual switch on your new cluster to ensure every specified virtual network's VLAN has been tagged on each interface? – Alex Berry May 07 '14 at 12:41
  • 2
    @AlexBerry My insight into the configuration details of the vSphere networking is limited - I only get to manage the VMs once they are configured. Of course, the switches do use VLANs and I would not expect the problem to be on the networking side however - once the interface has been disabled and re-enabled in the guest, it works flawlessly. But I will try and get the vSphere configuration details from the respective service owners. – the-wabbit May 07 '14 at 13:56
  • [Here](http://hardforum.com/showthread.php?t=1757236) are users with similar issues. I recommend cases with VMware and Microsoft for this, it does not seem to be something fixable/diagnosable by serverfault to be honest. (If you can't switch NIC) – MichelZ May 22 '14 at 08:57
  • I have switched NICs in my environment to vmxnet3 and continue to have the same problem. I've tried all the usual stuff like updating vmware tools and cleaning up the network configs. No VLAN tagging (two untagged VLANs on separate physical interfaces), and the setup is very basic. Several people have had this problem in strange situations but I've never found a straightforward solution online. – Todd Wilcox Sep 04 '14 at 17:12
  • @ToddWilcox do you have this reproducible? If so, could you try [disabling the power management for the network interface in the device manager](http://i.stack.imgur.com/vlKbm.png)? It seemed to help, although I do not know for sure due to the small sample size of restarts of the affected machines. And as they are productive, I cannot schedule an awful lot for testing. – the-wabbit Sep 04 '14 at 17:22
  • @the-wabbit Sorry, power management already disabled. – Todd Wilcox Sep 04 '14 at 17:24

3 Answers3

1

I have changed the power management settings for the NIC in the guest OS

power management settings

and recent restarts went without networking issues. I do not know for sure if it is causality or just coincidence, though - other things (like the vSphere patch level) might have changed in the meantime without me knowing.

the-wabbit
  • 40,737
  • 13
  • 111
  • 174
1

Check this solution it has worked for me-

Type below command in cmd of your vm-

netsh interface ipv4 show inter

You will get output like this- Cmd output

remember IDX- 11 which is assigned to your lan connection run below command-

netsh interface ipv4 set interface 11 dadtransmits=0 store=persistent

exit from CMD Go to Services.msc disable DHCP Client service

Restart the machine and check

Vino
  • 119
  • 8
0

Try "netsh int ip reset" to reset the IP stack, re-assign the static, reboot and see if it sticks?

Clayton
  • 4,523
  • 17
  • 24