I have two virtual machines on VMWare ESXi host. I want to drop an external IP (used for a site in IIS that we're moving to a new server) from the old one (Windows Server 2008 x64), and assign to the new (Windows Server 2008 Web R2 x64). In the past, this is easily accomplished by removing the IP from the old, shutting it down, assigning it to the new, and re-booting it. However, I've not had any luck with this configuration and I don't know if it's a Win2008 issue, or VMWare ESXi issue, or both:

  • vSphere Client indicates that IP address is correctly on the new server (and not on old).
  • New server has IP address
  • Bindings in IIS are correct
  • All machines in question have been rebooted.

But still:

  • Any host header pointed to the IP address in question times out.
  • Any host header pointed to a different external IP on the new server resolves without issue.
  • On the same network (from another server), host headers pointing to IP in question resolves fine!

I'm baffled! It's like there's a caching issue, but DNS caching is not the issue, configuration is not the issue (as works to a different IP or from a server on the same network). All said and done, a request from the outside still looks like it's attempting to contact the old server, even though the IP address is the same, it's just now on a different server (and the old one is turned off).

I would say it's the network configuration, or something, but I have had no problems in the past on the same network dropping an IP address from a Win2003 machine, and assigning to another Win2003 machine. This time around, the unknown variables are that I've never done it before with Win2008 and/or VMWare ESXi hosted virtuals.

  • Possibly resolved as after about half an hour, all domains resolved properly. Am now thinking it is a firewall caching issue (it's verified not an ARP issue). Therefore, will try again with another site and see if run into the same issue. What's weird is that same firewall, same network, same everything except no VMWare and Win2003, worked without delay in the past. – Ted Oct 14 '09 at 23:53

Seems much more likely to be an ARP issue to me - I'd be looking at ARP issues on whatever switch(es) your ESXi host is connected to first - especially if the uplink ports from the vSwitch\Port Groups VM's are likely to be different. If you are using NIC Teaming on relevant Virtual Switches on the ESXi host then you should also make sure that your switches are configured to support the same variant of 802.3ad as your ESXi version - this may no longer be an issue with ESX/i4 but for V3.5 VMware only supported static, not dynamic, 802.3ad/LACP.

One other thing you can do is see if something has gotten messed up with the MAC addresses assigned to the VM's virtual Nics - see this VMware Kb article for some info regarding how you can manually override the automatically assigned MAC addresses. This would only be a problem if the two VM's ended up with the same MAC address though, and I've no idea how that would happen if they hadn't been manually tampered with in the first place.

  • Helvick - thx for the very thorough answer. We know it's not likely the latter, but we're checking into the former possibility, particularly whether it could be ARP caching in the virtual switch in VMWare. Will update here if it is. – Ted Oct 19 '09 at 18:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.