I've no experience of this on Hyper-V, but I did have a possibly similar
issue using VMWare recently. The issue was that Multicast mode requires
routers that can cope with two multiple MAC addresses per port (or
something..?). We had to get the network guys to configure static ARP entries
across the routing infrastructure in order for the cluster address to be
reachable outside of the host.
It may be worth you investigating whether your issue is something like this,
rather than something related to Hyper-V itself? Also you may want to look at
adding a second NIC to the guests, and then using Unicast on the cluster NICs?
Hope this helps ;-)
"John" wrote:
> Hi,
>
> I have configured 2 nodes NLB (multi-cast) using Hyper-V guest Windows Server 2003 and I could ping the cluster ip address from hyper-V host or any other virtual machines hosted by this hyper-v host but NOT from any other machines outside hyper-v (they are all in the same subnet). Can someone help me to troubleshoot this issue?
>
> Here are the configuration:
> Hyper-V host:
> Adapter 1:
> 10.118.100.226/16 -- hyper-v virtual network
> Adapter 2:
> 10.118.100.228/16 - used for hyper-v mgmt
> Adapter 3:
> 10.117.100.8/16 -hyper-v virtual network
>
> Guest VM1:
> NIC1: 10.118.100.231/16 + cluster IP: 10.118.100.234/16
> NIC2: 10.117.100.9/16
>
> Guest VM2:
> NIC1: 10.118.100.232/16 + cluster IP: 10.118.100.234/16
> NIC2: 10.117.100.10/16
>
> NLB configuration:
> operation mode = Multicast
> Port rules = Cluster IP Address = All, Start = 0, End = 65535, Protocol = Both, Mode = Multiple, Affinity = single
> converged successfully ... cluster ip can be accessed from hyper-v host or any other virtual machines (other than VM1 and VM2)
>
> From machine outside hyper-v host,
> > ping 10.118.100.226 OK
> > ping 10.118.100.231 OK
> > ping 10.118.100.232 OK
> > ping 10.118.100.234 Timeout --- cluster IP
>
> I saw most of the post on the net is using unicast with Hyper-V, Is multicast NLB mode supported with Hyper-V guest VMs?
>
> Thanks very much!
> John
>
>
>
|