I have a 2 node NLB (for ARR) cluster running. Both machines run Server 2012R2 and are guests on VMWare ESXi Hosts and can be moved between them as needed.

They are set up using IGMP multicast, the Cisco switch they all connect to is set up to accept this. There is a static ARP entry on the firewall/router for this network.

I can use NLB and view sites etc fine for a while, but randomly it will stop working. I open NLB manager and I can only see the local host. This is the same on both hosts.

When I run the scripts suggested in :

https://social.technet.microsoft.com/Forums/windowsserver/en-US/9262c47a-cfe1-4696-9442-dd8d82651891/nlb-manager-donot-show-the-other-host?forum=winserverClustering

I cannot see the other node.

What can I do to resolve this. Both hosts are multihomed, the NLB adapter is in the DMZ and there is another management network adapter. NLB's adapter is not publishing itself to DNS.

The firewall on the NLB Nodes is disabled. There are no obvious entries into event viewer - can anyone suggest some event IDs to look for?

Thanks

Hi,

Have changed anything recently?

Could you please change the mode to unicast? If it works, please check the multicast settings of the network devices.

Besides, here is an article about how to troubleshoot NLB, it may be helpful:

http://social.technet.microsoft.com/wiki/contents/articles/7329.network-load-balancing-nlb-survival-guide.aspx#VMWare

Best Regards.

There have been no changes, however the VMs could have moved between hosts.

We do not want to use unicast as this is sat on our live network and cannot afford to flood the network with unicast packets.

Thanks for the links.

There have been no changes, however the VMs could have moved between hosts.

We do not want to use unicast as this is sat on our live network and cannot afford to flood the network with unicast packets.

Thanks for the links.

Maybe I know what caused your issue. To be sure I have to ask you two questions:

1. Do you have multiple physical switches interconnected with each other?
2. If so, could it be that one VM moved between hosts and because of that connected via another seperate physical switch?

If that is the case than you HAVE TO configure your router (e.g. L3 switch) as an IGMP Querier. The exact problem and solution is described in the following Cisco white paper:

Multicast Does Not Work in the Same VLAN in Catalyst Switches
http://www.cisco.com/c/en/us/support/docs/switches/catalyst-6500-series-switches/68131-cat-multicast-prob.ht