I'm dealing with the following situation: I have 2 servers running Windows 2012 R2 and 1 server running Windows 2008 R2. I need to build a Microsoft Network LoadBalancer cluster using the 3 servers; servers are connected to a cisco Nexus Switch; servers are using Intel as well as Broadcom Nic cards. The servers are in the same subnet and the Network LoadBalancer VIP of the newly formed cluster is also in the same subnet. As long as wireshark is running on the servers the cluster is working (setting the nic interfaces to promiscuous mode) - members are detecting / seeing each other and passing each other the Primary / Master role when configured via the Network Load Balancer UI. How can this be possible?

Another strange thing: if I take one of the member offline from the cluster and run a Wireshark capture, I can see the Microsoft Network Load Balancer heartbeat broadcast messages being received from that host; nevertheless, as soon as I add the host again in the cluster, and I restart Wireshark, then the Network Load balancer cluster is working but in the Wireshark capture, I don't see any Network Load Balancer hearthbeat messages received any longer.

Moreover, each time I take any host offline from the Network Load Balancer cluster or add it to the Network Load Balancer cluster, Wireshark stops the formerly ongoing capture and displays a notification message stating that the ongoing capture cannot continue as the NIC formerly in use is no longer available.

  • 2
    I can see how this question seems like it belongs here, but I'm almost positive it really belongs on serverfault.com Jun 25 '15 at 17:18
  • Hi, I'm not so sure if the issue is more server side or network side ... but, for a start, I would like to better understand the behavior seen using wireshark on the server ...
    – Ottootto
    Jun 25 '15 at 17:22
  • 2
    rule number one, if your server is misbehaving, don't run wireshark on the server... span a port on the switch in case there are driver problems with server Jun 29 '15 at 0:05

Are the servers virtualized? If VMWare, then there are specific considerations for configuring monitoring, search VMWare's documentation site for packet capture on vSwitch.

Here's a link to a Cisco document on configuring NLB for Cisco 6500 series and 4948 switches. http://www.cisco.com/c/en/us/support/docs/switches/catalyst-6500-series-switches/107995-configure-nlb-00.html?mdfid=270638922

I found this in the release notes for 9000 series running 7.0(3)l1(1), dated 01Feb2015 at http://www.cisco.com/c/en/us/td/docs/switches/datacenter/nexus9000/sw/6-x/release/notes/70311_nxos_rn.html, under "Limitations":

"Cisco NX-OS Release 6.1(2)I2(2b) supports flooding for Microsoft Network Load Balancing (NLB) unicast mode on Cisco Nexus 9500 Series switches but not on Cisco Nexus 9300 Series switches. NLB is not supported in max-host system routing mode. NLB multicast mode is not supported on Cisco Nexus 9500 or 9300 Series switches."

Are the switches 9300s?

VMWare has instructions for implementing MS NLB in unicast mode: http://www.vmware.com/files/pdf/implmenting_ms_network_load_balancing.pdf

Microsoft's MSDN web page on the topic: https://msdn.microsoft.com/en-us/library/bb742455.aspx?f=255&MSPPError=-2147217396

Microsoft TechNet page about this topic https://technet.microsoft.com/en-us/library/cc732592(WS.10).aspx#BKMK_B

HTH, Jason

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