Load Balance Node communication Error

  • Thread starter Thread starter Guest
  • Start date Start date
G

Guest

I have two Advanced servers in a Load balance team, they
have to run in Unicast mode because of the way the
Network is setup.

The servers have a problem talking to each other and
browsing the network. I have put them through their own
switch (level 2) which solved the problem until they are
rebooted.

When they have been rebooted the servers can no longer
talk to each other, and it takes a lot of fiddling around
putting the network cables into different switches etc,
before they will work properly again.

Can anyone help??
 
First of all, how many NICs are in each server? I am assuming just one. When
you say that the servers have problems "browsing the network," what exactly
do you mean - are you initiating communication FROM one of the cluster hosts
to a non-cluster host? If so, you must ensure that the traffic is sourced
from the Dedicated IP address of the node. To do so, first make sure that
you have configured each host in the cluster with a Dedicated IP address in
the NLB properties. Then, in the TCP/IP properties, make sure that the
Dedicated IP address is listed FIRST in the list of IP addresses (hit the
"Advanced" button in the TCP/IP properties UI to see the full list of IP
addresses). If the Dedicated IP address is listed first in the IP address
list, that is that address that will generally be used by default for
out-going traffic.

Now, as for cluster hosts being able to talk to each other, that is
unsupported in a single NIC configuration in unicast mode. If you must use
unicast mode, you will need to have a second NIC in each host that you can
use for intra-cluster communication. The reason that a second NIC is
required is that in unicast mode, all NLB hosts have the same MAC address -
the cluster MAC address - which precludes TCP/IP from distinguishing the
hosts at the link layer.

Hope this helps.

Cheers, Sean
 
Back
Top