Server 2003 NLB

[Lee_K_81]

I'm having a bit of trouble getting Windows Server 2003 network load balancing working correctly.

 

At the moment it's running in a virtualised setup for testing / learning how; but will eventually be actual servers.

 

I've gone in to network load balancing manager and set up a new cluster called "tcservers" - ip x.x.x.1

and added 2 virtual hosts each with 2 network cards. I've selected Local Area Connection 2 for both, and given:

tc_server1 the ip x.x.x.100 (conn 2); priority = 1

tc_server2 the ip x.x.x.101 (conn 2); priority = 2

 

i've set a rule for the cluster of start: 3389; end:3389; Protocol: both; Mode: Multiple; Load: Equal; Affinity:None;

 

If i try and Remote Desktop to tcservers it puts me through to tc_server1, when it trys to send me to tc_server2 it times out. I've added entries in DNS for both.

 

On tc_server2 checking the local area connection, shows a lot of out-going traffic, but virtually nothing coming in (8 packets in 17hrs)

 

Any ideas?

[Lee_K_81]

Ok, bit of an update and a *bump* to see if anybody knows anything about Network Load Balancing today.

 

I'm having a bit of trouble getting Windows Server 2003 network load balancing working correctly.

 

At the moment it's running in a virtualised setup for testing / learning how; but will eventually be actual servers.

 

I've gone in to network load balancing manager and set up a new cluster called "tcservers" - ip x.x.x.1

 

Still same

 

and added 2 virtual hosts each with 2 network cards. I've selected Local Area Connection 2 for both, and given:

tc_server1 the ip x.x.x.100 (conn 2); priority = 1

tc_server2 the ip x.x.x.101 (conn 2); priority = 2

 

Ok, should have been Local Area Connection 1.

The NLB was set for unicast.

 

i've set a rule for the cluster of start: 3389; end:3389; Protocol: both; Mode: Multiple; Load: Equal; Affinity:None;

 

Also added rules for port 80; no affinity

port 443 single affinity.

 

If i try and Remote Desktop to tcservers it puts me through to tc_server1, when it trys to send me to tc_server2 it times out. I've added entries in DNS for both.

 

Ok, so when i put it into multicast, the load balancing seems to work to an extent. I've used the following guide to set up.

 

http://www.microsoft.com/technet/prodtechnol/windows2000serv/support/nlbcontb.mspx

 

The web requests seem to balance adequately, and both servers serve the pages. BUT and this is a big but as its the whole point of the load balancing, remote desktop does not work at all.

 

On tc_server2 checking the local area connection, shows a lot of out-going traffic, but virtually nothing coming in (8 packets in 17hrs)

This was because of unicast and is expected.

 

Any ideas? Or does anyone have any suggestions for alternatives. I think ideally, a linux box sitting infront of the servers polling them to see which is busier and sharing the load that way. Or am i clutching at straws.

 

Thanks in advance