Configure NLB on W2K3 EE

Discussion:

Configure NLB on W2K3 EE - what wrong?

(too old to reply)

Michal Ch

2005-11-10 20:00:26 UTC

Hi all
I've 2 server Dell PowerEgde 1855 connected to PowerConnect 5324
Each server have 2 NIC with Teaming (Adaptive Load Balancing)
First server have IP 192.168.1.22 on teaming card (virtual card)
Second server have IP 192.168.1.23 on teaming card (virtual card)

I would like to create NLB (cluster IP ) - 192.168.1.10
If I create NLB with Unicast mode working very slow - don't connect every
time.

If I create NLB with Multicast mode I have strange problem.
In Affinity - single
From network 192.168.1.0/24 ping on IP 192.168.1.10 and 1.22 and 1.23
working without problems.
But when connect with mstsc.exe to 192.168.1.10 I always connect to server
192.168.1.22
But I shutdown server 192.168.1.22 I couldn't connect to TSfarm and only
192.168.1.23 respond.

From network 192.168.0.0/24 ping on 192.168.1.22 and 1.23 working without
problems,but doesn't work 192.168.1.10
But when connect with mstsc.exe to 192.168.1.10 I couldn't connect to server
192.168.1.22
But I shutdown server 192.168.1.22 I couldn't connect to TSfarm and only
192.168.1.23 respond.

In Affinity - none
From network 192.168.1.0/24 ping on IP 192.168.1.10 and 1.22 and 1.23
working without problems.
But when connect with mstsc.exe to 192.168.1.10 I always connect to server
192.168.1.22
But I shutdown server 192.168.1.22 I couldn't connect to TSfarm and only
192.168.1.23 respond.

From network 192.168.0.0/24 ping on 192.168.1.10 and 1.23 working without
problems,but doesn't work 192.168.1.22. Bit sometime ping on 192.168.1.22 is
working
But when connect with mstsc.exe to 192.168.1.10 I always connect to server
192.168.1.23
But I shutdown server 192.168.1.23 I couldn't connect to TSfarm

Could you tell me what is work with my configuration?

Have a nice day,
Michal

Mark Bradley

2005-11-11 00:56:03 UTC

Permalink

Not sure if it helps but I have always had the same problem with NLB when
connecting to a cluster from a client that has had an existing connection.
The session will always connect to the same node.

If it is a new connection from the same session it will connect to the same
node even if the node is offline.

What I have found is that if the cluster has both nodes online then it will
load balance different client connections to it but has no intelligence if a
server is down.

I may be wrong but I don't think it can allow fault tolerance for clients
already connected.

Post by Michal Ch
Hi all
I've 2 server Dell PowerEgde 1855 connected to PowerConnect 5324
Each server have 2 NIC with Teaming (Adaptive Load Balancing)
First server have IP 192.168.1.22 on teaming card (virtual card)
Second server have IP 192.168.1.23 on teaming card (virtual card)
I would like to create NLB (cluster IP ) - 192.168.1.10
If I create NLB with Unicast mode working very slow - don't connect every
time.
If I create NLB with Multicast mode I have strange problem.
In Affinity - single
From network 192.168.1.0/24 ping on IP 192.168.1.10 and 1.22 and 1.23
working without problems.
But when connect with mstsc.exe to 192.168.1.10 I always connect to server
192.168.1.22
But I shutdown server 192.168.1.22 I couldn't connect to TSfarm and only
192.168.1.23 respond.
From network 192.168.0.0/24 ping on 192.168.1.22 and 1.23 working without
problems,but doesn't work 192.168.1.10
But when connect with mstsc.exe to 192.168.1.10 I couldn't connect to server
192.168.1.22
But I shutdown server 192.168.1.22 I couldn't connect to TSfarm and only
192.168.1.23 respond.
In Affinity - none
From network 192.168.1.0/24 ping on IP 192.168.1.10 and 1.22 and 1.23
working without problems.
But when connect with mstsc.exe to 192.168.1.10 I always connect to server
192.168.1.22
But I shutdown server 192.168.1.22 I couldn't connect to TSfarm and only
192.168.1.23 respond.
From network 192.168.0.0/24 ping on 192.168.1.10 and 1.23 working without
problems,but doesn't work 192.168.1.22. Bit sometime ping on 192.168.1.22 is
working
But when connect with mstsc.exe to 192.168.1.10 I always connect to server
192.168.1.23
But I shutdown server 192.168.1.23 I couldn't connect to TSfarm
Could you tell me what is work with my configuration?
Have a nice day,
Michal

Entei

2005-11-11 22:03:21 UTC

Permalink

Hi Mike,
I am pretty new with NBL. I like to setup a print cluster for 2 nodes.
Each has 2 NICs. Could you tell me how to connect private network.
Should I connect these 2 private NICs to a sperate hub/switch? I am
confused. If you have any other info how to setup print cluster,
please let me know.

Thanks,

TL

Post by Mark Bradley
Not sure if it helps but I have always had the same problem with NLB when
connecting to a cluster from a client that has had an existing connection.
The session will always connect to the same node.
If it is a new connection from the same session it will connect to the same
node even if the node is offline.
What I have found is that if the cluster has both nodes online then it will
load balance different client connections to it but has no intelligence if a
server is down.
I may be wrong but I don't think it can allow fault tolerance for clients
already connected.

Mark Bradley

2005-11-12 20:40:03 UTC

Permalink

You should use a separate network for the private network either using
separate switches or if close enough a crossover cable will work just as well
with less things to fail.

Cheers

Mark

Post by Entei
Hi Mike,
I am pretty new with NBL. I like to setup a print cluster for 2 nodes.
Each has 2 NICs. Could you tell me how to connect private network.
Should I connect these 2 private NICs to a sperate hub/switch? I am
confused. If you have any other info how to setup print cluster,
please let me know.
Thanks,
TL