TS Session Directory and NLB

[Guest Shane Franchina]

Hi,

 

I have 5 terminal servers running Windows Server 2003 Enterprise Edition R2

SP2. They are running NLB and together form a NLB cluster. We also have a

terminal services session directory running on another server in the

organisation and all servers are a member of this session directory.

 

Each terminal server has 2 network interfaces, one for the clients to

connect to (and the one that NLB is enabled on), and the other one is a

management interface.

 

The issue that we are having is with the configuration of one particular

terminal server. In it's session directory configuration screen, there is no

option to select the correct interface for redirection. The interface simply

is not listed! On all other servers, there are two NICs listed (and we are

able to select the correct interface for redirection), however, on this one

particular server, it is not there. Disabling this management interface

causes the list to go totally blank with no interfaces listed for redirection

at all.

 

When we disable NLB, the interface appears in the list. It's almost as if it

will not allow us to redirect to an interface that has NLB enabled on it,

even though this works fine on all 4 other servers.

 

The problem can be temporarily resolved by manually editing the registry

(HKLM > SYSTEM > CurrentControlSet > Control > Terminal Server >

ClusterSettings and changing the "SessionDirectoryRedirectionIP" key to the

correct IP - in this case, 10.2.56.5). This works. However, upon enabling the

management interface, this key *automatically* changes back to the management

interface's IP address!

 

Basically what this means is that when a session is disconnected on this

particular terminal server, and a client goes to reconnect to it (from

another server), they are redirected to the Management interface of this

terminal server which the client does not have access to (it's on a different

subnet, and locked down to prevent unauthorized access).

 

I have taken the following steps to try and rectify this problem:

 

1. Deleted the cluster totally and recreated it - no difference.

2. Uninstalled and reinstalled terminal services on affected TS machine - no

difference.

3. Updated the firmware and drivers for the NICs in the affected server - no

difference.

4. Formatted the affected server and reinstalled all patches, service packs,

drivers etc - no difference!

5. Manually editing the registry as stated above, but this does not

permamently solve the problem - enabling the management interface causes it

to change back by itself.

6. Tried swapping the interfaces around, reconfiguring the IPs, NLB etc - no

difference.

 

The servers are all HP DL380 G5s. They have identical hardware and software

configurations - same amount of RAM, hard drive, network cards etc. The

affected server has had it's mainboard replaced once by HP due to DOA but

there was no operating system installed on it at this time. To us it seems

unlikely that hardware could be a problem (although indeed the replaced

mainboard is the only difference between the terminal servers), but as

stated previously we have updated all the drivers and firmware on the system

to no avail.

 

Has anybody ever come across a problem like this one.

 

Any help you may be able to provide on this would be very much appreciated.

 

Regards

 

Shane Franchina