Remote Desktop Connection Broker R2

Hey Pete, I too had problems getting my 20008 R2 terminal servers working through the Big IP. R2 changed a lot of things, and F5 has still not updated their deployment guides for R2 yet. When I called in with my issue, we had to change numerous things from the deployment guide just to get the web access pat to work. Still coudln't get remote app work, so I gave up and went a different route. Hopefully somebody will see this and finally get their engineers to start working on R2 deployment guides/solutions.

Hi carter91,

 

Thanks for your reply. I may well have a look at using microsoft's built in load balancing, however there are downsides to that from a management point of view. Can I ask what route you took?

Sorry for my late reply, dont' get on here very much.

 

 

If you still need the info, we just went native Microsoft load balancing. It works quite well.

 

 

F5 STILL has yet to update the informaiton for R2 remote desktop services. They need to get on teh ball. I know it presnted a lot of changes, but this type of service is pretty popular that people want to put behind their F5s.

A new Deployment Guide for Windows Server 2008 R2 Remote Desktop Services should be online in the next day or so.

 

 

In the meantime, to avoid the error message you're seeing:

 

 

1) make sure that the DNS name (FQDN) of the RD Session Host virtual server -- the farm name, in other words -- is correctly resolvable from the RD Web Access server(s) where you're configuring the RemoteApp applications. This can be done via DNS or a local hosts file.

 

2) Using the same IP address as that which you've used for the RD Session Host servers, create another virtual server that listens on TCP port 135. That virtual should have a pool resource whose members are the RD Session Host farm servers, also on TCP port 135. In most topologies, you'll want to enable SNAT Automap (or use a SNAT pool) for this virtual server.

 

3) That virtual server IP address must be routable from the RD Web Access servers. (You state that the Connection Broker and Session Host servers are on the same subnet, but not where your Web Access servers are, or if the virtual server addresses are on the local subnet or one that is reached by traversing the LTM).

 

 

When you configure RemoteApp applications, the RD Web Access servers try to contact the farm via RPC. If using Microsoft NLB, presumably this part of the configuration is taken care of automatically. With a third-party device like F5 LTM, you have to set up the name resolution and traffic flow manually.

 

 

As far as RD Connection Broker goes, it's not much different than TS Session Broker. What problems are you seeing? In all our testing, configuration and behavior of persistence on BIG-IP LTM was identical whether provided by TS Session Broker or RD Connection Broker.

Many thanks for your advice. I look forward to reading through the guide and trying out your advice above over the next few days.

Thank you. I have just given this a quick try on one test farm and it works perfectly.

 

 

To answer you questions we had all servers involved in the same subnet so routing was never really an issue. As to what problems we have had, the publishing of the applications using the virtual server address to the web access server. Everything else seems to work fine using the pre R2 guide.

The new RDS Deployment Guide (which also covers the steps listed above) is now online at http://www.f5.com/pdf/deployment-guides/f5-microsoft-remote-desktop-services-dg.pdf

 

I'm also having problems getting Server 2008 R2 RDS running via the F5.

 

We are only using the first scenario from the Deployment Guide having only a Connection Broker (no Web Access or Gateway Servers). I just followed the steps of the Deployment Guide, but the main problem we currently have is that we don't get any entry in the persistence table.

 

Any idea what's the reason for this or how I can further troubleshot this (e.g. tcpdump)?

 

Thank you!

 

 

Ciao Stefan :)

 

I might have an idea what's my problem. If I disconnect my RDP-session and try to reconnect then I first might end up on a different server, but then this server sends back the session token from the Connection Broker to the client.

 

With tcpdump I see the Cookie (e.g. msts=254347274.15629.0000) but for whatever reason the BIG-IP is not recognizing this and making the required persistence table entry.

 

Or is the RDP-persistence profile working without entries in the persistence table at all?

 

On thing came in my mind right now, is this maybe a problem when working with Route Domains as the "%RD-ID" is missing. Or can the BIG-IP handle this correctly?

 

Thank you!

 

 

Ciao Stefan :)

 

Hi Pete Cranny,

 

Could you share withnus how do you fix the error.

 

I have the same when trying to add RDSH farm as remoteAPP source on the broker server.