Thick Client user with Clustered env

[#1]

We've recently set up a clustered env and changed our config file from Random to DocumentLoad. But when a user opens a document through the thick client, it goes to the server they request. So even if the document is loaded in memory on one qvs, if you specify the other, it will load on that server as well.

Any idea on why that would occure? When talking through this with some folks, I was thinking that pointing the Open In Server path to the master server would allow the server to define which qvs to use.

Using Access Point, it works great.

[danielrozental]

I believe the balancing in the AP is achieved through fixing the IP address in the links, QVS doesn't actually do anything, it's the AP that implements the balancing strategy.

When you use the QV Developer it doesn't go through the AP so no load balancing strategy is implemented.

[#3]

So does anyone out there run QV Developer (thick client) with a clustered environment?? If so, what is your approach to eliminate duplicate documents in memory?