Well its 2 minutes later and I found a problem - the wierd error happened because I was writing continuously... not a good thing. If I have the datasocket cluster read operating continuously and then just execute the write now and then - it appears to work just fine.
As for the wierd error - I can't say its very descriptive but maybe there was just too much information being passed to the server to quickly, I don't know.
(by the way, both the read and write vis are running on the same machine)