How many Cows are you running? One thing I've found in my work is that they are vulnerable to packet collision, and you can run out of physical ram if you are starting several Cows simultaneously, and you have large textures. If you have the latest update, then that eliminates the handle issue that the update twice removed introduced. Try starting a render with 1 Cow, wait until all the frame and texture data has been sent, and it is rendering frame 1, before starting the next Cow. If this is a resource issue, that -may- solve it. If it is a network issue, having the operating times of the Cows offset from each other should take care of it.