we've got a very serious network rendering problem: sometimes during a
network rendering task some computers in our render farm won't render any
images but sending the manager a image ready message that causes the manager
to log that frame as rendered and giving that render machine the next frame.
The fatal thing is that these crazy computer finshes one frame after the
other in only a few seconds and so the hole renderjob will be killed by that
machine. All other computers can only render their first submitted frame and
after finishing that they couldn't get a further frame to render. So the
render job is destroyed by this one computer in the farm (it isn't everytime
the same machine). Do anyone now this problem? Finally I must say that this
problem ocures only with big jobs where the ram usage is very high over the
physically installed amount and we are having this problem since MAX 3.0 but
3.1 makes it not better.