#Net Rendering farm
We just got a new farm setup with 5 P5/66s and they really seem to render at a
blazing speed (about 20% faster than our P5/60s). The problem, however, is
this:
Whenever we set the whole farm up to render a single animation, there are
usually no more than 2 machines that actually load the project file and get
going on the render, the other three (sometimes four) machines timeout because
they've been waiting to load the project file too long.
This is on a Lantastic network, and all the required maps are correctly stored
in a central location on the server and set to read-only, and they are
definitely not the problem. When we set up a net render queue for <ANY>
machine, and add machines to the slave farm incrementally, it works just as
advertized.
Is there a way to make the farm machines wait longer for the project file
and/or disregard the "file in use" flag? The best solution that I can think of
now is to set up an external program on all master machines that will make the
project file read only, but the person setting up the net render will have to
race the network and try to get the file marked read only before the farm
machines get a chance to lock it.
There's gotta be a better solution(?).
-MF