|01 January 2013||#1|
Re-rising of an old problem: Stuck buckets in DBR
I've been having this stupid Distributed Bucket Rendering issue with Mental Ray, seemed to be solved from time to time - but eventually it is back again.
You know when there is no connection between machines or it is somehow blocked by firewall, translation takes a long time and eventually Max comes up with an expected error like: Host blabla-PC not found.
When network connection is not a problem, it starts translation immediately then starts rendering, but very often, just after rendering starts, (probably) the slave buckets stuck at certain point and while waiting their completion (which is actually not a real rendering calculation) render takes longer than rendering on a single machine.
I thought the problem was gone after raising my system ram to 32 GB on each machine however just today same problem is back again.
Rendering starts- Master goes on fine, while 12 cores of the slave leaves 12 black buckets on render window until master finishes all the work, just then slave actually starts rendering -which is noticable from the sound of the CPU fan (and the performance level in task manager) and makes rendering longer.
I've had the issue almost any network system I worked on. This must be something very common.
I have an 8 port gigabit hub. Two machines are connected to it plus a network printer. Master machine is connected to the Internet via a wireless adapter. The other machine is configured to use Master machine's internet connection.
I couldn't ever find the reason to this.
I started to think Mr rendering itself on the master machine is blocking some network transfer to trigger the start of the slave. It even locks the mouse during rendering. Nowadays even priority reduction doesn't even help.
|Thread Closed share thread|