Arras - distributed rendering : Infrastructure question

[nyue]

I have briefly read this page regarding arras and distributed rendering.

I would like to clarify if the nodes and coordinator needs to share a common file system e.g. NFS ? or does the coordinator package up just sufficient information for each package of work for each node?

I'd like to throw together a couple of boxes for a quick test and want to know upfront how much setup I need to perform just to kick some tires

Thank you in advanced.

[toshi-kato]

Hi, Nyue,

Basically, the coordinator only does the dispatch of computation to the backend hosts and does not do anything about the scene data itself. And all the backend computation needs to open the data (textures, geometry files, etc) by the same location definitions, and you need to manage them. I think there are several different ways to do this, but the easiest way is just using NFS and mounting the network drive from all backend hosts, then storing the scene data on them. In this case, you have to specify the data location as an absolute path (not a relative path). Other solution would be copy entire scene data to the local disk of each backend hosts, but I think using NFS would be sufficient for most of the production work.

Toshi