Vray Spawner In a WAN

[c-boogie]

I've been running a small farm in my office for about 10 years now. We have about 15 dedicate machines and about 20 more office computers that kick on at night. We use backburner for the most part, but often take over a few of the dedicate boxes to run spawner for larger or more complicate still renderings. This all works just fine.

 

Our corporate HQ has decided to supplement the farm with an additional farm in our main office. They have another 12 dedicated servers up there and are set up in a similar fashion to my office. The idea is that our rendering resources will eventually be consolidated there and will be opened up to everyone in the company.

 

So my first task is to get Spawner running in this WAN environment. On it's most basic level, I have it running just fine (unc paths, permissions, services, etc.) I have that working like a champ. However, the issue I'm running into is figuring out how to make this work more effectively. When I was using the dedicated machines on my LAN, access speed for my local machine and the dedicated boxes were equal. Now in this WAN scenario, if my scene is large enough or has a lot of linked assets (proxies, larger maps, xref's, etc.) the machines in the remote office take 20-30 minutes to show up in the vray log window.

 

So . . . what exactly is Spawner transferring to each machine? The entire scene? Just the maps and network assets? i think i just need a cleared understanding what is happening "under the hood" when spawner is running. Does anyone out there run a system like this? If so, any tips or lessons learned? Any insight would help.

[Eric]

I don't know what I'm talking about. That being said, what if you tried locating your working projects to a sync'd location, like a Google Drive folder on your desktop so all the other computers are sync'ing that folder locally. Then, tell Vray to NOT "Transfer missing assets" in the Distributed Rendering Settings dialog box.

[c-boogie]

That's an interesting idea. Unfortunately (for me) the parent company doesn't allow any automatically synchronizing applications like that. Security or bandwidth issues . . . take your pick I guess. But all of our offices are connected through fairly substantial connections. Not nearly LAN speeds but not DSL either (I think my office has a 50mb fiber connection to the main office). So I can store my files here on my server, and they are accessible in the main office (and vice versa). I've tried storing all of my scene assets (maps, proxies, etc.) on the servers in the main office (so that they are accessible faster to the render nodes up there), but opening my scenes here are painfully slow.

 

I guess I'm wondering how much of the scene is transmitted each time. Is my local machine holding on to the geometry and calcs (LC and Irr) and only sending the appropriate data to the render node on a bucket by bucket basis? Or is it transferring the ENTIRE scene to each render node and then only sending back the information for the buckets it renders?

 

One thing that the previous post reminded me of is that we're still on Vray 2.4 still. Part of this upgrade is moving to Vray 3 and it sounds like there are additional features in the DR field that might be helpful (like "trasfer missing assets").

[Francisco Penaloza]

There are several technicalities involved in what you are trying to achieve that goes beyond what is typical setup.

Your first problem will be make Backburner run as a service, and that will require some IT twiking in your network permissions.

I would recomend to post your question in Chaos group forum, there, the same people from Chaos can guide you with the correct workflow.

 

If you are planning to move to VRay 3 remember that V-Ray now charge for render nodes, just remember that so you can plan your budget accordingly.

Best regards.

Edited June 26, 2015 by fco3d