Max 2016 + Vray Next distributed rendering issues

[m0dus]

Hi!

We've five workstations over here with five full (gui + node) licenses, all running on 3ds max 2016 with the latest stable Vray Next installed. When rendering distributed everything starts fine, spawners can be found, starting, loading the scene and start rendering, but without a clear reason some of them suddenly drop. They start again and continue rendering and dropping again. Sometimes they hang on the last bucket so the render cannot finish. There's no difference in behavior between rendering through backburner or from max directly (the frame buffer finishes though), and all the workstations show the same behavior. It doesn't make a difference from which workstation we're starting the render. The issues started appearing this year. Anyone any idea? Thanks!

[vlado]

If you look at the log files on the render slaves, are there any warnings or errors? What is the message that you get when a server drops?

Best regards,
Vlado

[m0dus]

Hi Vlado,

The logs don't show any errors or warnings. The server itself doesn't drop, the buckets do. So for instance we're trying to render a big scene (7000px wide, at this moment, my workstation has 40 threads, everything starts ok, but after a few minutes it starts dropping buckets, until there's basically no cpu activity anymore, and they appear again after a few moments. It looks like it's rendering in bursts.

But we do save the vray raw image. Can that be the issue? That it has to save the information from the buckets first before sending new data to render? We'll test it today ourselves and keep you posted.

Best,

[m0dus]

Hi Vlado,

As said, we tested it. And it's indeed the saving of the vray raw image together with the distributed rendering. It renders fine when we turn the option off, the spawners are performing as expected. Is there anyway to avoid this? Because we do like the vray raw image . Thanks!

Best.

[Zdravko.Keremidchiev]

I tested the latest Stable build (28928 and Max 2016 but I can't reproduce the issue. Does it happen with particular scenes? Do the same nodes fail or it's on random? How many machines participate in the rendering process? Can you reproduce it with one slave only? Have you done any modification to the network after upgrading V-Ray?