Not sure if this is a thinkbox|deadline thing at all (but you guys are the easiest accessible so figured to check here first).
When using Vray RT over my network it seems that some slaves get stuck at the first pass
The buckets of this slave somehow take a long long time to arrive, they stay black while the CPU of the slave is continuously running at max capacity. In attached example Slave #2 is running at full CPU while all other slaves are already (10 minutes long) done with the rest of the image.
Eventually (usually 5-15 minutes) it does arrive and from then on all goes smoothly and it doesn’t happen again.
So basically my question would be: is this an issue that could be deadline related somehow or should I contact vray etc first?
I suppose there is some small chance this could be related to Deadline, but I don’t think it is, as all we are doing is starting up the spawner and telling it the file to render, then letting it run. Can you try to grab the render log from vray so we can see if any errors pop up?
ok, good to know.
i’ll get you the log tomorrow.
last time i looked it was just telling me 1 node was still busy and at another time it told me the node timed out.
funny thing is that each time it’s a different node.
I very much doubt it is us here, but you never know. I would enable the V-Ray “DR Bucket” Render Element, so you can see if the buckets are always the same Slave? Or do you know that for sure already? Does this Slave have any major different hardware that would explain this? Is it connected to the same network / same speed network card, etc?
Sometimes, V-Ray DR / V-Ray RT DR, just needs a purge and restart. This can sometimes include, having to make sure any temp/locally cached files are all purged (under user %TEMP%) and any other cached data, such as transferred assets. Are you using the V-Ray DR Settings to “Transfer/cache external files?”
I know for sure it’s different slaves (could see it via the desktopwidgets as in the preview)
most of the time on the older half of the farm, but i’m sure to have seen it at least once on a newer node.
All assets are indeed transferred. will check how the temp folders are doing tomorrow.
I will also have a look if it’s only happening in 2017 or also in 2015 as I didn’t have it there before (at least before the latest vray version).
Cool. Also, try using 50% of the machines you were using last time there was troubles. It might just be a combo of too much scene data, external data, bandwidth (net/disk), combined with the amount of ‘communication’ the master machine is having to partake in. Another way to test this scenario, is make sure the MASTER machine does NOT do any rendering (“Use local host” = False) checkbox setting in DR settings.
