Robert Kelly

On our work machines i have found a network write of a VDB, it streams to the server in small increments slowing it down in a big way. writing locally then moving it is faster. i may roll this into our cache node or look for a environment variable to solve the problem

we found one thing that helped. in the hython script that loads the file to render, the pipeline guys put a 10 second pause, this gave the hython session some time to catch up to the loading of the hip file and not error.

I think we have just got this issue in our studio’s render farm. When it happens a gob will get into the state on 4 out of 5 jobs. Render jobs that don’t come from IFD’s rendering on karma CPU or XPU, redshift and even geo processing. The last thing in the logs is..

Rendering frames 1010-1019

Then nothing. The node will use 1% cpu and not unlock unless it times out. It’s hard to debug as ther is nothing else in the logs.