Hi there, I am trying to overcome the 16 threads limitation per worker and used 6 of them, it did the job on the rendering side, having multiple frame rendered at once BUT most of the exr written on disk where all wrong or almost null in term of size. Anyone seen that before ?
Running on a 128 threads node 128 RAM and writing on a NAS 10Gig network
Help welcomed indeed 
Thank you
T
Hello!
The issue with the output being null or are written all wrong it could be overloading local storage throughput or the render node NAS connection.
The logs from frames that have failed will be helpful. The Worker logs will be a mess with all the threads. You can use this github script to split up each thread if needed. Feel free to share the logs here or you can open a new ticket with us with the logs.
1 Like
I will look for that and get back to you, thank you for the feedback
T