This is exactly what I was going to ask for. I would suggest cutting up the frame in to 32 equal slices (assuming less than 32 nodes) or, number of slices = total number of nodes times four (or so) and each node works on a slice until finished at which time it asks for another slice until all done. The bar to create a cheap render farm out of a few number of old computers has never been lower, but the way it works now it is often not a benefit.

@James 1903 did the trick!

Windows 10 ver. 1903 seems to have done it! My nodes are showing hard working CPUs now. This is NOT, unfortunately, giving a very big performance boost. I suppose the GPUs were working fine all along, and if you've read my previous findings in this thread, compared with GPUs, the CPUs are not adding a lot of muscle, relatively. But, it's good to hear the fans winding up in the node computers, they show a lot of CPU utilization now that both are fully up to date. THANK YOU!

@James No, the GPU, in both cases was being reported at about 8%, but I could tell it was busy. I don't trust TM for GPU reporting, that's why. In the case where the CPU was NOT asked to render, it was at maybe 20+%. In any case where the CPU IS asked to render, it's close to maxed out.

CPU-only took 3 min 20 sec. Had to try it.

@James GPU only render was 80 seconds.Task Manager showed moderate CPU usage and about 8(!) percent GPU (Nvidia 1060), but I don't particularly trust the TM for GPU usage stats, the GPU fan was blowing pretty hard. Adding 3 out of 4 of the physical CPU cores to the mix (I've learned to leave one core for the system, it helps), so six "threads" and the GPU only shaved 20 seconds (one minute render time). CPU fan deafening. TM again showed about 8% GPU usage but I don't believe it. This was conducted in Blender, using Cycles, not Crowdrender, Win 10 1903.

Thanks for the quick reply. Master and one node are on 1903, second node is 1803. I run updates frequently but MS has been slowly rolling out 1903 and haven't offered it up for that computer yet. I do know how to force it though, so I probably will. However, both nodes are very similarly under-performing. I'm happy to help troubleshoot.