Farming from mapped network drives? Or move plots to farming rig?

I now have 4 plotters running and the destination is to mapped network drives on my farmer. The network transfer is turning into a bottleneck so I was thinking, instead of transferring them, why not just land them on an external HDD on each of the plotters. Then I can map a network drive from the farmer to each plotter and farm the plots over the network.

Would this add latency to the farmer and cause me to possibly miss out on proofs found?

Edit: I’m in a Windows environment.

Try it and see, is the best advice I can give you. You can see harvester time in the logs if you set log level to INFO, and you get WARNs if they are over 5s now. It’s unlikely to be a problem unless you start getting to very large numbers of drives and network points.

I run a mix - each plotter has 8x3TB SAS HDD as a local “cache”, and they overflow onto 1 (soon to be 2) farmers using “stacks” of disks (I do mean EXTERNAL bunches of 4xsmaller HDD separated with felt floorpads to allow airflow with a 12cm fan attached with elastic connected with an Quadtoplus to a SFF-8087 mini SAS output - NO ENCLOSURE) using SAS expanders - over about 20 i’ll probably deploy an old mining server PSU with a 5v Step Down DC-DC to supply the ever increasing 5v requirement rather than trying to draw increasing power from the PC

It is not particularly power or space efficient but it is capital efficient - so my farmer builds an ever expanding local SAS farm plus the “cache” store of about 20TB on each plotter accessed over Samba - my “eligible” check are mostly decent around 1-2 seconds for 3-4 checks

The Plotters are Linux, each HDD is individual (no RAID) and i combine capacity using mergerfs - so no RAID like FS - i can treat the disk individually or combined - it means i can just take a few disk out and carry them over without worry about breaking volumes but i do lose the ability to parallel write at the end of each plot (unless i can configure mergerfs to round robin reliably)