Got home last night and found all the hosts still running, but issues with six GPU slots. Three GPU (980Ti+2x1070) with hung download, one 980Ti running downclocked, one 1080Ti with "paused by user" (gremlins???), and one 980Ti halted with error. After rebooting the host, the 980Ti spins up fans, but no longer recognized by the host. Replaced the "bad" 980Ti with a spare 970 and resumed folding on the slot. Will investigate the "bad" 980Ti on a spare host, but looking like RMA required. Production is back 10MM+ ppd again. Good news is that the old UPSs with new batteries kept all the hosts running for about 2 weeks unattended.