Anyway, after searching for posts containing the word "dumping" I find that other folk have been annoyed by having their donations spurned by servers that "did not like" them. so I am pretty sure there's a donor relations issue here that you might care to raise with the "developmental" folk at F@H, quite apart from technicalities. Back in the dark ages before the moon landing, the operator of the IBM 1401 would politely give me a thing called a "console card" whenever my Fortran II programs brought das blinkenlights to a sudden halt. Half a century later, "Server did not like results, dumping" is rather less helpful than a console card, and just possibly not the nicest way to advise a donor that their computer's work has been wasted, no matter what the reason. Since servers do not have emotions, it makes sense to issue a warning that tells it like it is, ie that the submitted results have failed validation checks, just in case it is the donor's machine or comms that is the problem. But since my scanning of the posts suggests that it may also be (shock! horror!) a mistake at the F@H end, it would be polite to hand over a default number of points, by way of thanks, not least because they cost F@H next to nothing. And some rephrasing of the warning message would do no harm.
Points are no big deal for Team appepi's collection of antique devices, mind you. Z601, Z602, Z603, Z604, Z605, Z802, Z803, Z441 and Z442 haven't had a proper workout since they were fired from their nice air-conditioned corporate leases onto eBay, the i5-750 (PC2) in the attic had nothing at all to do except play music now and then, while the refurbished 2005 Dell Inspiron 6000 (Pentium M) and the 2008 Dell 910 (Atom 270) were happy to switch from Ubuntu 18.04LTS and Radio Paradise to Windows 10 1909 on a SSD through a USB2 port and grind out a few points just for the hell of it. I'm also running my 2013 Toshiba ultrabook in "Light" mode which allows it to run cool enough for 24/7 work. The only ones that don't get a gig are this ASUS Zenbook, which is reserved for non-Folding duties like writing this post, and PC0 in the kitchen, which is an ASUS 1015PX netbook whose CPU fan doesn't work any more, and has been a bit dodgy ever since I poured that cup of coffee into it. And in fact it's just the ASUS GTX1060 6GB in Z603 that's doing the heavy lifting, detuned to about 93% with GPU Tweak II via an aggressive fan profile to keep it cool. No doubt this will all come to an abrupt halt when the next power bill arrives, or one of the nice old beasts dies of heat exhaustion, but until then the points are just a way of checking that the machines are working. However, more up-to-date donors might reasonably be offended if the offering of their pride and joy is unceremoniously dumped just because a server "doesn't like it".
Also, I think the shortage of WU's may have revealed a glitch in the algorithm that (as far as I can tell) progressively increases the delay between a failed request and the next. When output from one of the Z's drops off, I usually find that it had gotten itself into a state where the inter-attempt time has blown out to 5+ hours. When I switch the client to IDLE and then back to "While I'm working", they almost immediately pick up a WU. This happens with both CPU and GPU WUs. It might be worth the developers introducing some randomness into the waiting times, to prevent this.
