should some DNS TTLs be lowered?

[florinandrei]

This has become a pattern, repeated many times:

Client gets stuck trying to get new WUs. Reboot system. Client gets a new WU immediately.

I know this could be a bunch of coincidences, but still, it's happened so many times, it starts to look like a pattern. So many times already the GPU fans went silent for hours - and then immediately the client gets busy when I reboot.

Maybe the client gets stuck on some server? And when you reboot, a DNS lookup is done, it gets a new IP for that hostname (or something) and connects to a server that actually has WUs? If that's the case, lowering the TTL on the corresponding hostnames would solve the problem.

I'm folding GPU only, on Windows 10.

[Jesse_V]

I don't think it has anything to do with DNS, since the only DNS hostnames involved are assign1.foldingathome.org and assign2.foldingathome.org, which always point to the two Assignment Servers. Each AS will point you to a Work Server (WS) to get the workunits, and you'll then return them to the WS, or if that fails, to the Collection Server (CS). The WS and CS are referenced with raw IP addresses, so no DNS is used there at all. The trouble is that both the AS and WS are under very heavy load from the massive amount of new people connecting to them for workunits. When you restart your computer, you reset the timer on the delay between pings of the AS. Pausing the software, waiting about 30 seconds, and then hitting Resume should do the same thing.

Related threads:
viewtopic.php?f=61&t=33193
viewtopic.php?f=61&t=32424