Page 1 of 1

Fails to find work unit over and over and over again

Posted: Tue Aug 12, 2025 3:27 pm
by Woodward
On my M2 Mac, I had about one good month of folding, and then it just started to do this every single time I try to use the application.

Log from the console:
15:21:28:I1:WU82:Retry #3 in 8 secs
15:21:36:I1:WU82:Requesting WU assignment for user WoodwardIII homelab team 1066329
15:21:36:I1:OUT8:> POST https://assign4.foldingathome.org/api/assign HTTP/1.1
15:21:36:I1:OUT8:< HTTP/1.1 503 HTTP_SERVICE_UNAVAILABLE
15:21:36:E :OUT8:HTTP_SERVICE_UNAVAILABLE: {"error":"No appropriate assignment"}

I've already tried reinstalling the application as well as messing around with tons of different network settings, and I can't seem to get anything to work.

Thoughts?

Re: Fails to find work unit over and over and over again

Posted: Tue Aug 12, 2025 3:48 pm
by Woodward
Note: I'm posting this here because I don't understand how to navigate these forums. This is probably not a category.

Re: Fails to find work unit over and over and over again

Posted: Tue Aug 12, 2025 4:11 pm
by muziqaz
Yes, there is shortage of WUs.

Re: Fails to find work unit over and over and over again

Posted: Tue Aug 12, 2025 4:14 pm
by Joe_H
There is a shortage of GPU WUs, but that shouldn't affect your M2 Mac. What is your CPU thread setting on that Mac? Recently most projects were set to require a minimum of 3 CPU threads, if you have yours set to 2 you wouldn't get work.

Re: Fails to find work unit over and over and over again

Posted: Tue Aug 12, 2025 11:25 pm
by arisu
There seems to be an issue with CPU WU allocation. Several people have reported this. I've brought it up internally and it will be investigated.

Re: Fails to find work unit over and over and over again

Posted: Wed Aug 13, 2025 5:31 pm
by Woodward
Plot thickens, works fine with 4 CPU cores enabled, but not 3. (I want 3 because given the thermal constraints of the device, it's actually faster)

Re: Fails to find work unit over and over and over again

Posted: Wed Aug 13, 2025 5:34 pm
by Woodward
I've been manually bumping it up to 4 to get something assigned and then back down to three again. It's kind of stupid, given my RPIs can get units with 4 cores, but it literally takes 5x as long to run a unit on them (over my Mac with 3), even with the extra core.

Re: Fails to find work unit over and over and over again

Posted: Thu Aug 14, 2025 5:21 pm
by daiko
Woodward wrote: Wed Aug 13, 2025 5:34 pm I've been manually bumping it up to 4 to get something assigned and then back down to three again. It's kind of stupid, given my RPIs can get units with 4 cores, but it literally takes 5x as long to run a unit on them (over my Mac with 3), even with the extra core.
I have found the same limitation—it seems 4 cores is the minimum now to get a WU from the assignment server. I have an old 5K iMac that also performs better on 3 cores, so I have taken the same approach of backing it back down to 3 after acquisition. Unfortunately, using 4 cores for acquisition often nets a WU from the Disordered Proteins projects (18803 or 18806), which perform pretty miserably on a few of my machines. :roll:

Re: Fails to find work unit over and over and over again

Posted: Thu Aug 14, 2025 6:39 pm
by muziqaz
Minimum is still 3. You guys need to understand that we do not go out and release projects specifically for super low end. You cannot expect researchers target that market, when their job and research papers are on the line. When we are testing these projects we see how they run with mid range and high end set up. We can never predict how various people will be running folding. Some might pause once in a while, others pause overnight, others fold a bit, then do other things, then fold a bit again. If we freely allow super low end CPUs (core count) to get every project, we will end up with loads of WUs timed out.

Re: Fails to find work unit over and over and over again

Posted: Thu Aug 14, 2025 7:40 pm
by arisu
daiko wrote: Thu Aug 14, 2025 5:21 pm
Woodward wrote: Wed Aug 13, 2025 5:34 pm I've been manually bumping it up to 4 to get something assigned and then back down to three again. It's kind of stupid, given my RPIs can get units with 4 cores, but it literally takes 5x as long to run a unit on them (over my Mac with 3), even with the extra core.
I have found the same limitation—it seems 4 cores is the minimum now to get a WU from the assignment server. I have an old 5K iMac that also performs better on 3 cores, so I have taken the same approach of backing it back down to 3 after acquisition. Unfortunately, using 4 cores for acquisition often nets a WU from the Disordered Proteins projects (18803 or 18806), which perform pretty miserably on a few of my machines. :roll:
Most likely it's just coincidence and not a real limitation. I'm getting CPU WU shortages with 8 cores from time to time as well.

Re: Fails to find work unit over and over and over again

Posted: Thu Aug 14, 2025 7:46 pm
by daiko
muziqaz wrote: Thu Aug 14, 2025 6:39 pm Minimum is still 3.
Are you sure about that? I have several machines (a few of which are very capable CPU machines), none of which will receive a WU if the cpu count is set to 3. It may just be that the available pool of WUs does not have a sufficient number of small atom count projects, but I haven’t seen a WU served that requires less than 4 cores for days, personally.

Re: Fails to find work unit over and over and over again

Posted: Thu Aug 14, 2025 7:54 pm
by muziqaz
daiko wrote: Thu Aug 14, 2025 7:46 pm
muziqaz wrote: Thu Aug 14, 2025 6:39 pm Minimum is still 3.
Are you sure about that? I have several machines (a few of which are very capable CPU machines), none of which will receive a WU if the cpu count is set to 3. It may just be that the available pool of WUs does not have a sufficient number of small atom count projects, but I haven’t seen a WU served that requires less than 4 cores for days, personally.
Yes, I'm trying to drop it down to 2 threads now