Project has no work since May

Do you happen to be running only Windows? It's been months since either of my Windows computers got any Rosetta work.

Hi Bill,
Maybe you can catch some of those Rosetta Work Units with a Batch file
Like this - Upd-Rosetta.bat
@echo off
cd /d c:Program FilesBOINC
:loop
cls
echo "boinccmd.exe --project https://boinc.bakerlab.org/rosetta/ update"
boinccmd.exe --project https://boinc.bakerlab.org/rosetta/ update

TIMEOUT /T 600
goto loop

Of course you would need to leave a terminal window open for that

I tried the Windows Task Scheduler Loop - but the Win 10 task scheduler doesn't
seem very reliable. I can't seem to scale to 5 minute intervals very well.

Regards,
Ian

Which indicates their CPU's are busy doing work other than BOINC.
If this shouldn't be the case, you can use Task Manager (or better yet Process Explorer) to see what else is making use of the CPU.
Or you can make use of the local settings (or separate web based preferences) to limit the number of cores/threads BOINC can use so that BOINC processing isn't impacted by the other work on the system that's taking up CPU time.

xx days + 0.01 additional days is best.
The larger the additional days setting, the lower the cache falls from it's full capacity before it refills...

Fair enough if you've got reliability concerns. You have to do whatever you need to to remain successful.
By all means run a smaller cache than 2 days if you need to as well - I got the impression one or two of your machines run a pretty large cache already and 8hr runtimes may push you past Rosetta's 3-day deadline

Got my last new task at 09:22 UTC, only some resends after that moment.

I still have over 600 tasks cached (630-650 was the normal interval while there was a constant supply) but I don't expect resends to be able to hold the line for long. Was a nice run though, let's hope there will be more in the new year. Until then I'm moving over to SiDock and WCG as secondary.

Yes, just a little longer than expected.
The number of tasks you have seems like a lot, so I took a look to see you have several PCs running, but the majority are running tasks for barely more than 2hrs, for which you're getting very little credit, while using up tasks faster than expected.
I don't mean to drag you into the discussion we're having elsewhere, but it will make an enormous difference to how long batches of tasks last if you change your Rosetta Target CPU runtime explicitly to 8 hours here, matching the time Boinc schedules unstarted tasks for, and limit your cache to 2-2.5 days at most to ensure they complete within their deadlines.
The limited number of tasks we get at a time will last a lot longer and can be shared out among more users, while you'll see a massive credit boost per task and a smaller credit boost overall.

I managed to get the results page to load and can confirm that PV jail has fewer inmates. Used to have around 1400 tasks pending, now down to under 1200. Among them are still some that were reported back in August.

I'm glad you've got some information. At least it is doing something, from what you report.