Several of my machines running SPT WUs at priority 0 on one core have been given 1000 WUs today! I now have nearly half the 11,000 active SPT WUs when I'd expect to have about 20.

The other three cores are running WUs from another project.

I think the issue is at your end...

The machines are a mixture of 1GB Pi3 and Pi3Bs, and 2GB Odroid C2s all running 64-bit Ubuntu 18.04 limited to a single WU because of their relatively limited RAM sizes.

I created the file /var/lib/boinc-client/projects/boinc.tbrada.eu/app_config.xml containing:

<app_config>
  <app>
    <name>spt</name>
    <max_concurrent>1</max_concurrent>
  </app> 
</app_config>

Which was working as expected until this morning!

I have a couple of Odroid C2s and an XU4 running the project at priority 100, and they currently have more realistic number of WUs (40 to 60 each).

Jon

The madness might have started on Saturday: one of the C2s running SPT and PADL at priority 100 has 1000 SPT WUs, and downloaded most of them on Saturday...

Jon

I have no idea what and why that happened. If you want to abort them, go ahead, or just let it do it's thing (which would be best).

---

Several of my machines running SPT WUs at priority 0 on one core have been given 1000 WUs today! I now have nearly half the 11,000 active SPT WUs when I'd expect to have about 20.

The other three cores are running WUs from another project.

Jon

This can happen if your other active projects, ones set to receive tasks, are also set at a zero resource share, Boinc THINKS you must mean something else so it fills your cache with the project you are running. What I do is attach to a project that rarely has tasks and set it to a resource share of 10 or 25, I sometimes use World community Grid as it still has settings for projects that long ago stopped sending out tasks. Boinc is a program and as such sometimes makes assumtions that are just whacky.

Several of my machines running SPT WUs at priority 0 on one core have been given 1000 WUs today! I now have nearly half the 11,000 active SPT WUs when I'd expect to have about 20.

The other three cores are running WUs from another project.

Jon

This can happen if your other active projects, ones set to receive tasks, are also set at a zero resource share, Boinc THINKS you must mean something else so it fills your cache with the project you are running. What I do is attach to a project that rarely has tasks and set it to a resource share of 10 or 25, I sometimes use World community Grid as it still has settings for projects that long ago stopped sending out tasks. Boinc is a program and as such sometimes makes assumtions that are just whacky.

Yes, that is what had happened!

LHC was at 100 (max 3 WUs), TBRADA was at 0 (max 1 WU). and Universe was at 0 (max 3 WUs). LHC sixtrack work is intermittent, and ran out of work a few days ago. It looks like as each machine ran out of LHC work, instead of downloading three Universe BHSPINv2 WUs, it downloaded 1000(ish) SPT WUs.

I've changed Universe to 100, and aborted most of the SPT WUs, and as LHC is out of work I've suspended it.

Thank you.

Jon

If possible, please submit this issue to https://github.com/BOINC/boinc/ and link it here. If it already exists, also link it here.
Thank you, and I am sorry for the incident.

If possible, please submit this issue to https://github.com/BOINC/boinc/ and link it here. If it already exists, also link it here.
Thank you, and I am sorry for the incident.

This is a long standing known Boinc issue from the early days and it's never been fixed, they are rewriting Boinc to be 64bit only so maybe they will fix it then. The existing 32bit version will NOT be dropped though. Just over time fewer and fewer projects are providing 32bit apps so Boinc is trying to clean things up and root out some old problem causers.

I hope that dropping 32 bit support is a joke.