I have installed the latest version of Boinc (8.2.4) and followed the instructions for Docker (Podman) setup and everything is working normally for "regular" project tasks.

I have found that when running docker-based project tasks (BOINC Central), despite the overall CPU usage being below the thresholds; and overall CPU % usage being relatively low (in task manager); the PC fans kick in and run at full speed. I have managed to partially solve this by creating an app_config file for BOINC_Central to specify only running 1 project at a time. When my PC had no other work to do, the PC was running acceptably. However, because the CPU usage for BOINC projects is below my defined thresholds, when other project tasks become available (e.g. World Community Grid), it is running those as well until the CPU threshold is reached, and the PC fans are kicking in at full speed and getting very hot (CPU usage in task manager is still relatively low).

Is there any way of configuring BOINC so that if it is running a Docker-based task, it DOES NOT attempt to run any other tasks? I don't want to have to prevent BOINC central from running on my device, but if I can't resolve this issue, I will have no option but to exclude it from my project list.

I have also noticed that when my machine goes to sleep, on restarting, all of the docker-based tasks rest back to 0% progress. This seems wrong?

I'm running on Windows Pro 11 (24H2) 64-bit
Thanks.

I don't know if you wish to do this, but one way to have only BOINC Central run is to have your other projects set to "no new tasks" (so only BC tasks are downloaded), or to suspend them altogether until you're finished with your BC tasks and are ready to run your other projects. As for "run only BC tasks even if other project tasks have been downloaded", I'm not personally aware of a way to 'force' BOINC to do so.

As for 'progress resets to 0', sounds like a checkpointing issue (or it doesn't commit checkpoints to disk at all).

---

https://boinc.berkeley.edu/dl/

As for 'progress resets to 0', sounds like a checkpointing issue (or it doesn't commit checkpoints to disk at all).

The current CPDN work has a long interval between checkpoints because they already use up a lot of disk space, especially if a network problem causes delays in uploading the intermediate results during computation. This may well be true of some of the other more resource hungry projects also.

Thanks for those suggestions Dave & JLDun. Suspending other projects will at least prevent tsks being pulled down. I'll look into that option.
With regards the resetting to 0, I also wonder whether because they are docker tasks, and so running in a Virtual Machine (possibly all in memory - the vmmemwsl task uses a lot of memory), the status isn't getting stored within boinc / project folders and when the pc goes to sleep, the VM is then getting "wiped", so has to restart again.

BOINC Central and CPDN do NOT play well together. So much is not correct with the docker app such as CPU time I see people with zero and others with like 350x runtime), memory usage, etc. See my post
https://boinc.berkeley.edu/forum_thread.php?id=15546&postid=116480#116480

boinc / project folders and when the pc goes to sleep, the VM is then getting "wiped", so has to restart again.

I haven't used docker but I have used BOINC instances in VMs and certainly if using virtual box, sleep does not lose VMs.

Thanks for that Dave.
I found today that on waking my PC up from sleep, the BOINC Central tasks had indeed "remembered" their status. I then had to reboot my PC and on restarting, the progress had reset to 0%. After a few minutes of running, so there was some progress in Boinc, I restarted Boinc Manager. On restarting the progress had reset to 0%. This suggests it's related to restarting of Boinc Manager that is "loosing" the progress of the tasks running in the docker app.

This suggests it's related to restarting of Boinc Manager that is "loosing" the progress of the tasks running in the docker app.

You could enable checkpoint_debug in event log options. You can then find out whether the tasks are checkpointing. Some task types do this as often as once a minute, others such as CPDN (the project I am most involved with) it is very much longer. Any task that hasn't checkpointed when you quit BOINC or reboot will inevitably go back to 0 progress.
You may be able to find out something on individual project sites about how often checkpoints are saved.

Edit: I don't think it is a problem with checkpoints not being saved when using docker.