Running version 7.16.11 x64
OS is Windows 10 home
Project is Milkyway@Home

Issue is that the project launches tasks that completely ignore the "computer in use" limitations, running at 100% of the CPU. Deliberately suspending that tasks releases the CPU hold (those are reportedly 4 to 5 hours total tasks), but smaller 7 minutes tasks are still loaded and processed despite the computer being in use, however only on one core +GPU. The whole project has to be manually suspended to regain full control.
Moreover, this rather perplexing notice was issued at 5 am, while the process was allowed to run in an attempt to see if one rogue task being cleared would allow the next one to behave according to the settings (it didn't):

Milkyway@Home: Notice from server
Milkyway@home Separation needs 71.53MB more disk space. You currently have 0.00 MB available and it needs 71.53 MB.

For the record, that computer has gigabytes a free disk space, and the disk utilisation from the BOINC manager acknowledges this:
* used by BOINC: 94.39 MB
* free, available to BOINC: 34.76 GB

How can a task disregard the computer in use restrictions?
Is this an issue only with the Milkyway@Home project (the only current one, since SETI@home terminated two years ago) in which case I would be willing to switch project; or one that applies to all BOINC projects, in which case I would consider terminating my contribution unless this issue gets fixed.

(Edit: even if I joined the board only today, I am an old timer with BOINC, with over 26 million units performed)

Some questions on your system to get more information.

What version of the manager are you using? The current version is 7.20.2.
How many cores are you using, and are you using H/T?
On the computer preferences tab > Computing
What is the use at most % CPU - mine is set at 60% -- H/T
What use at most % CPU time - mine is set at 100%

On the computer preferences tab > Disk & Memory
Are you limiting the use of memory, or do you have the 'Leave at least # GB free' checked? Make sure this is checked with a large amount of spave
and BOINC will use as much space as it needs. I have mine set at 10 GB for many years with no problem. Just make sure the other 'use' options
are unchecked.

From the msg from the server --
Milkyway@Home: Notice from server
Milkyway@home Separation needs 71.53MB more disk space. You currently have 0.00 MB available and it needs 71.53 MB.

it looks like memory is being restricted somehow.

Even though I'm running win 11 Pro, these settings are carried over from 10.

---

Check the setting on Boinc manager how you are allowing the GPU to run. The GPU has settings to run: Always, Based on preferences or Never. You should use Based on preferences.

Said right there already 7.16.11.

I don't care that there is a newer version, if I do not update the manager, that is not a license for the task to disregard the settings that was respected until two days ago.

When allowed to run, one task takes 5 cores, and another one takes the last core plus the GPU, running 100%.
And no, I am not using hyper threading.
And no, I am not limiting memory either.

When the computer was not in use, I allowed 90%; but it does not matter what setting I give. I just tried 20%, and what I get is something that swaps between 100% load for 1 second, and 0% for 4 seconds. But that is while the computer is in use.
That is my issue.
I don't care it taking 100% when I am not using the computer; but it ignores keyboard and mouse inputs, assuming that I am never using the computer.
When I am typing, I expect the load of BOINC to drop to 0% CPU. It does not.

Everything is (and always had been, for years) set to "based on preference".

It is just that the system currently ignores the running preference "Suspend when computer is in use". I used to have "'In use' means mouse/keyboard input in the 3 minutes" and have increased this to 5, with no effect.
It is as if BOINK fails to recognize mouse and keyboard activity.

Could you please post the contents of your global_prefs.xml and global_prefs_override.xml files?

For the record, I left the system to carry on the project overnight, the task being marked has having about 2 hours remaining.
It is assumed that it completed, and that a new task was uploaded.

This morning, I found that the current task is marked as "suspended, computer is active" which is what it was supposed to be. My task manager also shows the CPU being chiefly idle, since I am not using resources to any extreme degree at this very moment; again, as it should be.

The only thing that changed is that this would be a new packet.

It is therefore suggested that some tasks might be configured in a manner that disregard the computation preference.

(groan... as if you could see something I can't)



This file has been untouched since 2019:

<global_preferences>

<source_project>http://setiathome.berkeley.edu/</source_project>
<source_scheduler>http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi</source_scheduler>
<mod_time>1149332521</mod_time>
<idle_time_to_run>3</idle_time_to_run>
<cpu_scheduling_period_minutes>60</cpu_scheduling_period_minutes>
<hangup_if_dialed/>
<work_buf_min_days>0.1</work_buf_min_days>
<max_cpus>2</max_cpus>
<disk_interval>60</disk_interval>
<disk_max_used_gb>100</disk_max_used_gb>
<disk_max_used_pct>50</disk_max_used_pct>
<disk_min_free_gb>0.1</disk_min_free_gb>
<vm_max_used_pct>75</vm_max_used_pct>
<max_bytes_sec_down>0</max_bytes_sec_down>
<max_bytes_sec_up>0</max_bytes_sec_up>
</global_preferences>



This file was last updated yesterday, as I was adjusting parameters trying to keep the task from overriding settings.

<global_preferences>
<run_on_batteries>0</run_on_batteries>
<run_if_user_active>0</run_if_user_active>
<run_gpu_if_user_active>0</run_gpu_if_user_active>
<idle_time_to_run>5.000000</idle_time_to_run>
<suspend_cpu_usage>25.000000</suspend_cpu_usage>
<start_hour>0.000000</start_hour>
<end_hour>0.000000</end_hour>
<net_start_hour>0.000000</net_start_hour>
<net_end_hour>0.000000</net_end_hour>
<leave_apps_in_memory>0</leave_apps_in_memory>
<confirm_before_connecting>0</confirm_before_connecting>
<hangup_if_dialed>1</hangup_if_dialed>
<dont_verify_images>0</dont_verify_images>
<work_buf_min_days>0.100000</work_buf_min_days>
<work_buf_additional_days>0.200000</work_buf_additional_days>
<max_ncpus_pct>90.000000</max_ncpus_pct>
<cpu_scheduling_period_minutes>60.000000</cpu_scheduling_period_minutes>
<disk_interval>60.000000</disk_interval>
<disk_max_used_gb>100.000000</disk_max_used_gb>
<disk_max_used_pct>50.000000</disk_max_used_pct>
<disk_min_free_gb>0.100000</disk_min_free_gb>
<vm_max_used_pct>75.000000</vm_max_used_pct>
<ram_max_used_busy_pct>50.000000</ram_max_used_busy_pct>
<ram_max_used_idle_pct>90.000000</ram_max_used_idle_pct>
<max_bytes_sec_up>0.000000</max_bytes_sec_up>
<max_bytes_sec_down>0.000000</max_bytes_sec_down>
<cpu_usage_limit>90.000000</cpu_usage_limit>
<daily_xfer_limit_mb>0.000000</daily_xfer_limit_mb>
<daily_xfer_period_days>0</daily_xfer_period_days>
</global_preferences>