Advanced search

Forums : Technical Support : MT VB units get stuck at 0.100% for 20 mins and then get aborted. CPU stayed idle.
Message board moderation

To post messages, you must log in.

AuthorMessage
2DFe3jP8aH5yqsZe4vgGmiFFkfa5

Send message
Joined: 26 Jun 18
Posts: 2
Credit: 311,320
RAC: 3
Message 22910 - Posted: 1 Dec 2021, 7:50:36 UTC
Last modified: 1 Dec 2021, 7:55:45 UTC

Just to note I'm not downgrading VB because I'm using other OS applications for it and they work fine and don't want to brake that. I'll just wait for VB or cosmology@home to fix the problem.
11/30/2021 11:59:27 PM | | Project communication failed: attempting access to reference site
11/30/2021 11:59:28 PM | | Internet access OK - project servers may be temporarily down.
12/1/2021 12:01:11 AM | Cosmology@Home | Sending scheduler request: To fetch work.
12/1/2021 12:01:11 AM | Cosmology@Home | Requesting new tasks for CPU and NVIDIA GPU
12/1/2021 12:01:12 AM | Cosmology@Home | Scheduler request completed: got 44 new tasks
12/1/2021 1:02:43 AM | Cosmology@Home | Aborting task camb_boinc2docker_2515834_1638337626.737253_0: exceeded elapsed time limit 1857.19 (86400.00G/46.52G)
12/1/2021 1:02:59 AM | Cosmology@Home | Computation for task camb_boinc2docker_2515834_1638337626.737253_0 finished
12/1/2021 1:02:59 AM | Cosmology@Home | Output file camb_boinc2docker_2515834_1638337626.737253_0_r595011334_0.tgz for task camb_boinc2docker_2515834_1638337626.737253_0 absent
12/1/2021 1:03:42 AM | Cosmology@Home | Aborting task camb_boinc2docker_2515386_1638337608.459948_0: exceeded elapsed time limit 1857.19 (86400.00G/46.52G)
12/1/2021 1:03:52 AM | Cosmology@Home | Computation for task camb_boinc2docker_2515386_1638337608.459948_0 finished
12/1/2021 1:03:52 AM | Cosmology@Home | Output file camb_boinc2docker_2515386_1638337608.459948_0_r2044516695_0.tgz for task camb_boinc2docker_2515386_1638337608.459948_0 absent


Windows 10
BOINC 7.16.20
Virtual Box v7.16.30
ID: 22910 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 187
Credit: 8,308,328
RAC: 292
Message 22911 - Posted: 1 Dec 2021, 9:13:42 UTC - in response to Message 22910.  

Change these setting on the Cosmology@Home Preferences page. Mzx # CPUs controls how many core get assigned per VM. Each VM will use about 2Gb of RAM. You can also control this using app_config.xml method per the FAQ section. Start with Max # job set low or you might run into the Vboxwrapper losing communication problem and work units starting later (about 24 hours)

Max # jobs No limit
Max # CPUs 4
ID: 22911 · Report as offensive     Reply Quote
.clair.

Send message
Joined: 4 Nov 07
Posts: 629
Credit: 12,068,402
RAC: 0
Message 22912 - Posted: 1 Dec 2021, 21:58:55 UTC

Threre are many crunchers that have problems like this with vbox across many different projects
and no one has a full fix for it
I see you have a 12 core cpu
vbox was found to have problems with using more than 8 cpus when cosmo started using it
so you don't get berid in workunits , set 8 for that as well
Max # jobs - 8 {is the same as `tasks in progress` on `tasks` page}
Max # CPUs - 8
ID: 22912 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 187
Credit: 8,308,328
RAC: 292
Message 22914 - Posted: 2 Dec 2021, 2:45:59 UTC - in response to Message 22912.  

I would set Max # CPUs to 2 or 4. You could even set it to 1 if you prefer. The work units complete so quickly with 2 or 4 cores assigned, I don't see any point in assigning more as it just mean more time spent creating and destroying VM images for each task ran.
My 4 core tasks took about 5 minutes and the 2 core are claiming about 10 minutes. CPU time per task is about the same.
ID: 22914 · Report as offensive     Reply Quote
2DFe3jP8aH5yqsZe4vgGmiFFkfa5

Send message
Joined: 26 Jun 18
Posts: 2
Credit: 311,320
RAC: 3
Message 22930 - Posted: 10 Jan 2022, 23:41:36 UTC

Noting is working. Sometimes the Linux in VB doesn't even run with a blank screen.
I'm just going to stop running Cosmology from Virtual Box.
ID: 22930 · Report as offensive     Reply Quote
Jesse Viviano

Send message
Joined: 29 Nov 14
Posts: 36
Credit: 1,193,092
RAC: 980
Message 22931 - Posted: 15 Jan 2022, 1:14:26 UTC - in response to Message 22930.  
Last modified: 15 Jan 2022, 1:15:24 UTC

Put the following code into Notepad:
<app_config>
    <app_version>
        <app_name>camb_boinc2docker</app_name>
        <plan_class>vbox64_mt</plan_class>
        <avg_ncpus>8</avg_ncpus>
    </app_version>
</app_config>

Then save the file as "app_config.xml" in C:\ProgramData\BOINC\projects\www.cosmologyathome.org .

This will force the "camb_boinc2docker" Cosmology@home application to use no more than 8 virtual CPUs, or vCPUs for short. There is a bug in Cosmology@home where it will crash if it uses more than 8 vCPUs per task. I noticed that your machine has 6 physical cores, so it can provide up to 12 vCPUs thanks to its hyper-threading capability. The project's administrators should configure the project to use no more than 8 vCPUs per task to work around this bug.
ID: 22931 · Report as offensive     Reply Quote
.clair.

Send message
Joined: 4 Nov 07
Posts: 629
Credit: 12,068,402
RAC: 0
Message 22932 - Posted: 15 Jan 2022, 20:38:58 UTC - in response to Message 22931.  

The project's administrators should configure the project to use no more than 8 vCPUs per task to work around this bug.

This can be done .
Go to your `Account` look in `Project Preferences` and select the number of CPUs to use there.
If you want to run multiple work units at a time divide the number of cpu/threads to match.
You can also select the maximum number of jobs `Max#Jobs` same as `tasks in progress` limit.
ID: 22932 · Report as offensive     Reply Quote

Forums : Technical Support : MT VB units get stuck at 0.100% for 20 mins and then get aborted. CPU stayed idle.