Message boards : Theory Application : 300.06 Theory Simulation (native_theory) 0.385C
Message board moderation

To post messages, you must log in.

AuthorMessage
Jim1348

Send message
Joined: 15 Nov 14
Posts: 602
Credit: 24,371,321
RAC: 0
Message 43163 - Posted: 1 Aug 2020, 21:43:39 UTC

My latest Theory show that they are using 0.385 of a core.
For a while, I was running 2 of them per core on a 16 core Ryzen 2700, but this has straightened out now.

Is this a bug or a feature?
ID: 43163 · Report as offensive     Reply Quote
Greger

Send message
Joined: 9 Jan 15
Posts: 151
Credit: 431,596,822
RAC: 0
Message 43165 - Posted: 1 Aug 2020, 22:42:06 UTC - in response to Message 43163.  
Last modified: 1 Aug 2020, 22:48:48 UTC

Boinc probably get wrong value from config to theory and new wrapper could start if several are stated with lower value then full core.
I have experience before that theory had low usage it started a new task. Happen a few times when theory was mt task, But not seen that got that value in boinc manager. It could possibly be able to continue but could break in long run.

A restart of boinc service could change state but as run native that is not good.

Task would start a single pid and add up more to until it feed core it could be several process pending or stalled or python script is busy. You could probably monitor task running or check pstree or possibly get something in runrivet.log

Could probably be both.
ID: 43165 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Nov 14
Posts: 602
Credit: 24,371,321
RAC: 0
Message 43166 - Posted: 1 Aug 2020, 22:49:02 UTC - in response to Message 43165.  

Thanks. I just got 13 of them all together in a bunch, but now they are back to normal.
So it must have been a startup problem, as I had recently attached the machine.
ID: 43166 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,090,946
RAC: 103,877
Message 43501 - Posted: 14 Oct 2020, 18:12:00 UTC
Last modified: 14 Oct 2020, 18:13:10 UTC

native Theory with 0.5 CPU (Download 20 Minutes ago).
Nothing changed, last two days always 1 Cpu.
https://lhcathome.cern.ch/lhcathome/results.php?hostid=10666284
ID: 43501 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,090,946
RAC: 103,877
Message 43502 - Posted: 14 Oct 2020, 19:18:53 UTC - in response to Message 43501.  
Last modified: 14 Oct 2020, 19:20:51 UTC

successful finished (0.5 Cpu)
https://lhcathome.cern.ch/lhcathome/workunit.php?wuid=147458027
21:00:44 CEST +02:00 2020-10-14: cranky-0.0.32: [INFO] Pausing container Theory_2390-1107867-54_0.
21:00:44 CEST +02:00 2020-10-14: cranky-0.0.32: [WARNING] Cannot pause container as /sys/fs/cgroup/freezer/boinc/freezer.state not exists.
No more seen with less 1 Cpu.
ID: 43502 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,090,946
RAC: 103,877
Message 43504 - Posted: 15 Oct 2020, 6:58:33 UTC - in response to Message 43502.  

Now two with 0.5 Cpu are running well - one is a sherpa 2.2.4.
Only problem is output.tgz is not deleted in the slot-folder after finishing and uploading.
Boinc 7.16.6 from CentOS 7.7 (CentOS Linux 7 (Core) [3.10.0-1127.19.1.el7.x86_64|libc 2.17 (GNU libc)].
WCG is cleaning the slot-folder correct!
ID: 43504 · Report as offensive     Reply Quote

Message boards : Theory Application : 300.06 Theory Simulation (native_theory) 0.385C


©2024 CERN