Message boards : ATLAS application : maybe stuck at 100%
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile H&D

Send message
Joined: 12 Aug 05
Posts: 2
Credit: 456,631
RAC: 0
Message 43136 - Posted: 30 Jul 2020, 16:58:04 UTC

After a long time I started to elaborate some WU but I see that nothing has changed. The processing was supposed to last about 2 hours but instead it is lasting more than 24 and now it has been 100% for about 12 hours. Some advice?
ID: 43136 · Report as offensive     Reply Quote
Christoph

Send message
Joined: 2 Apr 20
Posts: 5
Credit: 5,571,567
RAC: 0
Message 43138 - Posted: 30 Jul 2020, 17:44:10 UTC - in response to Message 43136.  

Hi,
I'm also facing the same problem, some ATLAS tasks just don't finish. A "normal" task is typically finished in less than 2h in my case.

I decided to abort these tasks if they last > 10h because they are blocking 8 CPU cores.
ID: 43138 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 15 Jun 08
Posts: 2401
Credit: 225,576,756
RAC: 120,943
Message 43139 - Posted: 30 Jul 2020, 17:55:39 UTC - in response to Message 43136.  

So far there are no logfiles that can be checked.
If you have installed the VirtualBox Extension Pack you can
- open your BOINC Manager
- select an ATLAS task
- in the left column click on "show the VM console"

This opens a window that shows the VM's screens.
Press "ALT + F2" or "ALT + F3" to switch to "ATLAS Event Monitoring" or the VM's top output.


BOINC's runtime estimation is useless in case of LHC VMs since the BOINC client can't directly look into the VM.
ATLAS sends out tasks with shorter or longer runtimes depending on the physics that has to be calculated.
Single events can have runtimes between a few seconds and 1/2 hour and each task calculates 200 of them.
ID: 43139 · Report as offensive     Reply Quote
Profile H&D

Send message
Joined: 12 Aug 05
Posts: 2
Credit: 456,631
RAC: 0
Message 43140 - Posted: 30 Jul 2020, 17:56:17 UTC - in response to Message 43138.  
Last modified: 30 Jul 2020, 17:58:36 UTC

Thanks for the information.
ID: 43140 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 15 Jun 08
Posts: 2401
Credit: 225,576,756
RAC: 120,943
Message 43141 - Posted: 30 Jul 2020, 17:59:47 UTC - in response to Message 43138.  

8-core setups are known to be inefficient.
If enough RAM is available the #core setup should be limited to 4 or less.
ID: 43141 · Report as offensive     Reply Quote

Message boards : ATLAS application : maybe stuck at 100%


©2024 CERN