Message boards : LHCb Application : Relative Efficiency of LHCb Tasks
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile ritterm
Avatar

Send message
Joined: 30 May 08
Posts: 93
Credit: 5,160,246
RAC: 0
Message 33159 - Posted: 29 Nov 2017, 15:07:36 UTC

I'm sorry if I've missed something elsewhere in the forums, but I wanted to ask a general question about what I'm seeing with LHCb jobs. I've noticed that recent tasks run a couple of jobs, wait ten minutes, and then shutdown. This seems very inefficient compared to CMS and Theory tasks (even though those jobs can have what seems to be hours of idle time). For example, here's part of the stderr output from Task 168175998:

2017-11-29 06:14:43 (15391): Guest Log: [INFO] LHCb application starting. Check log files.
2017-11-29 06:14:43 (15391): Guest Log: [DEBUG] HTCondor ping
2017-11-29 06:14:45 (15391): Guest Log: [DEBUG] 0
2017-11-29 06:16:06 (15391): Guest Log: [INFO] New Job Starting in slot1
2017-11-29 06:16:06 (15391): Guest Log: [INFO] Condor JobID: 5268.117 in slot1
2017-11-29 06:16:26 (15391): Guest Log: [INFO] Starting pilot in slot1
2017-11-29 06:18:59 (15391): Guest Log: [INFO] Job finished in slot1 with 700.
2017-11-29 06:19:26 (15391): Guest Log: [INFO] New Job Starting in slot1
2017-11-29 06:19:26 (15391): Guest Log: [INFO] Condor JobID: 5268.181 in slot1
2017-11-29 06:19:47 (15391): Guest Log: [INFO] Starting pilot in slot1
2017-11-29 06:21:17 (15391): Guest Log: [INFO] Job finished in slot1 with 700.
2017-11-29 06:31:30 (15391): Guest Log: [INFO] Condor exited with return value N/A.
2017-11-29 06:31:30 (15391): Guest Log: [INFO] Shutting Down.
2017-11-29 06:31:30 (15391): VM Completion File Detected.
2017-11-29 06:31:30 (15391): VM Completion Message: Condor exited with return value N/A.
ID: 33159 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Nov 14
Posts: 602
Credit: 24,371,321
RAC: 0
Message 33160 - Posted: 29 Nov 2017, 15:13:43 UTC - in response to Message 33159.  

I see LHCb jobs running from 18 minutes to 13 hours on an i7-4770. So it appears that it varies a lot.
ID: 33160 · Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 30 May 08
Posts: 93
Credit: 5,160,246
RAC: 0
Message 33161 - Posted: 29 Nov 2017, 16:01:21 UTC - in response to Message 33160.  

I see LHCb jobs running from 18 minutes to 13 hours on an i7-4770. So it appears that it varies a lot.

Yeah, I've had those long runners, too, but not in the last couple of days. Maybe my real question is about the nature or amount of the current LHCb work or how it's distributed.
ID: 33161 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 806
Credit: 651,667,947
RAC: 289,219
Message 33164 - Posted: 29 Nov 2017, 17:24:13 UTC

The shutdown after 10min is by design, the VM will try to get work from the CERN servers, if it is not able to get work after 10min it quits.

This way you only waste 10min. Of course if you chose to run only LHCb then you will not be doing any processing on your computer
ID: 33164 · Report as offensive     Reply Quote

Message boards : LHCb Application : Relative Efficiency of LHCb Tasks


©2024 CERN