Abnormally high run times

Message boards : Number crunching : Abnormally high run times

To post messages, you must log in.

AuthorMessage
Daedalus

Send message
Joined: 1 Aug 08
Posts: 39
Credit: 10,106,899
RAC: 508
Message 91726 - Posted: 17 Feb 2020, 13:34:33 UTC
Last modified: 17 Feb 2020, 13:39:52 UTC

Hello,

I sometimes catch tasks which run for an unusually longer time. My run time is set to four hours and those show low completion rates even after longer hours. I kill them as soon as i see them.

Does anyone else have the same problem. Is there an explanation ?

Edit: They also guzzle huge amounts of RAM but sounds like some bug.
ID: 91726 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 91727 - Posted: 17 Feb 2020, 14:35:12 UTC

R@h has a "watchdog" function, which will shutdown a task if it runs for more than 4 hours (of CPU time) passed the target runtime. So, they will get cleaned up.

Sometimes when more memory is consumed, tasks do not get as much actual CPU time as they normally would. The properties of the task show you the actual CPU time the task has used (which will always be somewhat longer than the "wall-clock" time).

If you can let things clean themselves up, it becomes much more clear to the project team that the specific WU had problems running normally. Otherwise, if you end the task, it is marked as being ended by the user and more easily goes unnoticed.
Rosetta Moderator: Mod.Sense
ID: 91727 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Daedalus

Send message
Joined: 1 Aug 08
Posts: 39
Credit: 10,106,899
RAC: 508
Message 91728 - Posted: 17 Feb 2020, 15:11:39 UTC
Last modified: 17 Feb 2020, 15:14:45 UTC

Thank you for your very quick answer. :)

I think i understand the problem, but my computer is not a dump.

Edit: I don't know if that still happens but in the pasts i had had WU's running 12 hours instead of 4.
ID: 91728 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Abnormally high run times



©2024 University of Washington
https://www.bakerlab.org