Rosetta WUs often hang. on linux

Message boards : Number crunching : Rosetta WUs often hang. on linux

To post messages, you must log in.

AuthorMessage
dasy2k1

Send message
Joined: 18 Feb 07
Posts: 2
Credit: 747,661
RAC: 0
Message 48103 - Posted: 29 Oct 2007, 13:36:33 UTC

when im running my multi project boinc on linux quite oiften a rosetta WU will hang at a certian point (quite often after 59min 59 seconds)

it still says running in boincmgr but the cpu time and % complete stay the same and my computer system monitor shouws the rosetta thread as taking 0% of processor time

the boinc sceduler seems unable to start another project in its place.
it seems that when boinc trys to suspend the WU to give the time to somthing elce, roseta is not exiting cleanly


the only way to get that core in use again is to manually abort the rosetta WU.

any ideas of what is going worng and how to fix?
ID: 48103 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Luuklag

Send message
Joined: 13 Sep 07
Posts: 262
Credit: 4,171
RAC: 0
Message 48114 - Posted: 29 Oct 2007, 16:25:59 UTC - in response to Message 48103.  

when im running my multi project boinc on linux quite oiften a rosetta WU will hang at a certian point (quite often after 59min 59 seconds)

it still says running in boincmgr but the cpu time and % complete stay the same and my computer system monitor shouws the rosetta thread as taking 0% of processor time

the boinc sceduler seems unable to start another project in its place.
it seems that when boinc trys to suspend the WU to give the time to somthing elce, roseta is not exiting cleanly


the only way to get that core in use again is to manually abort the rosetta WU.

any ideas of what is going worng and how to fix?


have a look in the linux preemption thread.

but it comes of your pc cant handle to swap projects, wich will happen each 60 minutes. looking at the info you provided, you could say make it that way that 1 core only crunches rosetta and let the other core(s) share in different projects.
ID: 48114 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Rosetta WUs often hang. on linux



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