I\'ve been testing some new Java code that can pre-empt 97% of CPU for 2-3 minutes, after which its CPU usage drops to 10-20% but it oversubscribes RAM by 80% (I\'m running Fedora Core 6 on an 866 MHz box with 256 MB RAM and this program needs 400 MB at times).
During a series of these tests MalariaControl suspended itself because (surprise!) \"computing resources were overcommitted\" but failed to restart afterward, eventually logging the messages:
Result blabblah is 8.45 days overdue.
You may not get credit. Consider aborting it.
It only restarted after I rebooted following a kernel upgrade.
As SETI@Home, my other BOINC project, continued to run when foreground tasks returned to normal I consider that this may be a MalariaControl problem.
Is this a known issue?
Martin
|