unbalanced load
Michal Szymanski <[email protected]>
| Newsgroups | gmane.linux.smp |
|---|---|
| Message-ID | <[email protected]> |
Hi, On 2.6 SMP kernels (running on multi-cpu machines) one can see the effect called "Light but Unbalanced Load" by Rick Lindsley in his article "What's New in the 2.6 Scheduler?" (http://www.linuxjournal.com/article/7178) If, say, 3 processes run on a dual-cpu machine, one of them tends to be bound to one processor, taking 99.9% of cpu time, while two other run on the second processor, dividing its time more or less 50%-50% I think this might be considered reasonable (although if the tasks are not of the same user, it might be unfair) if all the tasks have the same 'nice level'. But, I have noticed that even if one of the tasks is niced and two other - not, it happens that it is the niced process that occupies one cpu alone, getting 99.9% of the processor time. This seems to be explicitly contrary to what nice-ing the jobs should do. Is this 'affinity' or 'cpu-binding' issue controllable somehow (by setting some kernel parameters, for example)? thanks in advance for hints, regards, Michal. -- Michal Szymanski (msz at astrouw dot edu dot pl) Warsaw University Observatory, Warszawa, POLAND - To unsubscribe from this list: send the line "unsubscribe linux-smp" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html