> > That is for threads, isn't? I think you meant '1', or -1. thanks adr :)
> Sure: > > load averages: 6.17, 6.09, 5.94; up 67+01:27:45 15:47:04 > 65 processes: 6 runnable, 58 sleeping, 1 on CPU > CPU0: 54.8% user, 41.3% nice, 3.9% system, 0.0% interrupt, 0.0% idle > CPU1: 0.0% user, 0.1% nice, 7.3% system, 0.3% interrupt, 92.3% idle if you monitor this, does it change often? ie, we see that CPU0 is doing almost all the work in this snapshot, but over time, what's the trend? there's something weird going on with sched here. is hardclock() happening on all cpus properly (at hz rate)? i've seen very bad scheduler perf if that didn't happen many years ago (some change ad@ made in the netbsd-5 timeframe, IIRC), but it mostly lead to hangs. maybe since then it's turned into working less poorly not hangs :) i am not sure what sort of info we have about sched decisions in the recent time, but maybe there's something we can dump out. .mrg.
