Messages in this thread | | | Subject | Re: [RFC PATCH] sched: START_NICE feature (temporarily niced forks) (v3) | From | Mike Galbraith <> | Date | Wed, 15 Sep 2010 15:12:46 +0200 |
| |
On Wed, 2010-09-15 at 11:22 +0200, Ingo Molnar wrote: > * Mike Galbraith <efault@gmx.de> wrote: > > > Tangent: cgroups with classification user/pgid > > > > taskset -c 3 ./wakeup-latency& sleep 30 && killall wakeup-latency > > (taskset -c 3 make -j10 running) > > > > maximum latency: 256.5 µs > > average latency: 25.7 µs > > missed timer events: 0 > > > > Rather spiffy for this case. > > Ouch, nothing beats proper segregation of workloads. > > It would be _really_ nice to get that automatically somehow, for make -j > jobs. Perhaps a per tty auto-cgroup-classification kernel feature? > > We cannot really rely on distros to pull this off.
Dunno, wasn't hard.
I'm no good at scripting, but got mine cobbled together in fairly short order. With it all in userland, I can turn pgid classification on/off at the drop of a hat, and/or launch self reaping cgroups with shares set ala nice level with another script. Easy and flexible.
(I usually don't have cgroups enabled, cobbled pgid thing together to see if userland could deal with it easily enough.. seems so, if little ole /me can do it, anybody can)
-Mike
-- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |