Nathan,
Yes it was a typo though I have wanted a top that consolidated by cpuset.
What I was refering to was taskstats. getdelays gets to it via a socket.
/usr/include/linux/taskstats.h
>
>
> ----- "David Wright" <daw@xxxxxxx> wrote:
>
> > > OOC, did you find anything?
> >
> > No, Thanks for the info. With sles11 we now have cpu wait, disk wait,
> > swap wait, and memory reclaim wait.
>
> Hmmm ... I'm not sure which metrics you're refering to here. Are these
> general CPU utilisation metrics or are we still talking about tasksets
> (which are cgroups in kernel speak, no? - Documentation/cgroup/* in the
> kernel sources)? If not tasksets, do you know what the userspace/kernel
> interface for extracting these metrics is?
>
> thanks!
>
> --
> Nathan
>
|