Re: More Testing with 4000 disks

Andrew Morton (akpm@digeo.com)
Mon, 7 Apr 2003 14:15:51 -0700


Badari Pulavarty <pbadari@us.ibm.com> wrote:
>
> Hi,
>
> Here is more on testing with simulated 4000 disks.
>
> I mounted 4000 simulated (scsi_debug) disks. But
> when I tried to do IO on 200 of them (just reads),
> system is 100% busy. Here are the vmstat,
> 10-sec profile output and sysrq-t output.
>
> I am wondering why 100% sys ?
>

This doesn't make sense. You're showing 50% CPU time in schedule(), yet the
(undescribed) machine is doing only 11k context switches/sec.

How come?

What context switch rate and profile were you getting when performing IO against a
large number of real disks?

-
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/