disk diagnostics

[Pieter Dumon]

You're right: softdep did the trick. Turned it on on the /usr/obj
filesystem, and the rm of the whole world_i386 of 1.7 preview took
about 10 seconds....
Thanks! I didn't expect soft updates to make such a huge difference.

Pieter

On 7/27/06, Joerg Sonnenberger <joerg at xxxxxxxxxxxxxxxxx> wrote:
On Thu, Jul 27, 2006 at 05:46:49PM +0800, Bill Hacker wrote:
> >0.070u 0.476s 20:11.87 0.0%     313+264k 7+54102io 0pf+0w
> >
>
> Am I wrong in interpreting that said act took 7 1/100's of a second of CPU
> time for itself, needed just under half a second of system overhead, but
> needed 20+ minutes end-to-end to complete by the wall-clock?
>
> 'To be determined' if it was awaiting I/O, or if something else was denying
> it a place to sit down and eat its hamburger.
That's exactly what I expected. It needs almost no processing in
userland, a little bit more in the kernel to push the IO through the
various layer and most time waiting for the ATA controller. When you
repeat the same process with softdep, it should be a lot faster, since
it can write out many requests ordered. It would be even faster with a
proper disk scheduler, but that's (still) a dream.
Joerg