Jukka Marin writes:
> On Mon, Mar 12, 2001 at 06:23:02PM -0600, Greg Oster wrote:
> > Sure... we can go through the details off-line, but basically just:
> >  1) archive the stuff you want to keep
> >  2) build a new disklabel w/ larger values for sectors/track and tracks/cyl
> inder.
> >     The ones I suggested before might work reasonably well.
> >  3) apply that disklabel to raid0
> >  4) newfs /dev/rraid0e
> >  5) copy the data back on...
> 
> Is
> 
>         :dt=RAID:se#512:ns#62:nt#64:sc#3968:nc#22699:\
> 
> much better than
> 
>         :dt=RAID:se#512:ns#62:nt#16:sc#992:nc#90796:\

I'm not sure it would make that much of a difference, but others who know FFS 
better may be able to comment more...

> or should I relabel and format my RAID1 before populating it with real data?

Benchmark it and see :)  If you have the time, newfs the RAID 1, and time the 
extraction of pkgsrc.tar.gz.  Then change the disklabel as above, newfs, 
and time the extraction again.  Pick the one that does the best :) 
(Of course, multiple timings would be more accurate, but whether you decide to 
do that depends on how scientific you want this experiment to be :) )
I'd be curious to see the results, especially if one they were quite a bit 
different...

> I won't be able to do that again too easily after putting some 20 GB of data
> on the disk..
> 
> Thanks,
> 
>   -jm

Later...

Greg Oster