On 24.12.02, 13:33:28, webmaster@datazap.net wrote:
> Hi,
> 
> On Sun, 22 Dec 2002, Bernd Sieker wrote:
> 
> > e. g. (assuming sd1a and sd1e are the new / and /usr filesystems,
> > respectively, and are already newfs'd.)
> >
> > # mount -o async /dev/sd1a /mnt
> > # mount -o async /dev/sd1e /mnt/usr
> > # cd /mnt
> > # dump -0 -f - / | restore -rf -
> > # cd /mnt/usr
> > # dump -o -f - /usr | restore -rf -
           ^^
This should be "-0", too, of course!

> > # umount /mnt/usr
> > # umount /mnt
> 
> This sounds pretty strait forward. Although, I would like to add a
> separate partition for /var. Right now I don't have one. Could I mount
> that partition, and use dump|restore just like you described above?

Yes, just mount it under /mnt/var . restore operates file-oriented
and will put the files from /var of the old root filesystem under
the correct subdirectories.

restore creates file called "restoresymtable" on the restored
filesystem, which you might want to remove if you have made sure that
the restore was successful.

> 
> I have a book ("Unix Hinks and Hacks") that gives several examples of
> scripts to do what it calls hard drive "burn-in" (although, I don't
> understand a one of them), and it makes it sound very important. But I
> don't think that have ever heard of anyone done this type of a test.

Yes, I though that this was really meant as a stress-test. Might be
useful.

> 
> If I have time I will probably do some kind of test on the hard
> drive, because it is a used drive.

If you can mount it in a PC, try IBM's "drive fitness test", which can
do iterated media/seek tests on the whole medium. Let it run for a night
and see if it shows errors. You can download it from www.storage.ibm.com
somewhere.

Afair it will also test drives of other manufacturers.

> 
> Thanks for your help, I really appreciate it,
> Al
> 

-- 
Bernd Sieker

NetBSD: We put the "Net" in "BSD", and took out the "Free"
		-- Tom Harvey