Re: Mirror one 72 gig disk on another.

From: Juhan Leemet (juhan_at_logicognosis.com)
Date: 09/23/04 

Date: Wed, 22 Sep 2004 22:55:58 -0200

On Wed, 22 Sep 2004 07:47:29 -0400, ds wrote:
> "Juhan Leemet" <juhan@logicognosis.com> wrote in message
> news:pan.2004.09.22.06.04.47.813007@logicognosis.com...
>> On Tue, 21 Sep 2004 23:28:41 +0000, Darren Dunham wrote:
>> > ds <not@now.invalid> wrote:
[snippage]
>> >> prtvtoc /dev/rdsk/cAtBdCs2 | fmthard -s - /dev/rdsk/cXtYdZs2
>> >
>> >> Now, what *I* would like to know is, for SURE does this ALSO
>> >> make diskb bootable, i.e. copy the boot record?? My
>> >> understanding is that it does.
>> >
>> > No. All 'fmthard' does is to arrange the slices in the VTOC (rather
>> > like 'format' does).
>
> This is getting interesting (again ;-).
> Since prtvtoc | fmthard arranges the slices in the VTOC, like format
> does, does newfs then need to be applied to all slices of the
> pre-mirror diskb before mirroring? Or, does mirroring *also* create
> the ufs newfs slices?

As Darren said, newfs is not needed, and is a waste of time.

>> > Making a disk (actually a slice) bootable requires the creation of a
>> > bootable filesystem, and the loading of bootblocks. My assumption is
>> > that mirroring a bootable slice will create a bootable slice.
>>
>> Yes, Darren (and others) helped me out with mirroring on an Ultra2, and
>> using fmthard. I had to use the "installboot" command to make my mirror
>> bootable. In fact, first time, I forgot (oopsies!). Best test booting both
>> sides, before you have to rely on the alternate mirror boot.
>
> :-)
>
> Okay, if one forgets to use installboot before mirroring, can it be
> done after mirroring? And yet, isn't the bootblock part of the
> root slice (typcially s0)?

Well, you can run installboot on a disk even after the normal file system
has got contents on it. Why not on a mirror? I'm assuming that the metadb
is not wedged into that same space. I have it in a separate partition.

ISTR that is what I did. I forgot (er, or didn't know) the first time,
after creating my mirror on my Ultra2. Then I yanked disk0 and tested a
reboot... urk! Oh, yeah, put disk0 back in, run installboot, then tested
OK. That is yet another example to emphasize that you really have to test!
There's some poor unfortunate in a Linux group who's damaged a RAID1
mirror, can't boot (must not have tested?), and is running around causing
damage. At this point, poor fellow has probably toasted his data.
Expensive lesson in the school of life.

> I have read the Sun doc, and upon reflection, a lot of this is still
> not very clear. (Or, maybe I missed the nuances?)

Yeah, I have had to read it several times, too. Unlike some stuff from the
Dark Side which is "crazy making", tho, Sun information does come
together. I found that doing the RAID setup while following along with the
documentation (cheat ***) really brought things together (margin notes!). 

-- 
Juhan Leemet
Logicognosis, Inc.