mirrored root disk boot block

[Bridge]

After mirroring a root disk, does it still need to install the boot
block on the mirrored disk?
There are some documentations mentioned step of installing boot
blocks, are they correct? As resync from the root disk to the mirrored
disk should have a copy of the boot block, is that right?

Thanks!

[Darren Dunham]

Bridge <bridge_xue@yahoo.com> wrote:

> After mirroring a root disk, does it still need to install the boot
> block on the mirrored disk?

What is 'it'?

In general, you do not have to do anything with the boot blocks. They
are part of the metadevice, and are mirrored along with the rest of the
metadevice.

> There are some documentations mentioned step of installing boot
> blocks, are they correct?

Which documentation?

> As resync from the root disk to the mirrored
> disk should have a copy of the boot block, is that right?

Depends on the technology, but ODS/SDS/SVM or even VxVM with a mirrored
root partition will behave this way.

--
Darren Dunham [email]ddunham@taos.com[/email]
Unix System Administrator Taos - The SysAdmin Company
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. >

[Andrew Gabriel]

In article <3e949365.0307270836.71229256@posting.google.com >,
[email]bridge_xue@yahoo.com[/email] (Bridge) writes:

> After mirroring a root disk, does it still need to install the boot
> block on the mirrored disk?
> There are some documentations mentioned step of installing boot
> blocks, are they correct? As resync from the root disk to the mirrored
> disk should have a copy of the boot block, is that right?

I don't know for sure, but SVM can't simply be copying all 32
sectors at the start of the disk into the mirror as that would
destroy the mirror's VToC. It might be that it doesn't sync
any of the first 32 sectors across, but I'm guessing.

On Solaris x86, the boot code doesn't live in the root filesystem,
so in that case you would certainly need to install it on the
mirrored drive if you ever wanted to boot directly from that drive.

--
Andrew Gabriel
Consultant Software Engineer

[Rob Stampfli]

In article <3e949365.0307270836.71229256@posting.google.com >,
Bridge <bridge_xue@yahoo.com> wrote:

>After mirroring a root disk, does it still need to install the boot
>block on the mirrored disk?
>There are some documentations mentioned step of installing boot
>blocks, are they correct? As resync from the root disk to the mirrored
>disk should have a copy of the boot block, is that right?

I asked this same question a few months back, and the answer appears
to be "no" (unnecessary) for Sparc architectures (presuming, of
course, that an installboot was previously run on the underlying FS
being mirrored) and "yes" for Intels. Of course, it shouldn't hurt
anything to run an installboot on the mirrored partition after the
fact.

On the Sparcs, installboot causes the boot loader to be copied to
consecutive sectors starting at sector 1 on the applicable media.
This is in an area of the slice that the UFS FS doesn't use, but
that Disksuite still manages. Thus, the loader is copied over
automatically when the new mirror is synced. (As an aside,
Disksuite does not manage physical block 0 of the disk, as this
contains the disk's VTOC which would not necessarily be identical
between disks and which would most definitely cause problems if
overwritten incorrectly. But, neither the UFS FS nor the boot
loader concern themselves with block 0.)

The Intels boot in a different way, and it is necessary to run
an installboot after the fact, as noted in the manual.

Rob

[Dennis Clarke]

> On the Sparcs, installboot causes the boot loader to be copied to
> consecutive sectors starting at sector 1 on the applicable media.

Makes me wonder if it is possible to boot a sparc system from a floppy.

just curious ... would be interesting .. useless but interesting, kinda like
fresh cut flowers. Nice to look at but just so much dead stuff in four days.

Dennis

[Rob Stampfli]

In article <Pine.GSO.4.53.0307271651000.6815@blastwave>,
Dennis Clarke <dclarke@blastwave.org> wrote:

>

>> On the Sparcs, installboot causes the boot loader to be copied to
>> consecutive sectors starting at sector 1 on the applicable media.

>
> Makes me wonder if it is possible to boot a sparc system from a floppy.
>
> just curious ... would be interesting .. useless but interesting, kinda like
> fresh cut flowers. Nice to look at but just so much dead stuff in four days.

Probably the main impediment would be that the Boot Prom has no
facility for reading a loader off a floppy disk in order to jump
to it and thus begin the boot sequence. I think you can boot off
tape, or at least could in the *Good Old Days* (TM).

Rob

[Darren Dunham]

Rob Stampfli <restamp@hotmail.com> wrote:

> In article <Pine.GSO.4.53.0307271651000.6815@blastwave>,

>> Makes me wonder if it is possible to boot a sparc system from a floppy.
>>
>> just curious ... would be interesting .. useless but interesting, kinda like
>> fresh cut flowers. Nice to look at but just so much dead stuff in four days.

> Probably the main impediment would be that the Boot Prom has no
> facility for reading a loader off a floppy disk in order to jump
> to it and thus begin the boot sequence. I think you can boot off
> tape, or at least could in the *Good Old Days* (TM).

Yes, you can certainly boot off tape. Of course to the hardware, that's
simply another scsi device that supports 512 byte block reads. There's
just no Solaris boot tape (that I know of).

Why is the floppy different? Just wasn't worth putting in the
(drivers?) necessary to do the work? That surprises me. I had just
assumed that the hardware would work if a sparc-compatible boot floppy
were created for some reason, but that appears to be false.

--
Darren Dunham [email]ddunham@taos.com[/email]
Unix System Administrator Taos - The SysAdmin Company
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. >

[Dennis Clarke]

> Why is the floppy different? Just wasn't worth putting in the
> (drivers?) necessary to do the work? That surprises me. I had just
> assumed that the hardware would work if a sparc-compatible boot floppy
> were created for some reason, but that appears to be false.

I am certain that it is possible, at least with the hardware. I have a SS20
with quad Ross procs running RedHat Linux 6.2 and it would boot with a boot
floppy. Without the floppy it would just sit there.

Makes me wonder if Solaris could/would do the same thing.

Still .. a useless curiosity.

Dennis

[Darren Dunham]

Dennis Clarke <dclarke@blastwave.org> wrote:

> I am certain that it is possible, at least with the hardware. I have a SS20
> with quad Ross procs running RedHat Linux 6.2 and it would boot with a boot
> floppy. Without the floppy it would just sit there.

Okay... you made me do my tests again, and I may just have a defective
floppy drive on the U-5 I used. I'll have to see what's up with it.

> Makes me wonder if Solaris could/would do the same thing.

I suppose it's theoretically possible, but it would be a lot of work
that I don't see anyone having a great need for.

--
Darren Dunham [email]ddunham@taos.com[/email]
Unix System Administrator Taos - The SysAdmin Company
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. >

[ultrasparc3@hotmail.com]

In <Pine.GSO.4.53.0307271651000.6815@blastwave> Dennis Clarke <dclarke@blastwave.org> writes:

> Makes me wonder if it is possible to boot a sparc system from a floppy.

OpenBoot supports booting from the floppy though there are bugs
on most of the newer sun4u systems. Solaris does not support being
booted from a floppy on SPARC but you are free to boot other standalone
programs.