From: bigby on
Hi all

I have knowledge of Linux but not much in Windows.

I would like to set up a home server with RAID-1 (2 disks) in Windows,
including boot, hence I need to leverage the fakeraid in the mainboard
(mainboard which I still have to buy) (*)

All will be nice except if/when the mainboard breaks.

So I would really like if someone could tell me that in case of failure
I will be able to replace the broken mainboard with a different
mainboard with a different controller, and that the new controller will
be able to at least read one of the two disks of the array like if it
was a single non-raid disk.

For this to happen, the metadata of fakeraid raid1 needs to be at the
end of the disk. Otherwise finding the partitions (skipping the metadata
at the beginning) would require a difficult and uncertain low-level
hacking and then MBR fixing.

So do you know if this is the case for usual fakeraids?

AND THEN:

After successfully booting with 1 disk only, I would like to be able to
even recreate the raid1 on the (different) fakeraid controller of the
second mainboard.

For this to be possible I need the fakeraid to allow me create the raid1
while keeping the data of at least one of the two disks. I.e. if the
first operation of raid-1 creation is filling the two disks with zeroes,
I am doomed.

So is this standard, for what you have seen?

Also in order to be able to port raid1's to another fakeraid, I guess I
need to leave enough space after the last partition so to accommodate
every possible fakeraid raid1 metadata. This is because if the first
mainboard's fakeraid takes only 4K for metadata but the second mainboard
takes 100MB, and my last partition touches the end of the disk after the
first raid1 creation, during the creation of the raid1 on the second
mainboard the first 99.99MB of the metadata would overwrite my last
partition.

So do you have an idea of how much space should I leave free at the end?


(*) or do you know if Windows (any version) allows software raid-1
starting from boot-time? I don't think so huh? If yes, please can you
point me to a web tutorial/information?


Thank you in advance
From: Arno on
bigby <bigby(a)nowhere.org> wrote:
> Hi all

> I have knowledge of Linux but not much in Windows.

> I would like to set up a home server with RAID-1 (2 disks) in Windows,
> including boot, hence I need to leverage the fakeraid in the mainboard
> (mainboard which I still have to buy) (*)

> All will be nice except if/when the mainboard breaks.

> So I would really like if someone could tell me that in case of failure
> I will be able to replace the broken mainboard with a different
> mainboard with a different controller, and that the new controller will
> be able to at least read one of the two disks of the array like if it
> was a single non-raid disk.

That cpuld be problematic. But if you have Linux competence,
you can do something else: If the fake-raid is supported by
dm-raid (Linux Fakeraid driver), you can make a sector image
of the array to a third disk (or to the second one as
the new fakeraid with a different dm-raid setting if you are
daring). I really would recommend using a third disk, so you
still have a backup after adding the second disk to the new
raid.

> For this to happen, the metadata of fakeraid raid1 needs to be at the
> end of the disk. Otherwise finding the partitions (skipping the metadata
> at the beginning) would require a difficult and uncertain low-level
> hacking and then MBR fixing.

> So do you know if this is the case for usual fakeraids?

Usuallu not, because they are intentionally incompatible,
vendor lock-in and all that business "optimization".

> AND THEN:

> After successfully booting with 1 disk only, I would like to be able to
> even recreate the raid1 on the (different) fakeraid controller of the
> second mainboard.

> For this to be possible I need the fakeraid to allow me create the raid1
> while keeping the data of at least one of the two disks. I.e. if the
> first operation of raid-1 creation is filling the two disks with zeroes,
> I am doomed.

> So is this standard, for what you have seen?

No. As I said, vendors try to intentionally make this hard
or impossible.


> Also in order to be able to port raid1's to another fakeraid, I guess I
> need to leave enough space after the last partition so to accommodate
> every possible fakeraid raid1 metadata. This is because if the first
> mainboard's fakeraid takes only 4K for metadata but the second mainboard
> takes 100MB, and my last partition touches the end of the disk after the
> first raid1 creation, during the creation of the raid1 on the second
> mainboard the first 99.99MB of the metadata would overwrite my last
> partition.

> So do you have an idea of how much space should I leave free at the end?

Not a lot. Leave 1GB off to be really, really safe.


> (*) or do you know if Windows (any version) allows software raid-1
> starting from boot-time? I don't think so huh? If yes, please can you
> point me to a web tutorial/information?

I think windows has this "dynamic disk" concept in the professional
versions, but I have no experience with it and do not know whether
you can reliably boot from it with one disk gone or at all.

I think you should investigate dm-raid and in case the thing breaks
image over with it. Image-copies of windows typically work, at
least I have never had problems imaging my c:\ with linux
or restoring it.

I do not know whether dm-raid has a webpage, but
google(dmraid) gives you a lot og info. Also the
source is here

http://people.redhat.com/~heinzm/sw/dmraid/src/

and probably contains embedded documentation.

Arno
--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: arno(a)wagner.name
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans
From: Bob Willard on
bigby wrote:
> Hi all
>
> I have knowledge of Linux but not much in Windows.
>
> I would like to set up a home server with RAID-1 (2 disks) in Windows,
> including boot, hence I need to leverage the fakeraid in the mainboard
> (mainboard which I still have to buy) (*)
>
> All will be nice except if/when the mainboard breaks.
>
> So I would really like if someone could tell me that in case of failure
> I will be able to replace the broken mainboard with a different
> mainboard with a different controller, and that the new controller will
> be able to at least read one of the two disks of the array like if it
> was a single non-raid disk.
>
> For this to happen, the metadata of fakeraid raid1 needs to be at the
> end of the disk. Otherwise finding the partitions (skipping the metadata
> at the beginning) would require a difficult and uncertain low-level
> hacking and then MBR fixing.
>
> So do you know if this is the case for usual fakeraids?
>
> AND THEN:
>
> After successfully booting with 1 disk only, I would like to be able to
> even recreate the raid1 on the (different) fakeraid controller of the
> second mainboard.
>
> For this to be possible I need the fakeraid to allow me create the raid1
> while keeping the data of at least one of the two disks. I.e. if the
> first operation of raid-1 creation is filling the two disks with zeroes,
> I am doomed.
>
> So is this standard, for what you have seen?
>
> Also in order to be able to port raid1's to another fakeraid, I guess I
> need to leave enough space after the last partition so to accommodate
> every possible fakeraid raid1 metadata. This is because if the first
> mainboard's fakeraid takes only 4K for metadata but the second mainboard
> takes 100MB, and my last partition touches the end of the disk after the
> first raid1 creation, during the creation of the raid1 on the second
> mainboard the first 99.99MB of the metadata would overwrite my last
> partition.
>
> So do you have an idea of how much space should I leave free at the end?
>
>
> (*) or do you know if Windows (any version) allows software raid-1
> starting from boot-time? I don't think so huh? If yes, please can you
> point me to a web tutorial/information?
>
>
> Thank you in advance

Bad thinking. You need a comprehensive backup procedure (meaning a
program plus some practices) to cover all of the events that could
corrupt all HDs in a RAIDset. And, with good backups, a failed MB
would need no more than MB replacement followed by restoring the
content of the RAIDset. (And, if the new MB is different from the
old MB, a repair/install of Windows will probably be needed; simple,
if you have a valid disk for the OS.)

Backups will also work if you decide to replace the HDs in your
RAIDset, to increase capacity or speed; or if you decide to
upgrade the MB or CPU or RAM or OS or ...

RAID is not a replacement for backup. Period.
--
Cheers, Bob
From: bigby on
Bob Willard wrote:
> bigby wrote:
> Backups will also work if you decide to replace the HDs in your
> RAIDset, to increase capacity or speed; or if you decide to
> upgrade the MB or CPU or RAM or OS or ...
>
> RAID is not a replacement for backup. Period.

RAID is a replacement for RAID!

I have never asked a backup. You read wrong.

I already have a backup for data but that takes lots of time for
restoring if the OS is gone. So I also wanted the RAID.

Do you know a type of backup that will reinstall all installed applications?
From: David Brown on
bigby wrote:
> Bob Willard wrote:
>> bigby wrote:
>> Backups will also work if you decide to replace the HDs in your
>> RAIDset, to increase capacity or speed; or if you decide to
>> upgrade the MB or CPU or RAM or OS or ...
>>
>> RAID is not a replacement for backup. Period.
>
> RAID is a replacement for RAID!
>
> I have never asked a backup. You read wrong.
>
> I already have a backup for data but that takes lots of time for
> restoring if the OS is gone. So I also wanted the RAID.
>
> Do you know a type of backup that will reinstall all installed
> applications?

There are dozens of disk cloners and system restore disk utilities that
could help here.

RAID only protects you against hard disk failure, nothing else. While
hard disk failures do happen, in my experience the great majority of
problems on Windows disks are either caused by users or by software
(including malware). Having your two disks raid'ed together just means
that these errors will be faithfully copied to both disks.