Cannot reactivate dynamic disk (offline)

Hello,

I have the following problem:

A few days ago Windows (2003) started to resync my dynamic raid 5 volume,
because I made a mistake (removed a wrong disk, but immediatly putting it
back in). Unfortunately at about 15% in the synchronization process, windows
crashed. After a reboot all drives are suddenly marked as Offline and I
can't get them back in an Online Mode. I tried both the disk managment snap
in ('reactivate disk') and diskpart ('online' command) without success. I
don't get an error message, but the disk is still in offline mode.

The disk data is still there. I can access them without problems in various
recovery tools, that work directly on the physical device(s). There are also
no physical errors on the disks and the event log shows no errors when I try
to activate a disk. There is also no problem with the controller itself, I
can access other disks on the same controller without problems.

I am quite confident that if I just get windows to accept the disks and put
them in online mode, that they'll work again with no problem. I just can't
figure out how to do that.

Any ideas?

Thanks,
Thomas Krause

Hi, Thomas

Did you get an error or anything before the crash, or was it a real
"crash-freeze" thing?

--Malu



"Thomas Krause" <forum.thomas.krause@gmx.de> wrote in message
news:5931AC3B-929E-4D54-80A6-022725602A6F@microsoft.com...[color=blue]
> Hello,
>
> I have the following problem:
>
> A few days ago Windows (2003) started to resync my dynamic raid 5 volume,
> because I made a mistake (removed a wrong disk, but immediatly putting it
> back in). Unfortunately at about 15% in the synchronization process,
> windows crashed. After a reboot all drives are suddenly marked as Offline
> and I can't get them back in an Online Mode. I tried both the disk
> managment snap in ('reactivate disk') and diskpart ('online' command)
> without success. I don't get an error message, but the disk is still in
> offline mode.
>
> The disk data is still there. I can access them without problems in
> various recovery tools, that work directly on the physical device(s).
> There are also no physical errors on the disks and the event log shows no
> errors when I try to activate a disk. There is also no problem with the
> controller itself, I can access other disks on the same controller without
> problems.
>
> I am quite confident that if I just get windows to accept the disks and
> put them in online mode, that they'll work again with no problem. I just
> can't figure out how to do that.
>
> Any ideas?
>
> Thanks,
> Thomas Krause[/color]

Hi Thomas,

if you're using the native dynamic disk feature implemented in W2K3, you
should open a CMD shell and run dmdiag -v.
This tool is part of the Support Tools, you could find them on any W2K3 CD
in the Support folder.

Dmdiag doesn't fix anything, it simply reads data from the private region of
the disk.
There's a MS internal tool named dmpss.exe which could perform maintenance
tasks on the private disk region.

This tool isn't public, not well documented and does require cryptic inputs,
so you SHOULD involve MS Support to help you with this issue.

"Thomas Krause" wrote:
[color=blue]
> Hello,
>
> I have the following problem:
>
> A few days ago Windows (2003) started to resync my dynamic raid 5 volume,
> because I made a mistake (removed a wrong disk, but immediatly putting it
> back in). Unfortunately at about 15% in the synchronization process, windows
> crashed. After a reboot all drives are suddenly marked as Offline and I
> can't get them back in an Online Mode. I tried both the disk managment snap
> in ('reactivate disk') and diskpart ('online' command) without success. I
> don't get an error message, but the disk is still in offline mode.
>
> The disk data is still there. I can access them without problems in various
> recovery tools, that work directly on the physical device(s). There are also
> no physical errors on the disks and the event log shows no errors when I try
> to activate a disk. There is also no problem with the controller itself, I
> can access other disks on the same controller without problems.
>
> I am quite confident that if I just get windows to accept the disks and put
> them in online mode, that they'll work again with no problem. I just can't
> figure out how to do that.
>
> Any ideas?
>
> Thanks,
> Thomas Krause
>[/color]

Hi,

I don't remember seeing any error before the crash, but this is already
several weeks ago, so my memory may be flawed.

In the meantime I managed to recover my data using a combination of "Raid
Reconstructor" and "Captain Nemo" from Runtime Software, which allowed me to
copy all data from the raid to other disks.

After restoring the data I experimented a little bit more and I found that
when first starting the disk managment console, I see disk read errors in
the event log. However the errors are always on the same 2 sectors and they
occur on every disk in the raid volume, which seemed strange to me. I tried
accessing the sector manually with another tool, but I found that the sector
is outside of the disks range. So the read error seems to be an out of range
error.

My theory is that somehow the ldm header (which is used by dynamic disks) at
the beginning of the disks got damaged. In this ldm header there is a
pointer to the start of the ldm database, which is stored at the end of the
disk. If this pointer points to an sector outside the range of the disc,
windows tries to read the sector but gets an read error. It then tries one
disc after another and always gets the same result. It will then mark the
disc at offline, because it can't find the ldm database.

I tried locating the ldm database myself by searching the end of the disk
for specific patterns that occur in the database (header names, etc.), but
have not had success. It was probably overridden during the synchronization
process or later on.

Since I already recovered my data, I saw no point in further researching
this issue. I believe the only way to recover such a volume would be to
write a program that creates the necessary ldm header and database
information from manual input and writes them to the disk. Unfortunately it
seems that there is no program that can repair or recreate ldm information.
The information on what to write where is available thanks to linux hackers,
that reverse engineered the ldm implementation in windows, but it seems
there are not enough people that need such a tool. I thought of writing that
myself, but since I got my data back I don't really need it anymore.

I believer after this experience and numerous other problems I had with
dynamic volumes over the years, I will invest in a good hardware raid 5
controller from 3ware/amcc. With windows I now feel less safe with software
raid 5 than with single disks without any redundancy :-(

Thanks,
Thomas Krause

"Malu Menezes" <mariamen@nonono.microsoft.com> wrote:[color=blue]
> Hi, Thomas
>
> Did you get an error or anything before the crash, or was it a real
> "crash-freeze" thing?
>
> --Malu
>
>
>
> "Thomas Krause" <forum.thomas.krause@gmx.de> wrote in message
> news:5931AC3B-929E-4D54-80A6-022725602A6F@microsoft.com...[color=green]
>> Hello,
>>
>> I have the following problem:
>>
>> A few days ago Windows (2003) started to resync my dynamic raid 5 volume,
>> because I made a mistake (removed a wrong disk, but immediatly putting it
>> back in). Unfortunately at about 15% in the synchronization process,
>> windows crashed. After a reboot all drives are suddenly marked as Offline
>> and I can't get them back in an Online Mode. I tried both the disk
>> managment snap in ('reactivate disk') and diskpart ('online' command)
>> without success. I don't get an error message, but the disk is still in
>> offline mode.
>>
>> The disk data is still there. I can access them without problems in
>> various recovery tools, that work directly on the physical device(s).
>> There are also no physical errors on the disks and the event log shows no
>> errors when I try to activate a disk. There is also no problem with the
>> controller itself, I can access other disks on the same controller
>> without problems.
>>
>> I am quite confident that if I just get windows to accept the disks and
>> put them in online mode, that they'll work again with no problem. I just
>> can't figure out how to do that.
>>
>> Any ideas?
>>
>> Thanks,
>> Thomas Krause[/color]
>[/color]

Thanks for the tip. The dmdiag tool and dmpss.exe are interesting tools...

If I dump the ldm configuration to a file using dmpss, I can see that the
database is completely zeroed out. The only thing intact is the backup ldm
header at sector 6 in the disk. However without the information in the ldm
database this will be of little help...

I should theoretically be able to make a new software raid 5 using identical
hard disks and then copy the ldm database from the new raid to the old raid
one disk at a time. Of course the ordering of the disks must be 100% correct
for this to work. However as I mentioned in my other post, I was able to
recover my data using other tools, so I don't really need this anymore...

Thanks,
Thomas Krause

"Ralf Kastl" <RalfKastl@discussions.microsoft.com> wrote:[color=blue]
> Hi Thomas,
>
> if you're using the native dynamic disk feature implemented in W2K3, you
> should open a CMD shell and run dmdiag -v.
> This tool is part of the Support Tools, you could find them on any W2K3 CD
> in the Support folder.
>
> Dmdiag doesn't fix anything, it simply reads data from the private region
> of
> the disk.
> There's a MS internal tool named dmpss.exe which could perform maintenance
> tasks on the private disk region.
>
> This tool isn't public, not well documented and does require cryptic
> inputs,
> so you SHOULD involve MS Support to help you with this issue.
>
> "Thomas Krause" wrote:
>[color=green]
>> Hello,
>>
>> I have the following problem:
>>
>> A few days ago Windows (2003) started to resync my dynamic raid 5 volume,
>> because I made a mistake (removed a wrong disk, but immediatly putting it
>> back in). Unfortunately at about 15% in the synchronization process,
>> windows
>> crashed. After a reboot all drives are suddenly marked as Offline and I
>> can't get them back in an Online Mode. I tried both the disk managment
>> snap
>> in ('reactivate disk') and diskpart ('online' command) without success. I
>> don't get an error message, but the disk is still in offline mode.
>>
>> The disk data is still there. I can access them without problems in
>> various
>> recovery tools, that work directly on the physical device(s). There are
>> also
>> no physical errors on the disks and the event log shows no errors when I
>> try
>> to activate a disk. There is also no problem with the controller itself,
>> I
>> can access other disks on the same controller without problems.
>>
>> I am quite confident that if I just get windows to accept the disks and
>> put
>> them in online mode, that they'll work again with no problem. I just
>> can't
>> figure out how to do that.
>>
>> Any ideas?
>>
>> Thanks,
>> Thomas Krause
>>[/color][/color]