Where are Disc Identities held ? What changes them ?

[alan-0000]

Is the Disk ID (e.g. 000FA830 or baf784e7-6bbd-4cfb-aaac-e86c96e166ee) held in the MBR or in the Disc Controller Firmware or where ?

One day I booted up and both my WDC and SamSung HDD's were down and Windows was in a bad way.

The Samsung was down because the WDC ID value had been changed and became a duplicate of the SamSung ID
Disk ID: 000FA830
so Windows put the Samsung Offline - Windows never misses a chance to make a bad situation worse

The WDC was down because its ID was that of an MBR style HDD, but its entire life has been spent as a GPT style disk with an ID such as
Disk ID: baf784e7-6bbd-4cfb-aaac-e86c96e166ee
hence the first partition was reported as RAW DATA and the subsequent partitions were all seen be windows as continuous Unallocated Space.

Windows always correctly reports which SATA channels each HDD and SSD is connected to,
BUT gets the Disk numbers wrong.

There is a race hazard.

When I boot up from a shut-down condition the numbers are always
0 WDC HDD
1 SamSung HDD
2 OCZ SSD

When I choose to Restart the OCZ SSD is always either Disk 1 or even Disk 0,
and the two HDD's drop down to make way.

On the day of disaster I Booted into the BIOS to look at various settings without making any changes before I let the BIOS launch Windows,
and that immediately resulted in chaos.

I assume that The Disk Numbers are affected by how long the various drives have been powering up whilst the BIOS is in control before Windows is launched,
and my sniffing around looking at settings in the BIOS has permitted an especially weird sort of chaos such as
reading the UNIQUEID of from Disk 0 which happened to be 000FA830 because the Samsung was there
and subsequently accessing Disk 0 finding baf784e7-6bbd-4cfb-aaac-e86c96e166ee because the WDC was there
and deciding to fix the error by correcting it to 000FA830.

I assume that the Disk number 0,1, or 2 is purely a Windows thing and is NOT determined by the BIOS, and therefore Windows is solely responsible.

That is my best guess - please tell me if there is anything else I should consider.

QUESTION :-
To minimize the risk of future chaos with variable Disk numbers,
should I always ensure a total shut-down for at least 10 seconds before booting into Windows,
especially after changing, or even pausing to look at, BIOS settings ?

Regards
Alan

[hdsentinel]

I can confirm that these IDs are NOT stored in the disk controller. These IDs are related to logical volumes (partitions) defined on the actual physical hard disks.
Yes, Windows uses these to identify the partitions / volumes to find out which should be used during the boot stage, regardless of actual hard disk position. These are stored in the partition table of the corresponding physical hard disk.

As you can see, sometimes the identification does not work as should - and in this situation, the system may not boot or may not be able to properly identify the volume on the actual physical hard disk.
This may happen (for example) if you clone the hard disk and use both the original and the new (cloned) drive as then the same IDs (which should be unique) found on both physical hard disks. This can cause confusion as then the OS can't differentiate the volumes.

The disk numbers (0-1-2-etc...) are independent from the BIOS, these are assigned by Windows, depending on the order how it can detect the actual hard disk controllers and drives on these controllers.

In the BIOS setup, you may check the "Plug and Play OS installed" setting which should be YES, allowing Windows to allocate resources of the hard disk controllers and devices. Also you may consider to set "Hard disk pre-start delay" or something similar, which allows some seconds before booting to spin up the hard disks properly. These settings may affect how the OS can detect and determine the order of the devices.

[alan-0000]

Thanks for the information on the BIOS settings which I have written with pen and paper.
I really wish I could access the Internet and this forum whilst within the BIOS

I will look in the BIOS after a good night's sleep.

To be sure we are on the same page regarding Disk and Partition Identities :-

When I launch CMD.EXE and run the command MOUNTVOL it lists

\\?\Volume{6496ab7f-6b8e-11e2-8a81-001e8c888f33}\
H:\

\\?\Volume{6496ab85-6b8e-11e2-8a81-001e8c888f33}\
D:\

\\?\Volume{45d0885e-6c47-11e2-a0a8-001e8c888f33}\
E:\

\\?\Volume{45d08866-6c47-11e2-a0a8-001e8c888f33}\
R:\

\\?\Volume{7374db70-7389-11e2-8a06-001e8c888f33}\
I:\

\\?\Volume{6496ab7a-6b8e-11e2-8a81-001e8c888f33}\
J:\

\\?\Volume{b8982f16-6a92-11e1-9d3e-806e6f6e6963}\
C:\

Partition C:\ is on my OCZ SSD
All the others are on my Samsung HDD
Do I correctly understand that the volume ID 6496ab7a-6b8e-11e2-8a81-001e8c888f33 is held in the partition table for J:\ (primary partition)
and the many other volume ID's ???????????-????-????-????-001e8c888f33 are held in the various Logical partition tables on the Samsung ?

When CMD.EXE launches DISKPART and I select the Samsung HDD and invoke UNIQUEID it reports for the entire Samsung
Disk ID: 58FBC9FC
Is that value written in the Samsung MBR ?

I never ran UNIQUEID on my WDC HDD when it was a healthy GPT style disk, but I assume that typically it would report for the entire WDC HDD
Disk ID: baf784e7-6bbd-4cfb-aaac-e86c96e166ee.
Would that have been written in the protective MBR or somewhere else ?

Windows damaged the GPT partitions and now interprets the disk as MBR style holding RAW DATA in the first partition - and the rest as unallocated space.
Now UNIQUEID reports the WDC as
Disk ID: 000FA830
Is this (wrong) value in the (protective) MBR ?

Regards
Alan

[alan-0000]

The disk numbers (0-1-2-etc...) are independent from the BIOS, these are assigned by Windows, depending on the order how it can detect the actual hard disk controllers and drives on these controllers.

In the BIOS setup, you may check the "Plug and Play OS installed" setting which should be YES, allowing Windows to allocate resources of the hard disk controllers and devices.

If it is Windows that causes the chaos of disk numbers being affected by race hazards,
it appears to me counter-intuitive to give Windows further scope for more chaotic control

Please explain how changing from NO to YES may give benefit.

I am using AMIBIOS version 2301 Dated 01/27/11
Under ADVANCED -> PCIPnP is
PLUG and PLAY O/S [NO]
With the explanation that
[NO] = the BIOS configures all Devices
[YES] = Allows the O/S to configure Non-Boot devices that are not needed for BOOT
Is this the setting you suggest I change

N.B. This Advanced section has a warning that any errors will prevent booting.

Also you may consider to set "Hard disk pre-start delay" or something similar, which allows some seconds before booting to spin up the hard disks properly. These settings may affect how the OS can detect and determine the order of the devices.

I could not find anything related to this in the BIOS.

Regards
Alan

[hdsentinel]

To be honest, I'm not really sure about the UNIQUEID command you mentioned ...

I can confirm the longer values like "\\?\Volume{6496ab7f-6b8e-11e2-8a81-001e8c888f33}\" are the GUID (globally unique identifier) of the volumes. These are associated with the volumes when created and they should never change.

The smaller IDs (58FBC9FC) may be the "volume serial number" you can see when you use the simple DIR command (and displayed in Hard Disk Sentinel when you select the partition in the bottom left area). This is also related to the logical volume (partition) but can change easily. This is not used in the boot procedure, only the other GUID.

From here, it is hard to tell what caused that Windows overwrite the original partition table and thus making the partition(s) unaccessible

[hdsentinel]

> Please explain how changing from NO to YES may give benefit.

This depends on the hard disk controllers where the drives installed.
Even if all drives are connected to the motherboard, it may have different disk controllers, for example simple ICHxx chipset, JMicron / Marvell controller and so. Usually these are marked with different colors.

The problem may be related that one or more hard disk controllers are not properly initialised in the BIOS and/or a device conflict may happen if Windows need to allocate different resources for the device. This is why it may be better to leave the not-required devices unavailable during the boot stage and let Windows allocate them. This may result that the hard disks/SSDs connected to them will have more stable numbers.

If you encounter problems during booting - you can switch back the setting without problems.


> I could not find anything related to this in the BIOS.

I'm sorry - most BIOSes have such or similar setting but not all BIOSes contain this.
Just an idea as it could help.

[alan-0000]

Thanks for the answers.

> To be honest, I'm not really sure about the UNIQUEID command you mentioned

This UNIQUEID is part of DISKPART which is the most User UNfriendly command line tool I have ever encountered.
These give information on DISKPART and UNIQUEID
http://support.microsoft.com/kb/300415
http://technet.microsoft.com/en-us/libr ... 10%29.aspx

When I launch CMD.EXE via "Run as Administrator" I can do things that Windows Disk Management is incapable off,
and this is how it presents UNIQUEID- after which DIR shows Partition serial numbers :-

Code: Select all

Microsoft Windows [Version 6.1.7601]
Copyright (c) 2009 Microsoft Corporation.  All rights reserved.

C:\Users\Alan>DISKPART

Microsoft DiskPart version 6.1.7601
Copyright (C) 1999-2008 Microsoft Corporation.
On computer: ALAN-DESKTOP

DISKPART> LIST DISK

  Disk ###  Status         Size     Free     Dyn  Gpt
  --------  -------------  -------  -------  ---  ---
  Disk 0    Offline         596 GB   571 GB
  Disk 1    Online          931 GB    25 GB
  Disk 2    Online           55 GB      0 B

DISKPART> SELECT DISK 1

Disk 1 is now the selected disk.

DISKPART> UNIQUEID DISK

Disk ID: 58FBC9FC

DISKPART> exit

Leaving DiskPart...

C:\Users\Alan>

C:\Users\Alan>DIR E:\o*
 Volume in drive E is E_SAM_E
 Volume Serial Number is 1894-33B0

 Directory of E:\
File Not Found

C:\Users\Alan>DIR D:\o*
 Volume in drive D is D_SAM_DATA_D
 Volume Serial Number is 5902-98C0

 Directory of D:\
File Not Found

As shown above my 931 GB Samsung HDD now has a Disk ID of 58FBC9FC
and this is unrelated to the serial numbers for Partitions D:\ and E:\ on this same Samsung HDD.

The UNIQUEID Disk ID applies to each physical HDD/SSD.
When a Disk is cloned the duplicate is given the same ID,
after which the computer should be shutdown and one of the two duplicates must be removed before a restart,
otherwise Windows goes ballistic because it cannot handle two disks with the same ID.
I do not know if Linux is more tolerant.

Originally all was well when Samsung had a Disk ID of 000FA830,
and the WDC HDD had a drastically different Disk ID.
But when there was a weird start-up session the Samsung ID of 000FA830 was copied to replace the WDC ID,
and a fraction of a second later I guess it was Windows that saw two off HDD with the same Disk ID and it chose to put the SamSung Off-Line.
It was happy to persist through several reboots with two disks using the same Disk ID so long as one was Off-Line.

With hindsight I recognise I should have put the WDC Off-Line as well and THEN put the Samsung On-Line and Samsung would have retained its Disk ID,
but instead I turned Samsung On-Line and Windows immediately declared a Disk ID conflict and changed its ID to 58FBC9FC.

> This depends on the hard disk controllers where the drives installed.
> Even if all drives are connected to the motherboard, it may have different disk controllers

I assume it is one controller - I have extracted this information from your HDS Pro v 4.30
[quote]
Hard Disk Number . . . . . . . . . . . . . . . . : 0
Interface . . . . . . . . . . . . . . . . . . . : S-ATA II
Disk Controller . . . . . . . . . . . . . . . . : Standard AHCI 1.0 Serial ATA Controller (AHCI)
Disk Location . . . . . . . . . . . . . . . . . : Channel 1, Target 0, Lun 0, Device: 0
Hard Disk Model ID . . . . . . . . . . . . . . . : WDC WD6401AALS-00L3B2

Hard Disk Number . . . . . . . . . . . . . . . . : 1
Interface . . . . . . . . . . . . . . . . . . . : S-ATA II
Disk Controller . . . . . . . . . . . . . . . . : Standard AHCI 1.0 Serial ATA Controller (AHCI)
Disk Location . . . . . . . . . . . . . . . . . : Channel 2, Target 0, Lun 0, Device: 0
Hard Disk Model ID . . . . . . . . . . . . . . . : SAMSUNG HD103SJ

Hard Disk Number . . . . . . . . . . . . . . . . : 2
Interface . . . . . . . . . . . . . . . . . . . : S-ATA II
Disk Controller . . . . . . . . . . . . . . . . : Standard AHCI 1.0 Serial ATA Controller (AHCI)
Disk Location . . . . . . . . . . . . . . . . . : Channel 0, Target 0, Lun 0, Device: 0
Hard Disk Model ID . . . . . . . . . . . . . . . : OCZ-VERTEX2
[/quote

Is it still worth evaluating stability after setting ""Plug and Play OS installed" to YES ?

If I leave a non-bootable CD or DVD in the optical player, would this have a similar effect to "Hard disk pre-start delay" ?

Which is likely to give the longest delay :-
CD or DVD ?
Blank or Music or Video ?

Regards
Alan

[hdsentinel]

Thanks, now I see and yes, I agree that the UNIQUEID is different from the volume serial number.
Sorry for the confusion, to be honest, I did not use UNIQUEID previously but it is interesting.

> Is it still worth evaluating stability after setting ""Plug and Play OS installed" to YES ?

Yes, I'd recommend to try that option. For most cases, this is set to YES by default, it is interesting why it is disabled.
However, I agree that if all drives are on the same hard disk controller (it seems) it may have no effect.


> If I leave a non-bootable CD or DVD in the optical player, would this have a similar effect to "Hard disk pre-start delay" ?

Yes, maybe it will pause the boot sequence for some time. Not sure if works, but maybe can help to allow the drives to be properly ready before starting the boot sequence.

> Which is likely to give the longest delay :-
> CD or DVD ?
> Blank or Music or Video ?

A scratched disk
CD or DVD which may be hard(er) to read may cause pause as then the optical drive can't quickly recognise that the disk is not bootable.
Also (non-scratched) RW disk may be useful, usually the optical drives may require some more time to recognise a DVD-RW disk, compared to an original CD/DVD (the content is not really important).

[alan-0000]

Thanks

I will try setting YES to "Plug and Play OS installed".
and if that fails I will try RW CD and RW DVD.

If I can get the same Disk numbers (0, 1, and 2) allocated regardless of whether I boot up from cold or a hot restart then I will be more confident that race hazards are removed.

I will be back

Regards
Alan

[alan-0000]

It takes 16 Seconds from switching on power before the BIOS starts announcing its actions,
and a further 7 seconds before I am offered the choice of a
normal Windows Start-up;
Windows diagnostics such as a memory test;
Macrium Reflect set of recovery choices.

Those times are not affected by "Plug and Play OS installed" set to either [YES] or [NO].
Those times are not affected by placing a commercially recorded Music CD or a blank Writable CD or a blank Re-Writable DVD in the player.

When Windows is running and I choose to Shutdown,
it takes 10 seconds from clicking the Shutdown option to the screen closing down and removing the Shutting down messages,
but another 2 seconds before the fans and the "chassis running lights" also power down.

When Windows is running and I choose to Restart,
it takes 10 seconds from clicking the Restart option to the screen closing down and removing the Shutting down messages,
and a further 13 seconds before the BIOS starts announcing its actions,
and a further 7 seconds before I am offered the usual start up choices.
The "chassis running lights" do NOT go out and the Restart seems to begin perhaps 5 seconds before the Shutting down is concluded.
The HDD have no chance to spin down or any need to spin up, and who knows how the SSD thinks about it all.

Regardless of "Plug and Play OS installed" setting and the use (or not) of CD or DVD,
The 55 GB SSD in SATA channel 0 is always reported as :-
Disk 2 after a cold boot from a total shut down,
Disk 1 (or even 0) after a restart.

Conclusion :- I do not yet have a solution.

A strange effect with a commercial Video DVD.
On a cold boot it still takes 16 Seconds for the BIOS to start announcements,
but there is a further 15 seconds before I am offered the usual start up choices.
For some reason the BIOS / Windows transition is stretched by 8 seconds to a total of 15 seconds whilst inspecting the contents of a large DVD.
THIS ONLY HAPPENS ON A COLD BOOT.
On a Restart this 8 second inspection delay does NOT happen - obviously because Windows still remembers the contents of the DVD,
which proves to me that a Restart does NOT make the computer forget its past in the same way as a shut-down.

QUESTION :-
Is it worth enabling the MARVELL OPTION ROM or disabling one of the MARVEL SATA CONTROLLERS ?

The BIOS shows under ADVANCED : ONBOARD DEVICES CONFIGURATION
MARVELL 6111 SATA CONTROLLER [ENABLED]
MARVELL 6121 SATA CONTROLLER [ENABLED]
MARVELL OPTION ROM [DISABLED]

Regards
Alan

[hdsentinel]

To be honest, I also have no other idea ....

The delays should allow time to spin up, but it seems this is not really problem as on cold start (when the drive just start spinning) there are no problems, "just" on reboot, when the drives are already active, spinning.

> On a Restart this 8 second inspection delay does NOT happen - obviously because Windows still remembers the contents of the DVD,

This is not really Windows issue.
The optical drive itself reads the contents of the disk and if there is no power loss (or open+close of the tray of the optical drive) the drive "remembers" for the disk inserted.
Windows is not active at that point, but the devices may be intelligent to remember their last status.

> Is it worth enabling the MARVELL OPTION ROM or disabling one of the MARVEL SATA CONTROLLERS ?

I'm not sure, but it may worth a try....

Did you check if there is a BIOS update available for your motherboard? Not sure, but if so, it may help.

[alan-0000]

Thanks for trying.

The BIOS was up to date 6 months ago - I will look again.

If I make any further progress I will update this topic - otherwise I will let it rest.

Regards
Alan