Cannot create a good disk image from a.gho file

[dencorso]

I don't know what went wrong. On my laptop I have a primary boot partition C:, and an extended partition which includes a logical partition for Win2k and another logical partition for WinXP, they are all visible and coexist nicely. ....

I must say that I find your system of allowing all the operating systems to see one another and having all these partitions visible at once very bizarre.

I do it too. I first setup my 4 HDD machine with the slave IDE as the boot device and true DOS 7.10 boots from its 1st (primary) partition, and lauches Win 98SE or GRUB4DOS. If GRUB4DOS is launched then it'll let me boot XP SP3, from the 1st (primary) partition of the master IDE or TinHat Linux or RIP Linux from images or go to the GRUB4DOS prompt. All four HDDs have extended partitions further subdivided in several logical partitions, althou the SATA HDDs lack any primary partitions on purpose. I let the DOS/Win 98SE assign the letters to all disks, and jotted them down together with their labels. Then, in XP SP3 I've used the "Disk Management Console" to assign all letters just as it is on 98SE, except for the boot disk, so that my boot partition is always C:, the inactive other OS partition is always D:, and all other partitions always get the selfsame letters. I also turned off System Restore in all partitions in XP SP3, but I've kept the Recycle Bin active on both OSes. This presents no problem when I boot XP after 98SE, but I have to delete (with deltree) all the Recycled folders during AUTOEXEC.BAT, before booting to 98SE after XP. I take the oportunity to delltree and recreate the Temp folders also, at the same place. And all partitions are FAT-32 and all OSses see every one of them. And no file from one OS is in the partiton belonging to the other, so even if I remove one of the bootable HDDs, the other continues able to boot as it is. In my eyes, its also quite straightforward...

But, in any case, as always, YMMV, of course!

[LoneCrusader]

I do it too. I first setup my 4 HDD machine with the slave IDE as the boot device and true DOS 7.10 boots from its 1st (primary) partition, and lauches Win 98SE or GRUB4DOS. If GRUB4DOS is launched then it'll let me boot XP SP3, from the 1st (primary) partition of the master IDE ..... all partitions are FAT-32 and all OSses see every one of them. And no file from one OS is in the partiton belonging to the other, so even if I remove one of the bootable HDDs, the other continues able to boot as it is. In my eyes, its also quite straightforward...

But, in any case, as always, YMMV, of course!

Although arranged differently, your setup still retains (IMO) the most important things. No files from one OS in the partition of another, and each OS is independent and does not require another OS or the partition/boot sector thereof to run.

I prefer putting the systems in partitions in chronological order (except I would install ME before 2K), and I don't like giving in to 2K/XP demanding to take over the first primary partition, whether installed there or putting their boot files there. I guess I have a touch of OCD haha

[jaclaz]

I must say that I find your system of allowing all the operating systems to see one another and having all these partitions visible at once very bizarre.

Well, I have been called names before, but it's the first time I've been tagged (indirectly) as "bizarre".

I find OBVIOUS that each system has to be visible AND run on a different drive, i.e. Windos 9x is ALWAYS on D:, NT is ALWAYS on E:, 2K is ALWAYS on F:, XP is ALWAYS on G:, ME is ALWAYS in the trashbin .

jaclaz

[dencorso]

BTW, I was running sdelete under Win98SE in a DOS window yesterday, it had looked like a DOS program. sdelete seems to work fine on FAT16/FAT32 partitions under Win98SE, even if the text "Cleaning MFT.../" came up for a long while.

The docu page http://technet.microsoft.com/en-us/sysinte...s/bb897443.aspx lists as system requirements WinXP and higher. Would this mean that sdelete doesn't work properly under Win2k or with NTFS 3.00 partitions by Win2000?

It depends of where you look, it seems. In all cases we're talking about v.1.51 and look what was claimed about in the SDelete v1.51 Info Page (as it was in Jan 02 2008): "SDelete works on Windows 95, 98, NT 4.0 and Win2K". So I think you've just stumbled in a case of previously-documented undocumented behaviour!

Changing subjects somewhat, seriously, there are various builds of the sdelete v. 1.51, which are best diferenciated by their PE Timestamp (which one can easily read with the MiTeC EXE Explorer):

Notable builds of sdelete.exe version 1.51

==========================================================================

PE Timestamp: 07/09/2005 08:00:48 PM Size: 49,152 bytes Pre-MS (-c does nothing, non digitally signed).

PE Timestamp: 11/17/2005 03:31:42 PM Size: 49,152 bytes Pre-MS (-c works OK, non digitally signed).

PE Timestamp: 08/16/2006 05:33:17 PM Size: 166,712 bytes Available now from MS (-c works OK, digitally signed).

So, except for the earliest build I've just mentioned (which is buggy and used to be distributed alongside the source, way back when), all other pre-MS takeover (Jul 18 2006) versions are, in truth, documented to work with Win 9x/ME... and since the code probably didn't change significantly, because the version is still exactly the same, all later builds still ought to.

[Multibooter]

I find OBVIOUS that each system has to be visible AND run on a different drive

That's my personal preference too, but there are several different issues which got intermingled in the last few postings, but I can only speak from my experience with System Commander:

1) Multiple operating systems sharing a single partition:

I had no problems with various DOSes co-existing on the same partition, provided they used the same code page. If I remember right, I even had various different Windows 3x installed on the same partition.

Starting with Win95 it became problematic to have multiple instances of Windows on the same partition because they used the same folder "Program Files" and "My Documents', and one Windows would then use the others' "Program Files" or "My Documents".

I actually had US Win95 and a localized Win95 running on the same partition. This was possible because localized versions of Windows would use different folder names, such as "Programme" under German Windows, instead of "Program Files" under US Windows. The 2 instances of Windows should therefore not have interfered with each other. But then the code page problems started. Folder and file names created under a localized Windows might contain special characters from their own code page. ScanDisk under US Win95 would detect the "bad" files names and clean them up/delete them. Various localized Win95 which I have seen, had come out of the box with some files having filenames containing characters specific to a non-US code page.

2) Visible partitions

In 1997 I was using System Commander, but was checking out several other boot managers. I installed one which was based on hiding partitions, I think it was the PTS/Paragon Boot Manager v2.5 , and I didn't like it, I wanted to have everything on my computer accessible to me, not just one partition.

Hiding partitions and selecting among various sets of boot codes are two different approaches. System Commander originally used the selection approach. If I remember right, the early versions of System Commander didn't have the ability to hide partitions. I assume V-Communications added this feature because competing products had it, and many people buy features.

3) Hidden partitions

The problems caused by using different code pages, for example a US and a Farsi windows, on the same computer can be nasty. One approach to solve this problem is to run the non-US operating system on an otherwise invisible partition.

My approach was to leave all partitions visible and to install the Windows with the non-US code page onto a removable Jaz disk. I had the Jaz disk only inserted when I would actually use that non-US operating system.

4) Always visible boot partition

With System Commander whatever non-boot code is on the boot partition C: is (probably) always shared between all operating systems, even if they are on partitions hidden from each other. This shared boot drive might lead to some minor interference, but never posed a serious problem to me.

Example: One Windows (e.g. on E:) crashed or hung. Windows or an application in it then wrote some stuff on the boot drive C: or C:\Temp\ If I would then immediately boot into another similar Windows (e.g. on F:), that Windows or the same application in it would process the crash logs etc on boot drive C:

But this would happen also if Windows 1 and Windows 2 were installed on partitions hidden from each another.

5) Intentional sharing between different operating system on different partitions

I had 2 instances of Win98 share the same fixed-size swap file on a special partition. I have several applications which were installed once under Win98, and which run fine under WinXP without re-installation, just by creating a desktop shortcut and by adjusting the path in the registry (e.g. a special version of Ghost), or by re-entering user settings.

Edited January 14, 2010 by Multibooter

[LoneCrusader]

Well, I have been called names before, but it's the first time I've been tagged (indirectly) as "bizarre".

...

ME is ALWAYS in the trashbin .

BTW I agree on the ME thing, was just using it as an example.

4) Always visible boot partition

With System Commander whatever non-boot code is on the boot partition C: is (probably) always shared between all operating systems, even if they are on partitions hidden from each other. This shared boot drive might lead to some minor interference, but never posed a serious problem to me.

Example: One Windows (e.g. on E:) crashed or hung. Windows or an application in it then wrote some stuff on the boot drive C: or C:\Temp\ If I would then immediately boot into another similar Windows (e.g. on F:), that Windows or the same application in it would process the crash logs etc on boot drive C:

But this would happen also if Windows 1 and Windows 2 were installed on partitions hidden from each another.

If one created a special always visible boot partition for System Commander, (as you said^) this is completely true.

For the record though, you could have System Commander hide it's own partition whenever you choose one of your OS'es from the menu, thereby allowing each Windows to see its system partition as C: on boot. This would eliminate having to install System Commander to one of the Windows partitions (I know you mentioned you preferred installing from/to DOS) and also possibly eliminate some of the quirky errors/problems that have been mentioned about System Commander (especially when making backup images and/or cloning drives/partitions, because each Windows install would be "standard" so to speak). The only thing you would be giving up is one primary partition, which isn't a big issue if you're using several logicals.

[Multibooter]

you could have System Commander hide it's own partition whenever you choose one of your OS'es from the menu, thereby allowing each Windows to see its system partition as C: on boot.

Hi LoneCrusader,

System Commander installs stuff on track 0 and on its install-to partition. I sometimes edit/replace those boot files which I want System Commander to load from the next boot onwards (e.g. autoexec.bat, config.sys or boot.ini. even io.sys) directly in the folders where they are saved/loaded by System Commander, e.g. in C:\SC\WINDOWSX\ for Windows XP. I also make a manual backup of C:\SC\ in case I want to reverse the changes in boot files made by the installation of new software. So I would not want to hide the partition where System Commander is installed.

[have System Commander hide it's own partition] would... possibly eliminate some of the quirky errors/problems that have been mentioned about System Commander (especially when making backup images and/or cloning drives/partitions, because each Windows install would be "standard" so to speak).

I doubt it. The error condition on which System Commander choked was caused by Ghost, plus the parameters I used with Ghost. Ghost created something bad on the cloned HDD. In my opening post #1 I stated:

Regardless of what I select, Save or Bypass, System Commander eventually stops loading any operating system at all and displays a cryptic err msg "Error Boot 2>" or similar.

The System Commander user manual of v9 (.pdf) on p.263 explains: "The codes given help identify the source of the problem. Generally, you're given the option to boot into one of the four primary partitions on the first drive. Several combinations we've seen are: ... The second "X" [here: the ">" in the error code "2>"] indicates the error code returned from the hard disk BIOS. It can indicate the hard disk or controller has some type of problem, or might indicate bad partition information on the disk... Boot 2> or Boot3>. This error indicates that the file SYSCMNDR.SYS could not be found in any primary partitions on the first drive. To fix this, boot from a DOS or Windows 95/98/Me startup diskette and at the prompt, type FDISK /MBR. This will have no effect on partitions, but installs the generic MBR boot loader. After your operating system is running, you will need to perform a full installation of System Commander."

This is the official work-around for the current problem, a poorly cloned HDD. You used the same work-around and re-installed System Commander on top of the existing installation after you created a bad disk image with SystemRescueCD / Partimage. BTW, System Commander is a dead product, but well documented.

The re-installation workaround suggested in the manual leaves the original question:

Why can Ghost not create a useful image? Did I miss any parameters? ... I did not use the sector-by-sector forensic -id option because creating and restoring an image in this "forensic" way may take 24 hours... Is there other software which actually can clone from a file a HDD containing tricky boot code by System Commander, and which doesn't alter important partition characteristics?

What else might not work on a poorly cloned HDD? Is the System Commander cloning problem just the tip of the iceberg? Edited January 14, 2010 by Multibooter

[dencorso]

The re-installation workaround suggested in the manual leaves the original question:

Why can Ghost not create a useful image? Did I miss any parameters? ... I did not use the sector-by-sector forensic -id option because creating and restoring an image in this "forensic" way may take 24 hours... Is there other software which actually can clone from a file a HDD containing tricky boot code by System Commander, and which doesn't alter important partition characteristics?

What else might not work on a poorly cloned HDD? Is the System Commander cloning problem just the tip of the iceberg?

Of course it's just the tip of the iceberg, but you might have to wait a long time before uncovering more of it. My rule of thumb is this: ONE single flaw detected means: image rejected, no exceptions. It helps avoid a bucketfull of grief later. Ghost is powerful, and can do lots of interesting things, but can only be mastered through long and careful experimentation, on a case by case basis. That's why a full True Image (blind sector-by-sector) is the one you do first, in the general case, when you simply cannot afford to make any assuptions. BTW, a true forensic image is a True Image (that ghost calls "raw", "-ir"), and the fact they call "forensic" the "-id" is most unfortunate, because "-id" normalizes the disk structure, on the basis of suppositions that may (and as we saw, more often than not, actually do) render the clone it generates unbootable... in fact, it's faster and preserves (hopefully) all the data (most of the time), but introduces bias, that, IMO, taints the proof-value of such images, so they are not truly forensic, at all.

[Multibooter]

I have successfully created and restored with Ghost v11.0.2 an image of the System Commander HDD using the "-ir" switch for sector-by-sector cloning.

System Commander booted fine from the HDD created with the "-ir" switch from the .gho image file. Neither PartitionMagic 8 nor Partition Table Doctor signaled any errors with the cloned System Commander HDD. The compressed .gho image file, created from a HDD previously cleaned with "sdelete -c", was about 1GB, using compression switch "-z9".

I have created with Partition Table Doctor 3.5 a "partition table backup file" of both the original System Commader HDD and of the HDD restored from the .gho image file. Both "partition table backup files" (about 10k in size) were identical when making a binary compare with Beyond Compare. I have also created with MBRWizard image files of track 0 of the original and cloned HDDs. Both image files of track 0 were identical.

Conclusion: Ghost v11.0.2 CAN create a good image file of a System Commander HDD if you use the "-ir" switch. No other disk imaging software I have tried could produce from an image file a good System Commander HDD.

The "dumb" sector-by-sector cloning approach worked, while software using a "smart" approach and fast shortcuts could not create a good copy of a System Commander HDD from an image file.

[jaclaz]

Conclusion: Ghost v11.0.2 CAN create a good image file of a System Commander HDD if you use the "-ir" switch. No other disk imaging software I have tried could produce from an image file a good System Commander HDD.

Which actually translates to:

ANY cloning application capable of a "forensic mode" will work, if used in "forensic mode".

The "dumb" sector-by-sector cloning approach worked, while software using a "smart" approach and fast shortcuts could not create a good copy of a System Commander HDD from an image file.

Because it is not "smart" enough.

In these cases you could take a mixed approach and supply yourself the missing smartness of the program, like:

• backup the needed sectors that NEED to remain as they are (forensic quality clone of a limited number of sectors)
  
• image the drive (in a " smart" way)
  
• restore the image (in a " smart" way)
  
• restore the backed up bunch of sectors
  

jaclaz

[dencorso]

@Multibooter: I think now is the proper moment to suggest that you reread my post #29, above, in this thread, and also do reread from post # 166, to the end, of Dave-H's recent thread, where we last discussed imaging at lenght, to form a good picture of it all (BTW, I've sent you a PM, too). With the added experience you've just amassed, I'm sure it'll be worth it. Of course, I'm also writing this here to leave a pointer to help any future reader to be able to easily find that other thread, where those posts are somewhat hidden at the end of a long discussion on not obviously related matters.

[Multibooter]

I have successfully created and restored with Ghost v11.0.2 an image of the System Commander HDD using the "-ir" switch for sector-by-sector cloning.

I am updating this older posting.

During the most recent deployment I had a 750GB source PATA HDD with System Commander plus several operating systems on it. As target HDD I had only a 200GB PATA HDD, the 750GB PATA HDDs are getting hard to find and are quite expensive. Because the target HDD was smaller than the source HDD, the -ir (image raw) switch wouldn't work. The 750GB source HDD had less than 200GB allocated to partitions, the remainder was unallocated space.

I have now successfully created and restored with Ghost v11.0.2 an image of the 750GB HDD with System Commander on it, using the "-ia -ib" switches for sector-by-sector cloning (instead of -ir), but additional work was required:

1) immediately after creating the HDD image with the -ia -ib switches I booted into Win98 to fix up the target HDD

2) under Win98 Norton Disk Doctor and Partition Table Doctor did not report any errors on the cloned HDD. When I ran PowerQuest PartitionMagic v8.01 build 1312 under Win98, however, the following err msg was displayed:

"PowerQuest PartitionMagic has detected an error 114 on the partition starting at sector 8385992 on disk 2. The EPBR is not positioned at the beginning of a cyclinder. If this is not corrected the operating system could cause data loss. PowerQuest PartitionMagic can easily be fix this problem by moving the EPBR to sector 8385930. Would you like PowerQuest PartitionMagic to fix this error?"-> Yes, then msg: "Success. The partition table error was successfully fixed!" This was repeated for most partitions on the cloned HDD.

PartitionMagic under Win98 did fix the cloned HDD fine, afterwards System Commander and all operating systems on the (smaller) cloned HDD worked fine.

Addendum: Error 114 is not mentioned in the PartitionMagic User Guide. In http://www.win.tue.nl/~aeb/partitions/partition_types-2.html#ss2.13 it is described as follows: "114 - Logical partition does not start one head away from EPBR [Extended Partition Boot Record]. If the EPBR is found at sector N, and there are 63 sectors per track, then Partition Magic expects the logical partition to start at sector N+63."

Edited December 4, 2010 by Multibooter