Partition restore auto translate disk geometry - system partition unbootable

Hi,

This is a follow up to this thread with the new cause identified.

I believe True Image will automatically translate the disk geometry embedded in the image to the disk geometry of the current destination drive during partition restoration.

There is no warning of the translation, and I do not find it documented in the user guide.

This will create problem if the partition is a bootable (XP I have tested) partition. It will not be bootable after the restore.

There will, however, be no automatically translation when you do a disk restore. You have to do a entire disk backup with all partitions, and then check the DISK-X box and thus selecting all partitions during restore.

This workaround is undesirable as it waste time and space for backing up the other partitions.


Hardware tested: IBM Thinkpad T22
Hard disk type: IBM 20GB 2.5" DJSA-220
OS: Windows XP SP2
True Image version: 8.0.937 and 10.0.4942

IBM (and possibly some other notebook vendors) BIOS will detect HDD with a geometry of 240 heads (or LARGE mode) instead of the more common 255 heads (or LBA mode).

This is true for their IDE Primary Master (normally internal HDD) and IDE Secondary Master (normally ultrabay removable HDD tray) WHEN it does a reboot.

After you boot up Windows, and insert a HDD via USB or Ultrabay, then it will be detected as 255 heads.

Normally Windows will be installed on internal HDD with BIOS detected as 240 heads, and also primary NTFS paritition boot sector also configured as 240 heads. This works fine.

When you do a image of the system partition, I am not sure if the NTFS geometry of 240 heads is saved.

However, when you restore this image to a hard disk currently detected as 255 heads (like insert the new hard disk via USB or ultrabay hot insert), then True Image will automatically translate from 240 heads to 255 heads in at least 2 areas:

(1) Partition table within MBR

Original partition table:

Disk: 0 Size: 19G CHS: 2584 240 63
Pos MBRndx Type/Name Size Active Hide Start Sector Sectors DL Vol Label
--- ------ ---------- ---- ------ ---- ------------ ------------ -- ----------
0 0 07-NTFS 10G Yes No 63 20,608,497 C: system
1 1 05-EXTEND 9.0G No No 20,608,560 18,461,520 -- <None>


Partition table after restore:

Disk: 1 Size: 19G CHS: 2432 255 63
Pos MBRndx Type/Name Size Active Hide Start Sector Sectors DL Vol Label
--- ------ ---------- ---- ------ ---- ------------ ------------ -- ----------
0 0 07-NTFS 10G Yes No 63 20,611,332 -- system
1 1 0F-EXTEND 9.0G No No 20,611,395 18,442,620 -- <None>


Please note old partition layout ended in 240x63 sectors cylinder boundary, while the new partition layout ended in 255x63 sectors cylinder boundary.

(2) NTFS partition boot record disk geometry

Please see attached Old and restored NTFS boot record:

Please note that no. of heads has changed from 240 to 255.

With these 2 changes, it seemed the new HDD will not boot into XP when you put the HDD into the notebook internal disk position, as BIOS will detect it as 240 heads.


My recommendations:

I am not sure whether Acronis will add any warning or document this important "feature", or add an option to enable/disable this automatic translation.

For IBM thinkpad users, or any other users with notebook that report 240 heads I would therefore recommend the following 2 ways to restore which works for me:

(1) Put new HDD in Ultrabay and reboot before restore under Windows. This is when your notebook is still working and you are creating a emergency bootup HDD.

(2) Swap new HDD with internal harddisk, and restore with Acronis recovery CD/DVD. This is when your internal drive is dead and you are replacing a new one.

Both ways will make sure you do a reboot and will make sure that the new HDD is detected as 240 heads before the restore.

Any comments are welcome.
Vince

 

Vince:

Well, you made me go take a look at my ThinkPad X41T. I was blissfully unaware of how the BIOS reports the disk geometry, but I found exactly what you found on your T22; the disk geometry is reported as 240 heads and 63 sectors.

I have not tried what you are trying with restoring an image to an external drive that is reporting a different geometry, but I have restored single partitions from images created with TI9 and TI10 to the machine many, many times without incident, including bootable NTFS, ext3, and FAT32 partition images.

Your recommendations for working with ThinkPads are consistent with the advice given on forum.thinkpads.com in that most people there recommend putting a new disk in place of the internal hard disk and restoring an image from an external drive (or one in the Ultrabay) to the internal disk.

Hopefully Acronis will reply with comments on your findings.

 

Mark and Vince,

Dan Goodell explains this laptop geometry problem in this thread..

http://radified.com/cgi-bin/YaBB/YaBB.cgi?board=general;action=display;num=1128609708;start=

 

Thanks, Brian. Very interesting thread!

Vince, do you see post #26 in the referenced thread? The byte being referred to (absolute byte 7E1Ah) is at an offset of 001Ah from the beginning of sector 63, which is the parameter {# of heads} in the partition boot record. So according to the poster "tnoone" you could directly edit this byte, changing it from 255 to 240 to make your unbootable image bootable again. This is easily done in DD10 by changing the parameter in the "View as NTFS Boot Sector" in the Disk Editor window.

It would appear from Dan Goodell's description that this is not a TrueImage (or Norton Ghost) issue, but fundamental to the way that the ThinkPad BIOS reports the drive geometry.

 

Mark,

Yes I came across some forums with similar advise, though the exact reason was not given.

I found that out the hard way - doing various tests on my own.

My rule is: Thinkpad will ONLY report a disk as 240 Heads when you reboot with the hdd in internal bay or ultrabay.

Make sure when you restore, the destination drive is detected as 240 heads.

If you hot swap a drive in USB or ultrabay, it will be reported as 255 heads. However, if a drive has already been formatted with 240 heads NTFS partitions, it will still work. It seemed only when you create a NTFS partition, or boot a system partition, that disk geometry matters. I have been using HDD formatted as 240 heads partitions on a USB bay detected as 255 heads for a long time and did not notice any issues.

Brian,

Thanks for the good link. Yes Dan did explain the symptom in this thread. But I would like to supplement with 2 rules on how to make sure it works.

Rule #1: if you use the Thinkpad to do image and restore, just make sure you use the above REBOOT rule.

Rule #2: if you use other computers to image and restore which don't report 240 heads. Then you can do a "disk image" and "disk restore" as stated in my original posts. True image will not do the translation in this case. However, it waste more time and space in imaging/restoring other partitions on the same disk.


Mark,

I have not tested this. It may boot as the author said, but I am not sure it will work correctly afterwards.

It is much more than the OS boot record 240 heads that is changed. If you look at my images of old and new NTFS boot record, many other fields are changed.

Also the MBR partition table is changed to end at 255x63 cylinder boundary, and I doubt some disk/partition management will work correct when the drive is booted to be 240 heads.

There is a good freeware utility Testdisk that will warn you of such discrepancy of geometry mismatch.

I would recommend using my two rules above for these reasons.

I don't agree to that. I think True Image is at fault.

Firstly as I tried, only partition image/restore will do this translation, but not disk image/restore.

This would lead me to believe that we could have implemented an option that allow the user to enable or disable the translation, if True Image found that the destination drive geometry is different than that of the image.

Secondly, this difference should be well documented, as it leads to unbootable partitions, which waste a lot of time for users to find out by themselves.

Thirdly, they may be other computers reporting some other drive geometry, not limited to 240 and 255. True Image should be able to handle those, or at least warn users of such differences.

Thanks
Vince

 

I think you're right on that one. After rethinking my previous reply and re-reading some of Dan Goodell's replies in the other forum, the reason that the author of post #26 had success is that after Windows boots it almost exclusively uses LBA for disk access and the geometry parameters don't matter except perhaps when using Disk Management to create/delete/resize partitions. The change in the partition boot record of the {# of heads} parameter will probably just make the NTFS partition bootable again and Windows will start OK, switch to LBA mode, and you'll think that all is OK.

However, you're right that the partition table is not repaired properly and the partition boundaries may not be correct, so at some point in the future if you fill up a partition or do repartitioning operations then this is going to come back and bite you.

I'm not sure about this one. I think what happens with TrueImage when you do a full disk restore is that it knows about all of the partitions and it knows that you want to restore everything on the disk to the state that was saved on the image, so it is safe to restore the partition table saved in the image. Restoring the partition table would carry forward the disk geometry settings that were present when the image was created.

In contrast, when TI does a single partition restore it will not replace the partition table with the one in the same image but rather just update the table entry for the partition being restored.

Maybe TI can do a better job of detecting changes in drive geometry and figuring out how to correct this; maybe not. I'm not sure how much is involved in doing this or if it would be foolproof but at least, as you say, a warning message could alert the user that something is wrong. It would be nice if Acronis support could comment on this.

 

Mark,

If you can do disk restore without translation, then I would believe you should be able to do partition restore without translation too.

Doing a translation is a big deal, as it may end up with a unbootable partition. As least you should let the user knows about the translation, better still have to option to choose.

Right now users are hitting a big wall without knowing the reasons.

I think it is an oversight in the True Image design.

Vince

 

Mark and Vince,

I've been thinking about this problem again. When you connect an empty HD in a USB external enclosure to your Thinkpad, how is the geometry described? As 240 or 255 heads?

 

Brian:

255 heads.

 

Thanks Mark. What implications does this have for cloning the internal HD of an IBM laptop to a USB external HD? I suspect the clone will fail. But is that true?

 

That's what I've heard although I never tried it personally. I've always restored images from a network share to the internal drive on the IBM laptop and that works perfectly.

The ThinkPads forum (forum.thinkpads.com) contains several "sticky" posts relating to the use of Ghost or TrueImage with IBM laptops. The recommended procedure is to remove the internal drive and mount it in a USB enclosure. Install the new drive in the laptop and clone from the external to the internal.

 

Mark, your suggestion led me to this post.

http://forum.thinkpads.com/viewtopic.php?t=38496

It seems the IBM laptop is happy if the clone is made when the destination HD is seen in its correct geometry.

 

Brian:

Yes; "correct geometry" meaning that the destination HD of the clone (the clonee?) must be seen with the same geometry as the ThinkPad BIOS will see it when it is later installed as the internal drive; in particular 240 heads for ThinkPads. I take it from the article that this will occur if the drive's partition table is written while the drive is installed inside the ThinkPad as an internal drive.

Talk about a circular thread though! The ThinkPad forum thread that you refer to above contains a reference in post #2 of the thread to the same Dan Goodell post that you referenced all the way back in post #3 of this current thread. What goes around comes around!

 

Cloner and clonee. I like it. Reminds me of an joke I heard a long time ago.