Discussion: Remapping bad sectors/clusters?

In another thread, regular poster iflyprivate raised an interesting issue:

I think we have talked about bad-sector handling, but I do not remember if we ever did get an authoritative answer?

I think there is two bad-sector scenarios that needs to be considered:

1. Bad sectors and sector remapping that is handled by the disk and the disk controller.
2. Bad sectors that are marked as unusable by the OS and the file system.

Ad 1) Transparent sector remapping is something most modern disk drives can handle. Anyone with a disk that supports S.M.A.R.T. can check this for themselves. One of the S.M.A.R.T. values (I don't remember it name) contains the number of sectors that the disk controller have remapped. It should be in the low 10's to 100 or so.

As I said, this is completely transparent to the OS and the filesystem, so whatever True Image does when restoring the image, the controller will handle any bad sectors on the disk.

Ad 2) How do True Image handle bad sectors managed by the file system? Can it read the bad cluster table in say NTFS and avoid imaging those clusters (sectors)? That solves the imaging problem, but what happens during the restore.

As I see it, True Image must assume that the destination disk is okay and that none of the free sectors are bad. Or do it perform a read-after-write for each sector, and emulate the mark-as-usable process in the same way that the OS/file system would do?

Discuss!

 

Thanks so much for translating my conceptual question into stark reality so it can be addressed once and for all.

Although I have limited technical knowledge I fully understand your category 1 statement about S.M.A.R.T. disk error handling by the drive controller. It makes perfect sense that that type of error handling is not a TrueImage8 concern and is not what I was originally questioning.

However, your category 2 is EXACTLY what I have been concerned about and thus far we have only gotten what we Americans may call the "good-old-runaround" from Acronis support on the specifics of this concern. In other words their replies have barely skirted the question of how TI8 addresses the technical realities you put forth in your Ad 2) hypothesis.

I really want to KNOW that TI8 is properly handling these spurious disk errors on the fly and that I can completely trust the results of every restore action as "disk error free".

 

OK, I would have two points of consideration:

1. How do AD1 and AD2 relate to one another: why does the operating system think that a sector is bad when the Smart function should have remapped it?

2. Can you point to any modern drives that do not have sector remapping?

 

Depends on how/when the controller can do its magic. If the controller only remaps on failed sector-writes, you have to consider what happens when a sector that is in use goes bad.

Take Windows as an example. CHKDSK'ing with the /R option (Locates bad sectors and recovers readable information) will add the cluster with the bad sector to its Bad Cluster table and will avoid using it in the future (or for as long as the partition exists). That is how I understand - I could be wrong though...

If the controller also re-reads and remaps on failed sector-reads in a way that is transparent to the OS, then CHKDSK will never see a bad cluster.

Observe, that remapping by the disk controller on failed-writes is fine when it comes to restoring an image, since any bad sectors on the destination disk will be transparently remapped. The only problem is that the Bad Cluster table (which is also restored) will now contain a false entry...

Define modern.

My old, 300 Mhz Dell laptop with a 6 GB drive (which I type this post on) is a good example. 3 days ago I used my lunch break to boot into Win2K's Safe Mode and start a file-level backup to an external drive. Everything went smooth - except I got an error message about a corrupt file in my Temporary Internet Files folder. As a result I scheduled a full CHKDSK on the next boot. And as expected, somewhere in phase 3 or 4 the same file popped up, and CHKDSK wrote that the Bad Cluster table was being updated.

When it was all done, I rebooted and deleted the offending file, secure in my knowledge that Win2K had marked the cluster (and the sector) as bad.

Clearly, this disk (reported as an IBM-DADA-26480 by the Device Manager), do not support remaping on failed-reads (or perhaps it is fresh out of spare sectors).

 

This sounds like you suppose that the metadata is simply restored intact. If that were the case, since the sectors are not mapped one to one, the bad cluster table would be the least of our worries.

 

Okay, so now we all know that you two gurus can talk technobabble about what-if scenarios as related to disks in general.

Back to the issue at hand: How does TI8 address the situation with the Bad Cluster Table entry and how serious is that when judging the reliability of a restored image?

We need to hear definitively from Acronis Tech Support in clear, easily understood language how TI8 actually addresses how it restores data under the above circumstances. We can't just guess because we end up chasing our "tales".

 

Sorry, but I am no guru. I am just a dabbling amateur who know enough about computers to cause a lot of damage.

But I digress...

@Acronis - we want to hear from you

 

@MiniMax

Sorry, I misunderstood. I thought that this was your thread and that you said "Discussion" and "Discuss".

 

I think any discussion is welcome here. Of course, only the Acronis folks can weigh in with an absolute answer to any questions raised.

I find their silence disturbing. These are simple questions that deserve clearly worded answers.......unless they have something to hide?

 

Hi, MiniMax

Lo Lo Lo, OTFLMAO

I think that goes or most us.

Take Care,
TheQuest

 

Hi - no misunderstanding, beenthereb4. Keep talking.

Could you expand on this? I have no experience with FAT and NTFS file systems. Are you talking about TI recreating file system structures and pointers in FAT and NTFS metadata to match the addresses of restored clusters?? Or??

 

Yes, if you pin an Acronis programmer down, they will use the term "sector based imaging". The file system is not copied, but recreated based on the content of the restored drive\partition. Bad cluster pointers, along with all the other pointers are discarded. As an long ago instructor, I was trying to lead you to figure it out for yourself by questions and suggestions (That was the best technique in the good old days). Anyway, this will be my last post on the subject.

 

I am sorry to hear that. Thanks for the information you have provided so far.

So ... True Image

1. does not simply copy sectors one by one to the image file.
2. needs to understand the file systems it is copying.
3. copies enough information from the file system to the image file to be able to construct an equivalent file system on the restored partition.
4. (except if the source partition is corrupted, in which case a sector-by-sector copy is used?)
5. when writing to the destination disk, relies on the disk/controller to do any necessary remapping of a bad sectors.

In other words, there will never be a problem with undetected bad sectors?