Re: Can Debian Backup ntfs File System?

[Axel Freyn <axel-freyn@gmx.de>]

Hi Camaleón,
> > But that is extremly dangerous...: you risk to loose/destroy
> > informations on the damaged filesystem (nobody guarantees, that scandisk
> > is not destroying data...). And especially defrag: this WILL destroy
> > data, which are in lost files (=sectors of the harddisk which seem to be
> > free according to the file system...). 
> 
> (...)
> 
> I've never heard neither seen that before and have worked with ntfs 
> volumes during many years. In fact, I've managed to "restore to live" 
> windows systems that were unable to boot up by following that procedure 
> (even an unexpected shutdown can make the OS to be unbootable -system 
> files tend to become easily corrupted- and checking the file system 
> structure solves the issue).
> 
> Scan disk (chkdsk) and defrag are the standard and recommended tools for 
> dealing with MS file systems problems. Maybe they cannot solve the 
> problem but won't aggravate it either. And before "wiping out" the hard 
> disk, I think it is at least worth a try.
The point I wanted to make is the following: Defrag shifts parts of the
files -- and in order to do so, it OVERWRITES unused parts of the disk.
Now: as soon as the filesystem is corrupted, one can't trust anymore in
the information which parts are "unused" -- so it can happen that defrag
(or any other tool which writs on the disk) overwrites
file-informations, which belong to a "lost" file, but which are believed
to be "free" and "unused" by the filesystem. In my opinion, this danger
is not limited to ntfs, but exists in every filesystem: as soon as the
filesystem is corrupted, one can't guarantee that the blocks marked as
"free" don't contain usefull information -- and that's why I wouldn't
write anything on the partition!

A similiar point is true for scandisk / chkdsk: Again, independent on
the filesystem: Once the filesystem is corrupted, the recovery tool has
to make assumptions about what is the "correct" information -- and if it
makes the wrong assumption, it WILL destroy data. For example: How can
chkdsk guarantee that, whenever it writes some information on the disk,
it uses a truely "free" part of the disc? maybe on this place was a
piece of an important file, which was lost by the file-system
corruption?

Of course the question is, how important the data are -- is it worth the
effort to first create a copy of the partition and then to work on the
copy?

Axel