On 9/14/05, andy <dancer@...> wrote:
> > Easily done. Mount your NT partition through samba, and run dd. An
> > example would be:
> > dd if=3D/dev/cobd0 of=3D/mnt/disk.image
> well at image creation time I will be running linux native
> not CL so I can mount the partition directly but I think
> it doesn't want mounting (I just need access to the device
> file in /dev I think).
> Presumably what I do is
> dd if=3D/dev/hda3 of=3D/somepathToFileOnSomeDisk
> is that correct ? What about block size ?
Ack.. yes.. of course.. I was still in "colinux mode".. you should be dd'in=
/dev/hda3 instead of /dev/cobd0. But yes, that syntax is correct. Usually i=
will work with no block size set (there is some default that generally=20
> a) For the linux-in-a-partition that also runs with CL on
> > w2k is there a size limit - does that partition have
> > to be <=3D 4G to run with CL w2k ? (that would be a shame).
> > No. This 4gig file limit only applies to FAT32, AFAIK. Even NTFS can
> > work with images larger than 4gig (I've run 6 gig images before). Also,
> > if you're running FAT32 on a colinux system, you can get around this 2
> > gig limit by using RAID devices if disk size is upmost importance (and
> That was my question. My daughter's machine has FAT32 so I presume I
> can only use up to 2G if I want to use it with CL on there.
> I could *build* it on an external disk but that subverts one aim which
> is to leave the running system on her machine as well.
I would seriously suggest using RAID0 in this case. You'd have to build a=
customer kernel, but I'm going to work on a way to properly=20
document this so that its pretty straight forward. It effectively takes a=
"list" of 2 gig files and just concatenates them into a large
disk. That way, you can make 10 or even 20 gig drives for Linux to use (as=
long as you don't mind a lot of disk files laying about).
The point is that since it will run on my box which is ntfs its a shame
> if it has to be limited to 2G. I'd move her to xp cept it aint so
> freely avail without license as w2k.
Aye. With the above comment, though, you should be able to break the 2 gig=
limit (at least on the linux side, not on the FAT32 side).
> I had thought of running her box on ntfs - anyone know, what does
> ntfs do functionally (above block level/fs access level) that's
> different to FAT32 ? I mean could I rsync/tar/whatever her system,
> format a partition as ntfs and put it on there ? Since I read
> in many places that linux doesn't support ntfs functionality I
> always presumed there was functionality related to the OS
> implemented in it differently to on fat32 so that you couldn't
> just lift a w2k system from one and plonk it on another and have it
> run. Is that correct ? What is it that isn't implemented
> in normal unix filesystem access to ntfs that would make that
> not work ?
NTFS offers a lot of advancements that some say slow it down and others say=
don't affect performance at all. Some of the list are:
1. Larger files (as demonstrated above)
2. File ownership (much like UNIX)
3. File security (like modes in UNIX)
4. Sparse files (effectively compressing blank space, much like MP3=20
5. Journalling to assist in file recovery if the disk is corrupted (a lot o=
people say this is what slows down the FS, but its debateable)
There are other advancements, but these are the big ones I can think of now=
I'm a little confused about your last question regarding UNIX and NTFS, but=
I think you mean "Why can't you mount an NTFS partition
in UNIX?". As far as I know, the format of NTFS is proprietary, and=20
Microsoft isn't giving out the white papers on the filesystem layout,
so a lot of people are figuring out the format by hand. Apparently, this=20
isn't as easy as it sounds, since it hasn't been successfully done
yet (well, thats a lie.. it has been done, and read support is available,=
but write support is shaky.. it works for those files that are overwritten
and not changed in size, but not for new files, or files that you want to=
overwrite and change the size of).