What would I do without partimage?
William Ballard
mostly booted in Debian and manage that godawfulness with partimage.
Every time I boot into it I restore a clean partimage of XP, let it puke
all over itself, then restore the cleanness.
It's the only thing that makes patching Windows remotely tolerable.
Eventually partimage will stop working on new versions of NTFS, and it
seems to not be maintained anymore. It was removed from Sarge.
Are there other tools that work like Ghost but in Linux? Partimage is
great.
--
To UNSUBSCRIBE, email to debian-us...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listm...@lists.debian.org
Arafangion
> I literally would be unable to use Microsoft Windows if I couldn't stay
> mostly booted in Debian and manage that godawfulness with partimage.
>
> Every time I boot into it I restore a clean partimage of XP, let it puke
> all over itself, then restore the cleanness.
>
> It's the only thing that makes patching Windows remotely tolerable.
>
> Eventually partimage will stop working on new versions of NTFS, and it
> seems to not be maintained anymore. It was removed from Sarge.
>
> Are there other tools that work like Ghost but in Linux? Partimage is
> great.
You could use dd, and compress the image, or use an emulator with a COW disk
image.
Marty
> I literally would be unable to use Microsoft Windows if I couldn't stay
> mostly booted in Debian and manage that godawfulness with partimage.
>
> Every time I boot into it I restore a clean partimage of XP, let it puke
> all over itself, then restore the cleanness.
>
> It's the only thing that makes patching Windows remotely tolerable.
>
> Eventually partimage will stop working on new versions of NTFS, and it
> seems to not be maintained anymore. It was removed from Sarge.
>
> Are there other tools that work like Ghost but in Linux? Partimage is
> great.
>
>
Maybe this is just a case where in Unix you don't need weird tools to do a
routine task, or what *should* be routine in any sanely designed OS.*
I've always just used the cp command, e.g. "cp /dev/hda>disk_image" or
"cp /dev/hda1>partition_image to back up a disk or partition, respectively.
(dd also works but I just don't trust it.)
I've never tried to gzip or bzip the resulting image file, but if that works
then I don't see much advantage using partimage or Ghost. They may be of marginal
value if they are smart enough to automatically detect and adjust to differing
drive geometries. I don't think cp or dd can handle that by itself.
*What prompts this remark is an industrial application I heard about, where
a large number of identically configured systems used "ghost" (or similar tool)
to reinstall a pristine copy of XP each time the systems were booted, i.e. at
least once per day! (True story.)
Roberto C. Sanchez
> On Wed, Dec 14, 2005 at 10:22:55PM -0500, Marty wrote:
>
>>*What prompts this remark is an industrial application I heard about, where
>>a large number of identically configured systems used "ghost" (or similar
>>tool)
>>to reinstall a pristine copy of XP each time the systems were booted, i.e.
>>at
>>least once per day! (True story.)
>
>
> awful to use it otherwise.
>
>
Tell me about it. Where I go to school, the lab machines are ghosted
weekly. I lucked out and managed to get a SUSE 9.3 Pro workstation
setup, which I must admit is quite nice. SUSE is very well integrated,
but I still prefer Debian by far. I jsut count myself fortunate that
they didn't force me to use a WinXP machine. Ugh.
-Roberto
--
Roberto C. Sanchez
http://familiasanchez.net/~roberto
William Ballard
> a large number of identically configured systems used "ghost" (or similar
> tool)
> to reinstall a pristine copy of XP each time the systems were booted, i.e.
> at
> least once per day! (True story.)
I load a fresh image of XP every single time I use it. It's unthinkably
awful to use it otherwise.
Michael Dominok
> Are there other tools that work like Ghost but in Linux? Partimage is
> great.
Maybe ntfsclone (part of the ntfsprogs Package) could be of use.
Paul Seelig
>
> Are there other tools that work like Ghost but in Linux? Partimage is
> great.
>
Partimage is nice but the command line based ntfsclone is far more
flexible. Just check out the man page for some usage examples. Here are
some i came up with:
Backing up a NTFS partition into a gzipped image file:
ntfsclone -s -o - /dev/hda1 | gzip -9 -c > winxp_hda1.img.gz
Recovering a partition works like this:
gunzip -c winxp_hda1.img.gz | ntfsclone --restore-image --overwrite /dev/hda1 -
A friend of mine uses this tool on a daily basis to clone a partition to
another disk using this command:
ntfsclone -s -o - /dev/hda1 | ntfsclone --restore-image --overwrite /dev/hdb1 -
Recovering a partition to a file works like this:
gunzip -c winxp_hda1.img.gz | ntfsclone --restore-image --overwrite winxp_hda1.img -
after that one can mount and browse it like a normal filesystem:
mount -t ntfs -o loop winxp_hda1.img /mnt/
Beat this, partimage! ;-)
The ntfsprogs package contains lots of other useful utilities a XP really
can't do without.
Cheers, P. *8^)
Martijn Marsman
There is also Ghost4linux (not the real Norton stuff)
and i must say, it works great ! :D
ghost multiple clients on a network, via ftp! try it out!
http://freshmeat.net/projects/g4l/
Met vriendelijke groet / With kind regards,
Martijn Marsman <mmar...@afab.nl>
System Engineer
AFAB Geldservice B.V.
Maxim Vexler
> On Thursday 15 December 2005 13:25, William Ballard wrote:
> > I literally would be unable to use Microsoft Windows if I couldn't stay
> > mostly booted in Debian and manage that godawfulness with partimage.
> >
> > Every time I boot into it I restore a clean partimage of XP, let it puke
> > all over itself, then restore the cleanness.
> >
> > It's the only thing that makes patching Windows remotely tolerable.
> >
> > Eventually partimage will stop working on new versions of NTFS, and it
> > seems to not be maintained anymore. It was removed from Sarge.
> >
> > Are there other tools that work like Ghost but in Linux? Partimage is
> > great.
>
> You could use dd, and compress the image, or use an emulator with a COW disk
> image.
>
Here is a cool one liner for the above :
dd if=/dev/hda bs=1k conv=sync,noerror | gzip -c | ssh -c blowfish
user@hostname "gzip -d | dd of=/dev/hda bs=1k"
For more info, read : http://slice.med.uottawa.ca/public/manuals/ImageDisk.html
>
> --
> To UNSUBSCRIBE, email to debian-us...@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listm...@lists.debian.org
>
>
--
Cheers,
Maxim Vexler (hq4ever).
Do u GNU ?
Jon Dowland
> I dream of the day that windows will use swap partition instead of a
> swap file.... sure it made sense to have a swap file that could adjust
> on the fly when drives were small... but with most machines having
> 40GB + these days I can afford to dedicate a pretty significant
> portion to a swap partition and not need it to resize itself
In win95 you could specify a fixed filesize for the swap file, and
dedicate a partition to just that file if you so wished. I imagine
things are much the same now.
By contrast, on my work desktop I forgot to create a swap partition
so I use a swap-file, when necessary.
--
Jon Dowland
http://alcopop.org/
Joseph H. Fry
> I largely prefer ntfsclone from the ntfsprogs package over partimage.
> Partimage is nice but the command line based ntfsclone is far more
> flexible. Just check out the man page for some usage examples.
> The ntfsprogs package contains lots of other useful utilities a XP really
> can't do without.
> Cheers, P. *8^)
Can you configure ntfsclone to clone an NTFS partition but not include the
swap file or other files of your choice?
I dream of the day that windows will use swap partition instead of a swap
file.... sure it made sense to have a swap file that could adjust on the fly
when drives were small... but with most machines having 40GB + these days I
can afford to dedicate a pretty significant portion to a swap partition and
not need it to resize itself
Joe
Joseph H. Fry
> On Thu, Dec 15, 2005 at 11:20:50AM -0500, Joseph H. Fry wrote:
> > I dream of the day that windows will use swap partition instead of a
> > swap file.... sure it made sense to have a swap file that could adjust
> > on the fly when drives were small... but with most machines having
> > 40GB + these days I can afford to dedicate a pretty significant
> > portion to a swap partition and not need it to resize itself
>
> In win95 you could specify a fixed filesize for the swap file, and
> dedicate a partition to just that file if you so wished. I imagine
> things are much the same now.
>
> By contrast, on my work desktop I forgot to create a swap partition
> so I use a swap-file, when necessary.
That's true... however using symantec ghost, or MS's deployment tools to clone
a machine configured as you suggest results in the swap file being placed the
destination machine's c: in most circumstances. There are ways around it,
but they are far more complicated than they should be.
If you had swap partitions, then the windows kernel could simply scan the
available partitions for valid swap partitions and activate them at boot.
This would allow you to have multiple windows installations that share the
same swap space, would make backup and cloning tools easier, and allow MS to
develop an optimized file system for swap... though I understand that swap
files are only minimally affected by the filesystem they run on.
I suppose it's not a major issue... however I do like the swap partition idea
that -nix uses.
Paul Seelig
> Can you configure ntfsclone to clone an NTFS partition but not include the
> swap file or other files of your choice?
>
I don't know because i never bothered... ;-)
Paul Seelig
> There is also Ghost4linux (not the real Norton stuff)
>
> and i must say, it works great ! :D
>
> ghost multiple clients on a network, via ftp! try it out!
>
> http://freshmeat.net/projects/g4l/
>
I downloaded it, bootet it and what did i end up with? With
partimage. This g4l seems to be nothing more than a bootable mini
linux with a ncurses frontend for some utilities and partimage...
For ntfs partitions, i prefer the more mature ntfs support of the
ntfsprogs and the added benefit of loop mounting an NTFS image file.
BTW, to save an image via network one can use whatever the pipe
permits. Here are some samples from the man page:
Backup an NTFS volume to a remote host, using ssh.
ntfsclone --save-image --output - /dev/hda1 | \
gzip -c | ssh host 'cat > backup.img.gz'
Restore an NTFS volume from a remote host via ssh.
ssh host 'cat backup.img.gz' | gunzip -c | \
ntfsclone --restore-image --overwrite /dev/hda1 -
Stream an image from a web server and restore it to a partition
wget -qO - http://server/backup.img | \
ntfsclone --restore-image --overwrite /dev/hda1 -
William Ballard
> For ntfs partitions, i prefer the more mature ntfs support of the
> ntfsprogs and the added benefit of loop mounting an NTFS image file.
> BTW, to save an image via network one can use whatever the pipe
> permits. Here are some samples from the man page:
>
> Backup an NTFS volume to a remote host, using ssh.
>
> ntfsclone --save-image --output - /dev/hda1 | \
> gzip -c | ssh host 'cat > backup.img.gz'
>
> Restore an NTFS volume from a remote host via ssh.
>
> ssh host 'cat backup.img.gz' | gunzip -c | \
> ntfsclone --restore-image --overwrite /dev/hda1 -
>
> Stream an image from a web server and restore it to a partition
>
> wget -qO - http://server/backup.img | \
> ntfsclone --restore-image --overwrite /dev/hda1 -
You can't mount an image that has been saved with --save-image.
I tried ntfsclone and it works about as fast as partimage, and it's definitely less
cumbersome that partimage; however the resulting gzipped image file from a 20GB
partition with 2GB of actual data was about 60MB larger: 840mb versus 780mb. The
partimage image was also gzipped.
I'm also going to file a bug against ntfsprogs that ntfsclone should be packaged
separately from the rest of ntfsprogs. ntfsclone is actually useful; the rest of those
programs are either unnecessary or flat dangerous. The only thing they have in common
is they involve NTFS.
The fact that ntfsclone is packaged with a tool called "fixntfs" or somethign who's man
page says "always run this after running any of the other utilities in this package
before booting or your NTFS partition will be completely destroyed" makes me feel
squeamish about ntfsclone, although as I said it's a different animal and people report
it as stable.
Paul Seelig
> On Fri, Dec 16, 2005 at 12:48:07AM +0100, Paul Seelig wrote:
> > For ntfs partitions, i prefer the more mature ntfs support of the
> > ntfsprogs and the added benefit of loop mounting an NTFS image file.
>
> You can't mount an image that has been saved with --save-image.
>
Yes, if you use the --save-image parameter. But if you restore such an
image to a file instead to a partition, you can use loop mount that file.
> I tried ntfsclone and it works about as fast as partimage, and it's definitely less
> cumbersome that partimage; however the resulting gzipped image file from a 20GB
> partition with 2GB of actual data was about 60MB larger: 840mb versus 780mb. The
> partimage image was also gzipped.
>
But can you restore a partimage image to a file and loop mount the result?
> I'm also going to file a bug against ntfsprogs that ntfsclone should be packaged
> separately from the rest of ntfsprogs.
>
I love package fragmentation... :-/
> ntfsclone is actually useful; the rest of those
> programs are either unnecessary or flat dangerous. The only thing they have in common
> is they involve NTFS.
>
ntfsresize and ntfsfix are some other nice components of the ntfsprogs
package. I've benefited from both various times.
> The fact that ntfsclone is packaged with a tool called "fixntfs" or somethign who's man
> page says "always run this after running any of the other utilities in this package
> before booting or your NTFS partition will be completely destroyed" makes me feel
> squeamish about ntfsclone,
>
:-))))))
Don't worry. The ntfsprogs should be safer than the ntfs support of partimage.
> although as I said it's a different animal and people report
> it as stable.
>
ntfsfix helped me quite a few times to fix a ntfs partition which the
native WinXP chkdsk couldn't repair anymore... ;-)
Szakacsits Szabolcs
nospam...@carolina.rr.com (William Ballard) writes:
> You can't mount a [ntfsclone] image that has been saved with
> --save-image.
If you want a compressed and mountable image then use ntfsclone without the
--save-image option and with a compressed filesystem.
> I tried ntfsclone and it works about as fast as partimage, and it's
> definitely less cumbersome that partimage; however the resulting gzipped
> image file from a 20GB partition with 2GB of actual data was about 60MB
> larger: 840mb versus 780mb. The partimage image was also gzipped.
Two possible explanations:
1) Both patimage and ntfsclone save the used blocks based on the block
allocation bitmap, however partimage doesn't have consistency check while
ntfsclone has. This means if your ntfs is inconsistent (which is
unfortunately more common than most people would like it) then partimage
will save less data than needed and obviously you will lose those.
2) partimage used a higher compression option than the one was used with
ntfsclone, which could be basically anything given that one can have the
image in a pipe stream.
> I'm also going to file a bug against ntfsprogs that ntfsclone should be
> packaged separately from the rest of ntfsprogs. ntfsclone is actually
> useful; the rest of those programs are either unnecessary or flat
> dangerous.
They are much less dangerous than cp, tar, partimage, parted, etc. Over the
last three years there wasn't even one report about damaged ntfs (using our
code) even if they are pretty widely used (directly or indirectly over a
million users).
Actually due to their reliability, several serious problems were discovered
at least in the previously mentioned utilities: tar trashes any 4+ GB
sparse files for over a year when the --sparse option is used, parted
sometimes still corrupts partition tables with head number 240, etc.
> The fact that ntfsclone is packaged with a tool called "fixntfs"
Ntfsfix currently is distributed to fix corrupted NTFS which were corrupted
by the Windows NTFS driver, not by the new Linux NTFS code.
Originally ntfsfix was developed by the new Linux NTFS developers to "fix"
corrupted NTFS which were corrupted by the NT4 NTFS kernel driver 5 years
ago and which driver was developed then abandoned by their developers. That
driver is not used for years now and write was disabled 3-4 years ago.
> or somethign who's man page says "always run this after running any of
> the other utilities in this package before booting or your NTFS partition
> will be completely destroyed"
This was NEVER in the ntfsfix manual page, your claim is absolutely untrue.
I wrote ntfsresize, ntfsclone, worked on ntfsfix and I've never released
non-stable code. Here is the ntfsfix manual which has nothing even close to
what you're saying:
http://wiki.linux-ntfs.org/doku.php?id=man:ntfsfix
As a matter of fact, I was who rewrote the "5 years old" ntfsfix manual
this year
http://cvs.sf.net/viewcvs.py/linux-ntfs/ntfsprogs/ntfsprogs/ntfsfix.8.in?r1=1.5&r2=1.6
because it still referred to the old, dead NTFS kernel driver which was
never developed, maintained and supported by the new NTFS developers and
which had write disabled in the last 3-4 years.
All the utils in ntfsprogs and the current kernel code was written from
scratch to also support W2K, XP, W2K3, Vista and nothing is shared with the
old, broken and experimantal NT4 NTFS driver.
> makes me feel squeamish about ntfsclone, although as I said it's a
> different animal and people report it as stable.
Yes and that's not by accident but due to a lot of very careful work. It
was supposed to be always stable since I publicly released it, almost three
years ago. Ntfsclone is intensively used and also crucial during
development and regression testing.
Szaka
William Ballard
> This was NEVER in the ntfsfix manual page, your claim is absolutely untrue.
You're right; it's not in the man page. If I had read only that man page I wouldn't
have made that statement. I never installed ntfsprogs before a couple days ago; I'll
find you the document which lead me to make that statement. I installed the sarge
version and distinctly remembering that some tool in the package related to fixing
things was necessary AFTER running any of the other utilities before booting into it to
prevent corruption.
I'll find you the reference and quote it.
However, great informative post, thanks. However, don't reply directly to me.
William Ballard
> I'll find you the reference and quote it.
It's in the readme. However, you're right, it doesn't say "after running any of these
utilities." It says:
NtfsFix - Attempt to fix an NTFS partition that has been damaged by the old
Linux NTFS driver. Note that you should run it every time after you have used
the old Linux NTFS driver to write to an NTFS partition to prevent massive data
corruption from happening when Windows mounts the partition.
IMPORTANT: Run this only *after* unmounting the partition in Linux but *before*
rebooting into Windows NT/2000 or you *will* suffer! - You have been warned!
See man 8 ntfsfix for details.
Maybe NTFS write support in Linux has gotten as safe as houses; I don't know. I don't
mess with the crap because I don't need to. All I need is to back up and restore
partitions and READ ntfs. Which is why in my world I'd rather have just ntsclone and
not the rest of it.
If you disagree, just bear in mind this is my opinion. Thanks.
Andrei Popescu
I've been eager for an answer to a question and i always forget to google about it :)
Is NTFS support safe to use for writing? Are all functions implemented?
If no time to answer could you just please direct me where to look for this information. I know this is constant development, but i would be very interested in trying it out.
Andrei
Szakacsits Szabolcs
On Mon, 19 Dec 2005, Andrei Popescu wrote:
> Is NTFS support safe to use for writing?
It depends on what NTFS implementation you use. There are almost a dozen
different one which people greatly confuse and they often credit one's
faults to an unrelated one.
If you use the one from the open source Linux NTFS project (but not the
original NT4 NTFS driver which were experimental, developed by others and
kept trashing NTFS) then yes it was always safer than ext3, xfs or
reiserfs, mainly because until very recently it implemented only file
overwrite without change to the file size and volume resizing so basically
nobody would have much chance to corrupt anything even if they had tried
very hard.
> Are all functions implemented?
Only read, not all writer. And probably some bugs will appear too as more
write related features will be implemented. Just like in case of any other
filesystems or softwares.
> If no time to answer could you just please direct me where to look for
> this information. I know this is constant development, but i would be
> very interested in trying it out.