On Thu, 26 Feb 2004, Anton Altaparmakov wrote:
> On Thu, 26 Feb 2004, Szakacsits Szabolcs wrote:
> > NTFS development? Regression testing? For example I've never had Windows
> > in my life until recently, just for the sole purpose to _validate_ the
> > Linux NTFS code I write. Soon after my disk crashed and I couldn't recover
> Yes, we could really, really do with a fully featured ntfsfsk utility in
> ntfsprogs. It wouldn't even need to fix anything, just report errors...
My disk physically died (IC25N040 in a IBM T40 laptop). It never spinned
up after a huge click.
Of course our tools work perfectly for software recovery (it took 6
minutes in my case) but because I didn't have anything on NTFS I had the
NTFS image on disk thinking that if the disk crashed I could recover
Windows quickly using IBM's XP Recovery CD set.
I was wrong. It took almost two months. IBM support is superior but
whatever helpful they were, the technology they provided didn't work.
I don't detail why, only mention when I managed to recover it in the end.
The 4 recovery CD's are apparently an IBM branded Disk Image from
PowerQuest (now Symantec). It uses Windows 98. The whole recovery process
took 1.5 hour attended. I had to sit next to the computer, replacing
CD's, hitting keys, reconfiguring BIOS, etc.
I tried the same with open source tools, ntfsprogs from a DVD (basically
ntfsclone the minimal image then ntfsresize to adjust to the underlaying
partition size). The recovery took 23 minutes unattended. Just put the
DVD in then start the recovery process and do whatever you want until it's
ready.
I guess the open source way could be still improved if the image is
compressed. I'm able to compress the XP image to about 50% so since the
bottleneck of the recovery process is the bandwidth of the DVD device thus
the recovery time could be reduced maybe to 12-15 minutes.
Proprietary recovery: 1.5 hour attended
Open Source recovery: 15 minutes unattended
IMHO there is a slight difference between speed, price and convenience ;)
Szaka
|