NTFS v FAT32

[LiamC]

IS NTFS more robust than FAT32.

Why I ask - last night I booted my PC and tried to burn some files to a CD. I got a lot of file read errors.

I tried chkdsk the drive (only possible at boot time) and it kept saying that the FAT32 drive was a non-XP partition and would Skip checking.

Booted into Win98 and tried to backup the most recent files to a ZIP archive, but it kept getting file read errors with invalid checksums and aborting.

Couldn't use safe mode as it took over 45 minutes to boot - the disk light would blink on and then sit dormant for 30 or more seconds.

Finally I installed a new copy of XP on a spare partition which allowed me to boot normally and chkdsk the partition - lots of errors and lots of cross-linked files -> but I can't save all the cross linked files (more than 80GB!) as I don't have enough free space.

I have backups, of important data and a lot of the rest can be downloaded again so I'm not too concerned. BTW, I checked the drive thoroughly with WD DataLifeguard tools and there is no physical drive problem so I'm assuming that the FAT got trashed somehow, which brings me to the question.

BTW, anybody no of any 3rd party tools that might assist in this situation?

 

[blakerwry]

NTFS is more robust than FAT... NTFS doesnt just get trashed cause it feels like it. It keeps a log to prevent that sorf of stuff.

anecdotally I've never had a problem with an NTFS partition, I have had weird problems with FAT since the beginning of windows.

 

[Mercutio]

NTFS is a journaling filesystem, and has multiple copies of its partition information stored elsewhere than the first sector of the partition. Those two facts alone make it better than FAT32.

 

[LiamC]

Thanks people. Does anybody know if Linux is finally able to read and write to NTFS partitions?

Merc, you aren't hinting that you are one of the elder gods are you? (Your sig) :-?

 

[time]

Liam,

You might care to try Zero Assumption Recovery. The author popped up on SR about eighteen months ago with a very early prototype, and he seems to have refined it quite considerably since then. You can do a decent amount of test recovery before forking over the US$99.

I haven't had occasion to use it personally; he just sounded pretty switched on, and there's some good advice on his website.

I also googled this list of repair tools for you.

GetDataBack looks interesting.

Or you could always throw real money at it with one of OnTrack's tools such as EasyRecovery DataRecovery.

How about you do a quick comparison and publish it here?

 

[Mercutio]

Ia! Ia! Cthulhu Fhtagn!

 

[time]

NTFS journalling doesn't have much to do with its robustness in terms of data integrity. Journalling file systems came about because scanning a drive can get seriously tedious with large amounts of data.

The added redundancy Merc mentioned has more to do with it.

And yes, Linux (Suse 9.0 at least) will happily play with existing NTFS partitions. I actually set up two SATA drives in RAID 1, formatted them with NTFS under Win2k, then was able to read and write to them seamlessly under Linux.

Mind you, IMHO you'd be better off sticking with ReiserFS.

 

[Mercutio]

ZAR does work very well. I tested it and reported positive results after its author posted on SR about it (and before superc started arguing with him). I was able to do a better job as a "control" (I zeroed out both copies of the FAT on a drive) with it than with a couple of commercial tools I had at the time.
I won't say I was persuaded by superc's argument, but Alexey's license - requiring Americans and Europeans to pay but not Russians - was somewhat disheartening.

I'm glad you reminded me of that program, time, I couldn't remember its name.

Easy Recovery Pro has been a godsend for me. I have a technician license for it. It's only one tool in my box, but if you're talking data recovery, it's a very good one.

 

[Mercutio]

I don't remember all the details, but with ReiserFS, it's apparently possible to get the FS in a state that leads to whole-partition data corruption. I know that's somewhat true of any FS, but in this case it's something the developer is aware of, and has essentially said "I don't care." I read many angry posts on /. on the subject and could probably find a decent link if I looked a little harder.

Given that linux has so many other choices for a journaling FS, why choose the one that might trash your data for some tiny performance gain? ext3, JFS and XFS are all options, too.

 

[sechs]

NTFS is more robust than FAT... NTFS doesnt just get trashed cause it feels like it. It keeps a log to prevent that sorf of stuff.

anecdotally I've never had a problem with an NTFS partition, I have had weird problems with FAT since the beginning of windows.

I've actually had Windows bitch about NTFS partition x being corrupt, but read and write to it perfectly.

And you can hard-reset your computer as much as you want, and Windows will still boot!

 

[LiamC]

Thanks time, but I'm without internet at home at the moment, otherwise I would do a review.

The reason I ask about Linux as all data will be on an NTFS volume and I'll just mount that rather than clone it to a ReiserFS/ext3 volume. BTW, I always found Reiser to be a lot faster than ext3 - at least in the 8 series Mandrakes v ext3 in RH


Merc

http://ars.userfriendly.org/cartoons/?id=20040125



Back up a few too for some godly humour...

 

[blakerwry]

I was surprised mandrake now can read/write to NTFS... i believe knoppix can read by default... i would probably be a little hesitant to do much writing, but I'd have no problem with a read only mount.

 

[time]

... with ReiserFS, it's apparently possible to get the FS in a state that leads to whole-partition data corruption ... I read many angry posts on /. on the subject and could probably find a decent link if I looked a little harder.

I looked into this a while ago and satisfied myself that ReiserFS has been sufficiently stable for at least a couple of years. There's a reference here to some corruption-preventing patches two years ago. It says "Please note, this corruption only affects ReiserFS in the 2.5 development kernel."

This piece from a little under two years ago reckons "at the time this article was written, it [ReiserFS in kernel 2.4.18] was nearly 3 months old and there still haven't been any major problems found in the code."

So I'd be grateful if you can find anything concrete. My searches in the last half hour came up with repeated rumours but little else. One idiot claimed that there were many more Google results for "ReiserFS corruption" than "ext3 corruption". I actually get slightly more for ext3. I also tried "ext3" excluding "corruption" (and the same for ReiserFS) and the count was virtually the same. :roll:

... why choose the one that might trash your data for some tiny performance gain? ext3, JFS and XFS are all options, too.

I also researched this a while ago, and found:

XFS is optimized for arrays. Otherwise, it can be slow. The article I linked to earlier mentions that "It turns out that XFS 1.0.x had the unfortunate tendency of frequently mangling recently modified files if your server happened to crash or unexpectedly lose power." Obviously that was fixed, but it demonstrates that corruption is by no means confined to a single FS.

JFS doesn't add anything new except lower performance. It looks like a dead end to me. Recent bug fixes include:
- Fix resize errors
- Fix possible trap/data loss when fixing directory index table
- Fix hang while flushing outstanding transactions under heavy load
- Avoid deadlock under very heavy load

Ext3 doesn't scale well, it occasionally struggles under heavy load, and the full journalling is just too slow for the vast majority of applications. But it's still a realistic alternative to ReiserFS.

Finally, again from the above article, "I'm very eager to start using ReiserFS again and I plan to use it as my root filesystem when I next reload my development workstation. I'm sure there are many other ex-ReiserFS users who will be moving back to ReiserFS now that things have calmed down in kernel-land. Frankly, it's quite hard to live without ReiserFS once you've seen how its small file performance can boost the performance of certain applications."

This observation is backed up in numerous benchmarks all over the place.

The only FS with significant new technology that I've found is the upcoming Reiser4. It won't suit many applications, but in most respects, this completely new design blows everything else into the weeds.

 

[LiamC]

FWIW, I only lost my entire Games folder and one of my code sub-folders (programmer).


Installing a fresh/minimal copy of XP on a spare partition and running chkdsk recovered the partition with the abovementioned losses. The good news is I now have up to date backups on CDRW and on a network drive - I just have to remember to keep it up to date.

To which end I am going to write myself a little proggy to backup to various media, and if I can find a good and/or royalty free compression component/plug-in/algorithm, so much the better.

One of the things this will have is the ability to take a snapshot and do a diff later on, and only backup what I select - or to compare from some reference folder and only backup the different/new/added files. Suggestions welcome