[FoRK] So why the fuck aren't checksums built into the fucking filesystem?

Stephen D. Williams sdw
Tue Jul 12 19:33:05 PDT 2005


He mentioned the FAT, i.e. file allocation table.  I'm not sure of the 
details for NTFS, but on some filesystems it is possible to run out of 
file or allocation range slots/bits before running out of space.  On the 
old System 5.2/3 filesystems, inodes were limited.  It was configurable, 
but in the System V.2 days the hard limit was 64K inodes.  I passed this 
running a Usenet server in 1992 and rewrote INN to use ARC archives for 
article storage.

sdw

Joe Barrera wrote:

> Aw crap. OK. The problem is that I have one tower that I use as a file 
> server,
> with everything else as laptops (albeit with a couple firewire/usb2 
> disks).
> And it's using several NTFS striped disks (60 or 80 GB each) to build
> ~300 GB virtual disks. So I'm going to have to do a lot of shuffling 
> to get
> this all over onto ext3, assuming it supports striping (I assume it 
> does).
> If it doesn't, then this is going to have to wait until I have a couple
> spare 300 GB disks. How irritating.
>
> BTW I don't think I've ever pushed any of my filesystems past 95%
> capacity, so I don't know if the bug you described would apply.
>
> - Joe
>
> Justin Mason wrote:
>
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>>
>> Joe Barrera writes:
>>  
>>
>>> Luis Villa wrote:
>>>
>>>   
>>>
>>>> Why are you putting mission critical data like mp3s on NTFS? :)
>>>>     
>>>
>>> Sigh. Legacy. Would Linux really do better?
>>>
>>> NTFS is supposed to be a journaling FS -- It really shouldn't
>>> suck as bad as it does.
>>>   
>>
>>
>> but it does.  FWIW, NTFS had an even nastier bug previously.... if an 
>> NTFS
>> v4 fs ran out of FAT space (or whatever it is that passes for an inode
>> index over there), it'd simply go haywire and lose track of the
>> filename->data mapping -- undeleteable files, files that reappear after
>> they'd been deleted, and files that appeared fine, but that had no 
>> data on
>> inspection.  The latter in particular made for lots of non-functional
>> backups in the case I heard of. ouch.
>>
>> Save yourself a lot of trouble and just use ext3 ;)
>>
>> In my experience, I've never had trouble with ext2 or ext3 filesystems.
>> Reiserfs, now, that's lost data -- and ext2 has had some hairy fscks,
>> but in the ext* case the data was always still there.
>>
>> - --j.
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v1.2.5 (GNU/Linux)
>> Comment: Exmh CVS
>>
>> iD8DBQFC1Er5MJF5cimLx9ARAmn5AKCG14eGIHnWNxyAFrKTKTzSjpQcugCgmNW1
>> fCw+2vi16LwLj2Bc5FcAJI0=
>> =jW2i
>> -----END PGP SIGNATURE-----
>>
>>
>>
>>
>>  
>>
>
>


-- 
swilliams at hpti.com http://www.hpti.com Per: sdw at lig.net http://sdw.st
Stephen D. Williams 703-724-0118W 703-995-0407Fax 20147-4622 AIM: sdw



More information about the FoRK mailing list