[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: libtrash in fedora Core 5



On Tuesday 24 January 2006 11:55am, Paul A Houle wrote:
[snip]
>     Actually,  it would be quite nice to be able to manipulate the
> desktop trash via the CLI...  Good for writing scripts.

Another reason why I suggested that the GUI's and any new CLI program should 
share the same mechanism.  Of course, one could do this today without a CLI 
trashcan, but having it unified would make things much easier.

>     As for alternative filesystems,  ext3 and ext2 are the only
> filesystems I'd run on a machine where I cared about file integrity.
> Reiser,  XFS and JFS have file integrity issues AS DESIGNED -- integrity
> issues that have caused problems in production systems that I run.

I could say the same about ext2 & ext3, depending on *when* I was talking 
about them.  All of these filesystems (including ext2/3) have been marred by 
horror stories of lost data in the past.  But that's where such stories 
(generally) belong, since the bugs that caused them are long past.  Heck, I 
have a couple of good friends that vehemently refuse to use ext3 for that 
reason.  I know others that say the same thing for each of reiserfs, JFS & 
XFS.

The problem is, nobody tolerates errors in their use of a filesystem.  Since 
errors here usually mean data loss, we demand (near) perfection of that code.  
Unfortunately, human being being what we are, new filesystems' code does not 
get well tested until it's "gold".  Then the horror stories pile up:  "They 
said this filesystem was stable, but it's *not*!  I know; it ate my data.  
What a horrible design."

So, please, if you are one who lost data to other filesystems in the past, 
remember, bugs get fixed.  don't blame the design.  In a lot of ways ext3 has 
the more archaic architecture of the 4 (and in other ways it's farther ahead, 
and *please*, don't start a flame war/discussion of that here).

Anyway ... I'm not sure that Jeff was right when he talked about this being 
because of a deficiency in ext3.  Compared to ext2, perhaps.  But, in 
general, it is harder to recover deleted files from journaled filesystems 
that it is from block ones.  That said, Jeff has a point 

>     Undelete is possible on ext2,  but ext3 shreds directory entries for
> deleted files as a mechanism that (get this) protects fs integrity in a
> crash.  You're trading undelete and a bit of performance for no fsck
> after a crash:  it can take forever to fsck a 300G volume,  so most
> people choose no fsck.

I didn't know that's what ext3 was doing. Interesting ... I would guess that's 
to prevent the recovery code from really borking things in case the journal 
is corrupted.  Is that accurate?

Probably not an on-topic question, eh? :)
-- 
Lamont R. Peterson <lamont gurulabs com>
Senior Instructor
Guru Labs, L.C. [ http://www.GuruLabs.com/ ]

Attachment: pgpLN97VLpfAw.pgp
Description: PGP signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]