Subject: Re: File system performance
To: Christoph Hellwig <chhellwig@gmx.net>
From: Jaromír <jdolecek@netbsd.org>
List: tech-perform
Date: 02/23/2001 21:09:13
Christoph Hellwig wrote:
> Ext2 sets a dtime field in the inode if the file is meant to be deleted.

Yeah, but what if the change didn't get into the disk (since it's
written asynchronously, this might happen quite frequently).

> > > I usually lose more data on NetBSD - but fsck doesn't even NOTICE
> > > all lossage.
> > 
> > Whose fsck doesn't notice?
> 
> fsck_ffs on NetBSD (without softupdates, that is the comparism) compared
> to e2fsck on whatever OS you want.
> 
> > What doesn't it notice?
> 
> File data.

How would you exactly know if the given block is 'correct' or 'not
correct' ?
e2fsck 'notice' file incosistency by reconnecting individual unknown
blocks under lost+found. Not quite nice, especially since you don't
know where the blocks belong :)

Of course, fsck_ffs warns if the file's inode says there are more
blocks in the file than what's allocated. It cannot detect
if a block was changed and the change did not get to disk. Same way,
e2fsck cannot detect such thing.

Unless I'm confused, of course :)

Jaromir
-- 
Jaromir Dolecek <jdolecek@NetBSD.org>      http://www.ics.muni.cz/~dolecek/
@@@@  Wanna a real operating system ? Go and get NetBSD, dammit!  @@@@