[sclug] Well, *that's* not supposed to happen...

Andy Hayward ach1000 at gmail.com
Thu May 24 19:31:33 UTC 2007


On 5/24/07, David Given <dg at cowlark.com> wrote:
> I do know that you have to explicitly enable the journal with tune2fs
> --- it doesn't happen automatically when you create the filesystem ---

You may be using the wrong command:

$ mkfs -t ext3 ..

or

$ mkfs.ext3 ...

> There seems to be no advantage to ext2 (and yes, I have checked that
> it's being mounted with the ext3 filesystem instead of ext2).

ext3 has plenty of advantages.

> The fastest and most reliable filesystem I've met so far is reiserfs, but it's
> a monster chunk of code and doesn't have a good reputation (a while back
> there was a hash collision issue that meant certain combinations of Debian
> packages couldn't be installed at the same time, for example.

Reiserfs is awful, and I wouldn't trust it with any data that I
actually cared about.

There are known attacks against all three hash functions used for the
directory index.

  http://marc.theaimsgroup.com/?l=reiserfs&m=110434667931228&w=2

And worse is Hans Reiser response to this:

  http://marc.theaimsgroup.com/?l=reiserfs&m=110435438120364&w=2

Secondly, if you have to run reiserfsck on a reiserfs filesystem that
contains an image of another reiserfs filesyste, reiserfsfsck can't
tell the two filesystems apart and will happily mix them up.

Both JFS and XFS have their strengths, but for the vast majority of
purposes ext3 (with dir_index enabled and data=journal) is sufficient.

-- ach



More information about the Sclug mailing list