ext3 Bug in 2.4.20

[Andrew Mathews]

Bill Campbell wrote:
<snip>
> 
> I think that ext3 offers something that XFS doesn't, journalling
> of the data as well as the directory metadata.  I found this from
> the postfix mailing list:
> 	http://www.stahl.bau.tu-bs.de/~hildeb/postfix/ext3.shtml
> 
> Bill
> --

Which leads to the question of why ext3, if it is a "true" journaling 
filesystem, still has the requirement of enforcing by default, the max 
mount counts for running fsck? Yes it can be disabled by running tune2fs 
-c0 but this shouldn't be necessary. There's also the issue of running 
ext3 under LVM which doesn't perform fsck's properly due to the fact 
that fsck tries to access the partition via the LVM device listed in 
/etc/fstab rather than using the device the kernel recognizes such as 
/dev/sda. This causes the buffer cache to get out of sync, effectively 
eliminating the benefits of a journal. This is where ext3 really comes 
up short in the filesystem debates, especially when compared to XFS, and 
as long as it continues the need to retain the hooks into ext2, those 
limitations and liabilities will remain. Joe Linux may never be 
concerned, nor have any need to be about issues like this, but they 
still exist nonetheless.
-- 
Andrew Mathews
---------------------------------------------------------------------
  10:10am  up 1 day, 23:31,  2 users,  load average: 6.23, 6.15, 6.15
---------------------------------------------------------------------
I don't want to bore you, but there's nobody else around for me to bore.