Re: [PATCH] Give kjournald a IOPRIO_CLASS_RT io priority

[Theodore Tso <tytso@mit.edu>]

On Sat, Oct 04, 2008 at 01:14:45PM -0400, Joseph Fannin wrote:
> On Sat, Oct 04, 2008 at 10:12:52AM -0400, Theodore Tso wrote:
> 
> > Anyway, here's the patch (against ext4, although could pretty easily
> > move this to ext3 --- but you can mount an ext3 filesystem as ext4,
> > although for the moment you do have to run the command "tune2fs -E
> > test_fs /dev/hdXX" first) [...]
> 
> Shouldn't he also be sure to mount the FS with the "noextents" option,
> or has ext4 learned not to to enable extents if they're not already
> enabled?

The latter; ext4 will no longer enable extents if not already enabled
as of commit e4079a11, dated July 11th.  That means 2.6.26 with any of
the ext4 patchsets (highly recommended if you want to use ext4) or any
mainline release post 2.6.26-git4/2.6.27-rc1.

A number of the ext4 tutorials/howto's are available on the web page
are out of date, including the one on IBM developer works.  The best
place to get up-to-date information on using ext4 is
http://ext4.wiki.kernel.org.

Regards,

						- Ted

P.S.  Note that there will be advantages to using even an unconverted
filesystem with the ext3 filesystem format with the ext4 code base,
even if you don't enable extents or any other ext4-specific feature;
the delayed allocation and soon-to-be-added inode table readahead
features (in the 2.6.27-rc8-ext4-2 patchset, and which will be pushed
during the next merge window) in the ext4 code base will improve
performance even on an "ext3 filesystem".