Re: Header cache
On Wed, Apr 02, 2008 at 09:45:36AM -0400, Ian Wat wrote:
> * Chris Bannister <mockingbird@xxxxxxxxxxxxxxxx> [2008-04-02 21:42 +1300]:
> > On Mon, Mar 31, 2008 at 08:10:04PM -0400, Ian Wat wrote:
> > > I'm having trouble with header cache on my box at work. I have the
> > > exact same config on my home box and it works fine. What happens is
> > > that mutt seems to consult the header cache but then discards it for
> > > some reason and then proceeds to download all the headers again.
> > >
> > > Home: Debian/i386 mutt 1.5.13 (2006-08-11)
> > > Work: OpenBSD/sparc64 mutt 1.5.16 (2007-06-09)
> > > I can post the output of mutt -v if desired.
> > >
> > > I have set $header_cache to a directory and noticed that the directory
> > > structure is different between work and home. However, that might be
> > > just differences between versions of mutt.
> >
> > What is the filesystem?
> > I know on ext3 you have to set something, maybe the same for the
> > filesystem on your OpenBSD/sparc64 box?
>
> The file system is FFS. What is it that you have to set for ext3? I
> had softupdates turned on but turning that off didn't help.
man tune2fs and search for dir_index. I hope I'm not steering you crook, but
I remember when the header_cache feature came available you had to do a
tune2fs -O +dir_index /dev/hdXX
and then
e2fsck -D /dev/hdXX
otherwise enabling header_cache would be a waste of time.
e.g On my system:
# dumpe2fs /dev/hda1 | grep features
Filesystem features: has_journal dir_index filetype needs_recovery
sparse_super
So dir_index is already a feature, although I don't remember turning it
on[1] maybe it comes as standard with ext3.
Hopefully someone will correct me if I'm wrong, it would be good for the
record.
[1] I would remember, because you have to run e2fsck -D /dev/hdXX on an
unmounted filesystem which would mean I would have had to boot from a
live CD in my case because everything is mounted on /, another good
reason to have /home on a separate partition. :-)
--
Chris.
======