Anaconda, grub and XFS

[Jeremy Katz]

On Sat, 2005-10-08 at 12:00 +0100, James Pearson wrote:
> On 10/7/05, Jeremy Katz <katzj at redhat.com> wrote:
> > On Thu, 2005-10-06 at 17:24 +0100, James Pearson wrote:
> > > There has been an on going issue with installing grub on an XFS
> > > partition - anaconda will hang at the installing boot loader stage as
> > > grub 'spins'.
> > >
> > > The attached patch for 'booty' works round this problem by remounting
> > > the XFS file system that contains /boot as read-only and then as
> > > read-write before running the grub install command.
> > >
> > > This patch replaces the current XFS freeze/thaw work round that fails to
> > > work (a lot) more often than not.
> >
> > This feels like a hack for the fact that xfs_freeze doesn't work as it
> > was designed to -- it is specifically for things which need the contents
> > on disk to actually be what the kernel thinks is there.  And the
> > continued need of random hacks like this make me more and more inclined
> > to just disallow the use of XFS as a bootable filesystem.
> 
> But this is a 'hack' that replaces a 'hack' that doesn't work ... so
> the total number of hacks in anaconda won't increase :-)

Yes, but it's a hack that was added at the explicit request of the XFS
developers with a "this will fix the problems".  And having to
constantly change it because they don't want to have the same semantics
as all of the other filesystems which are even pseudo-supported is a
waste of time.

Jeremy