On Mon, Nov 21, 2011 at 10:32:07PM -0600, Eric Sandeen wrote:
> On 11/21/11 6:23 PM, Dave Chinner wrote:
> > On Tue, Nov 22, 2011 at 10:53:16AM +1300, Richard Scobie wrote:
> >> Is there currently now any situation where xfs_check would be used
> >> in preference to xfs_repair?
> >> If not, perhaps xfs_check could be deprecated.
> > xfs_check is one of the ways we test that xfs_repair is doing the
> > right thing. Having two implementation that you can use to compare
> > results is a good thing.....
> What about for end users though? I'm not sure there's much need
> for end users to be comparing xfs_check against xfs_repair in general,
> anyway ...
Right, but that doesn't mean it needs deprecating as that implies
complete removal at some point in the future. I'd prefer to keep it
around as we get most of what it does for free as it uses the xfs_db
infrastructure to do all it's work.
> Often enough I see users using xfs_check just because it's there,
> and running into trouble... it seems reasonable to warn the
> casual user against it, or at least recommend xfs_repair -n
> instead. What do you think?
The patch to modify the man page to advise use of xfs_repair is
sufficient, I think.