On Thu, Oct 24, 2024 at 09:38:04AM -0700, Darrick J. Wong wrote: > On Thu, Oct 24, 2024 at 01:51:02PM +1100, Dave Chinner wrote: > > I'd prefer that we take the latter path: ignore the first patch. > > This results in more flexible behaviour, allows existing filesystems > > with this issue to work without needing xfs_repair to fix them, and > > we get to remove complexity from the code. > > Do xfs_repair/scrub trip over these sparse chunks that cross EOAG, > or are they ok? Repair trips over them and removes them (because they fail alignment checks). I haven't looked to see if it needs anything more than to consider the alignment of these sparse chunks as being ok. I suspect the same for scrub - as long as the alignment checks out, it shouldn't need to fail the cluster... -Dave. -- Dave Chinner david@xxxxxxxxxxxxx
