On Mon, Apr 07, 2003 at 02:19:03PM -0400, Greg Freemyer wrote:
> I do a xfs_fsr and mapcheck nightly.
this is bad
if you're fragmenting badly enough to need this, you should address
the cause of that --- otherwise, for small moderate amounts of
fragmentation the loss of file-locality is worse for many people
> Is there more I could do to test the FS validity?
a xfs_repair passed over a snapshot would probably be slow and
overkill
> Specifically, it is a shell script for doing a read-only fsck of a
> mounted FS.
> It uses a LVM snapshot to perform fsck on, then echo's an error to a
> log file if it encounters a problem
>
> The idea is that a system with long uptimes could verify the FS
> structure is valid from time to time.
yes, this is what i mention about ... a lvn/frozen fs snapshot and
xfs_repair but NOT readonly would work but is overkill
xfs_repair -n (read only) isn't very useful
> It does not support XFS, but maybe it could be expanded to do that?
there is nothing to prevent what i suggest working with xfs
> If so, maybe this would be good candidate for the contrib section.
i'm sure people would welcome such a thing if you were to provide it
--cw
|