[Tux3] nfsd wishlist

J. Bruce Fields bfields at fieldses.org
Wed Dec 3 15:25:49 PST 2008

By the way, I sent a couple wishlist items for nfsd to Daniel in private
mail that probably should have gone to the list.  For now my
understanding is that nfsd exports aren't even supported, but for the
future a couple things that would be useful on a new filesystem are:

	- a change attribute (see the new file-versioning stuff that
	  went into ext4, which I haven't hooked the nfs server up to
	  yet--my bad): an integer which increases each time the file
	  data or metadata changes (basically, whenever ctime would be
	  considered for updating), needed for correct cache coherency
	  with nfsv4 clients (otherwise we use ctime and run into
	  time-resolution problems, even on filesystems that support
	  high-precision times)

	- Anything to ease creation of new volumes on the fly, either in
	  the filesystem itself or a volume manager underneath.
	  Otherwise people try to export subtrees, which has a number of
	  problems, not least that we can't really enforce them as a
	  security boundary: when we get a filehandle, there's no
	  efficient way to check whether the inode it points to is
	  accessible under an exported subtree....  We'd really like to
	  get people to the point where they only ever support entire
	  filesystems, but we won't get to that point without easier


Tux3 mailing list
Tux3 at tux3.org

More information about the Tux3 mailing list