OP LOCK4: question about open_to_lock_owner4:lock_seqid

[J. Bruce Fields]

On Thu, Jul 03, 2008 at 03:41:29PM -0400, Rick Macklem wrote:
> I found this in the ietf archive:
> 	http://www.ietf.org/mail-archive/web/nfsv4/current/msg01253.html
> 
> It is specifically discussing the recovery case, but the basic problem is
> that the RFC doesn't tell you if the lock_seqid in the open_to_lock_owner4
> case should be incremented when the Lock Op fails, due to an error.
> 
> As such, some clients might increment it and I believe a server might want to
> accept the retry with a non-zero lock_seqid?

OK, sure, I'd forgotten that.

I'd still recommend the client just use seqid 0 in that case, and that a
server always just accept any seqid rather than check for this
particular case.

--b.