On 17 Oct 2017, at 14:26, Trond Myklebust wrote: > On Tue, 2017-10-17 at 13:52 -0400, Benjamin Coddington wrote: >> Ah, yuck. I read 8.2.2: >> >> When such a set of locks is first created, the server returns a >> stateid with seqid value of one. >> >> .. and went from there. Is this a conflict in the spec? >> > > Hmm... I note that the equivalent section 2.2.16 in RFC7530 has been > changed to remove the bit about the starting value of seqid being > undefined. > > So, if the spec allows us to rely on the seqid always being initialised > to 1, then we might at least be able to detect that we have to replay > the open. > One way to do so might be to keep a count of the number of outstanding > seqids, and then force OPEN_DOWNGRADE/CLOSE to wait until that number > hits 0 (or until a state recovery is started). I can take that approach for another spin, but due to other priorities, I'll likely not come back to this until next week. Thanks for your comments. Ben -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
