On Mon, 2018-04-09 at 10:26 -0400, Daniel Gryniewicz wrote:
> So, NFS4ERR_OLD_STATEID can only happen in one circumstance: when the 
> State presented by the client doesn't match the State that Ganesha 
> expects.  In this case, it's the sequence number that's off-by-one. 
> This could be the result of a replay, but the code checks for this, and 
> the owners must be different in this case.
> 
> So, what seams to have happened is that the old state was destroyed, and 
> a new owner got a new state, and then the client presented the old state 
> again.
> 
> I'm not an expert in this code, so maybe I'm missing something?
> 
> (Note, the code is the same in 2.7, so no bugs have been fixed in this code)
> 
> Daniel


More or less.

A stateid represents the open/lock/delegation/layout state of a
filehandle at a particular point in time. The sequence ID is how we
distinguish the period of time for that state.

In this case, the client presented a stateid that was older than the one
the server had. So we have a situation where the server's idea of the
state of the file is newer than the client's.

A single OLD_STATEID error is generally not a problem, and usually
indicates a benign race between an OPEN/LOCK and a READ/WRITE. Once the
client gets the new stateid, things should start chugging along again.
If it keeps going then that usually means that the client didn't get the
new stateid for some reason.

A wire capture of the event would be ideal, if only to narrow down
whether the server sent the wrong seqid here or the client ignored it
for some reason. Testing more recent ganesha code might also be helpful
if that's an option.

Good luck!
-- 
Jeff Layton <jlay...@poochiereds.net>

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Nfs-ganesha-devel mailing list
Nfs-ganesha-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs-ganesha-devel

Reply via email to