[pnfs] CAT (connect-a-thon) Draft-19 ?

[Marc Eshel]

Copy the pNFS list.



Benny Halevy <bhalevy at panasas.com> 
04/28/2008 12:45 PM

To
Robert Gordon <Robert.Gordon at Sun.COM>
cc
Glasgow_Jason at emc.com, "William A. (Andy) Adamson" 
<andros at citi.umich.edu>, Anatoly Pinchuk <apinchuk at bluearc.com>, Ricardo 
Labiaga <Ricardo.Labiaga at netapp.com>, "J. Bruce Fields" 
<bfields at fieldses.org>, Marc Eshel <eshel at almaden.ibm.com>, Benny Halevy 
<bhalevy at panasas.com>
Subject
Re: CAT (connect-a-thon) Draft-19 ?






On Apr. 23, 2008, 4:10 +0300, Robert Gordon <Robert.Gordon at Sun.COM> wrote:
> We're assuming that everyone will be testing using the
> draft-19 dot x-file...  It's a silly question, i know..
> 
> Robert.. 

We're practically at draft-ietf-nfsv4-minorversion1-dot-x-04.txt
(corresponding to draft-21).
Your question is a bit tricky because the actual dot-x differences
between draft-19 and draft-21 are quite subtle...
http://tools.ietf.org/rfcdiff?url2=http://tools.ietf.org/id/draft-ietf-nfsv4-minorversion1-dot-x-04.txt


Shows these meaningful changes:

a. struct devlist_item4, which was already stale in draft-19's dot-x file,
   was removed.

b. bool fss_absent; added to struct fs4_status.
   and that's referrals specific and was not implemented yet here.

c. SEQ4_STATUS_DEVID_DELETED_ALL flag was removed from the protocol.
   We haven't implemented this yet.

The question that most interests me right now is how people are going to
implement layout stateid, draft-19 or draft-21 (+ Spencer's latest text) 
style.
In particular, usage of zero stateid and lrs_present on LAYOUTRETURN 
response
for the last layout segment.

Benny