[OpenAFS] rxk5 Mainline Issues?

Marcus Watts mdw@bruson.ifs.umich.edu
Fri, 06 Nov 2009 15:13:41 -0500

> Date:    Fri, 06 Nov 2009 18:43:53 GMT
> To:      Derrick Brashear <shadow@gmail.com>
> cc:      Evan Macbeth <emacbeth@sinenomine.net>,
> 	 "openafs-info@openafs.org" <openafs-info@openafs.org>
> From:    Simon Wilkinson <sxw@inf.ed.ac.uk>
> Subject: Re: [OpenAFS] rxk5 Mainline Issues?
> On 6 Nov 2009, at 17:16, Derrick Brashear wrote:
> >
> >
> > Check out this message, the summary of the hackathon. It covers the
> > action items for rxk5 needed for suitability to standardize:
> > http://www.openafs.org/pipermail/openafs-devel/2009-October/ 
> > 017096.html
> Marcus has also summarised the issues to the afs3-stds list, along  
> with his feelings of how desirable / achievable each feature might be:
> http://michigan-openafs-lists.central.org/archives/afs3-standardization/2009-Oc
> tober/000512.html
> (There are additional issues, beyond those from the hackathon, on that  
> list - in particular, many sites indiciated at the Rome Workshop that  
> a flag day transition to rxk5 would not be achievable for them)
> S.

I should comment a bit about timeline.

At the hackathon, it turns out the openafs gatekeepers have a schedule
whereby 1.6 comes out in "6 months", and nothing regarding improved
security mechanisms can happen "to the mainline" before then.  So that
schedule has nothing to do with rxk5; it's an absolute constraint.

Given that, our plan (matt & I) for rxk5 is:
/1/ discuss the various points that came out around the hackathon.
/2/ work on splitting out patches - target: "6 months out" openafs mainline.
/3/ release several intermediate snapshots between now & 6 months out:
	goal: provide sites opportunity to experiment with technology.
/4/ produce one or more updates of the rxk5 protocol document.

Some of the points that came out aren't actually all that
particular to rxk5.  maxcalls needs a proposal of its own
and affects rxgk and rxkad equally.  Heimdal has already been
discussed - it's just a "code wrangling" issue.

Things you can do to help:
/1/ if you value security or improved security mechanisms,
	let the gatekeepers, and also the rest of us, know.  If there are
	compromises on security you are willing to make to get improved
	security sooner, say so.  The gatekeepers are going to make
	decisions on priority based on their impressions of your input,
	and their own input.  If you don't give input, you know what
	priority they'll assign.
/2/ participate in the discussions.
/3/ try an rxk5 snapshot.
...and, of course, if you want to do more: talk to Matt and I.
We'd be delighted to work with you.

					-Marcus Watts