[OpenAFS-devel] Moving Forwards

[Jeffrey Altman]

This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enig31E03B62FAF9D62145BDDA61
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

On 9/10/2012 12:17 AM, Troy Benjegerdes wrote:
> Two examples where it does appear the current gatekeepers are roadblock=
s
> because some or all of the gatekeepers have a vested financial incentiv=
e
> to ensure all development goes through their respective organizations:
>=20
> 1) specification of a wire protocol and incremental implementation, so =

> we can see that IPv6 support is progressing.

There is no requirement that IPv6 or other work go through Your File
System, Inc.  If you are willing to do this work for me without costing
me a dime I would be thrilled.  However, what I as CEO of YFSI is not
prepared to do is spend my money on work that no end user organization
is prepared to pay for.

> 2) rxgk (and the rxk5 implementation, which, to my understanding *worke=
d*
> but was dropped)

rxgk has been funded entirely by YFSI.  The protocol has been documented
and given to afs3-standardization for review but it has never received
sufficient review to pass a last call.  The vast majority of
infrastructure changes to OpenAFS necessary to deploy rxgk have been
committed to OpenAFS master.

One of the primary problems with rxk5 is that in order for existing
cells to deploy it, there must be a flag day.  Flag days make rolling
out a new security protocol unrealistic.

> Is there any public documentation of past code changes that resulted in=

> data loss and/or destabilization so I can write test suites?

* The acceptance of the Demand Attach File Service changes blocked the
release of 1.6 for years.

* Idle dead timeouts being added to the clients.  Fixed in 1.6.1.

* Copy on write data corruption introduced in 1.6.0 and Fixed in 1.6.1.

> So if I get a working rxk5 on the latest codebase, does anyone have any=

> ideas for a name/trademark better than 'TFS' (troy's file system) so th=
at
> there is no confusion between that fork and OpenAFS? If there are going=
=20
> to be stupid legal arguments that IBM owns the .xg files and I can't=20
> actually distribute a modification that moves forward from OpenAFS,=20
> then I'd also like to know now so I can start looking at other more=20
> open protocols to migrate my files to.

IBM has licensed the .xg files under the IBM Public License 1.0.  You
can ask a lawyer to review the license.

Regardless of what you wish to call your file system product, I will
advise you not to modify the AFS3 protocol outside of the
standardization process.  Do so risks interoperability problems.

> If the gatekeepers wish to remain relevant, than I would please request=

> they come up with a workable IPv6 wire protocol that can be incremental=
ly
> developed and deployed to get working isolated cells running v6 within
> 6 months.

The reason that no end user organizations are funding IPv6 work has to
do with the priorities of those organizations.  As CEO of YFSI I have
made the determination that there is not sufficient demand to warrant
working on IPv6 at this time.  There is a long list of other work that
is more important to potential YFSI customers.

I cannot speak for the priorities of other organizations.  Simply
because you want protocols and source code for your own business
prospects does not provide any incentive for me or anyone else to spend
money.

Things would be very different if end user organizations donated money
to a Foundation.  Then the Foundation would have resources that could be
prioritized based upon the wishes of those that contribute to it.
However, such a Foundation does not exist and not for lack of trying.
As a result there are no resources for the Elders (let alone the
Gatekeepers) to allocate to satisfy your request.

Looking through the commit history of the OpenAFS repository.  There are
a total of four minor patches that have been contributed by "Troy
Benjegerdes" between June 2005 and June 2006.  Yet here you are in
September 2012 demanding that I and the rest of the community put your
needs ahead of our own priorities?  How is that supposed to work?

Are you willing to commit to do the full work necessary to implement
IPv6 in the AFS3 protocol and OpenAFS implementation in a manner that is
backward compatible, does not require flag days, permits rollback to
prior OpenAFS versions when necessary, and passes all standardization
requirements?  If so, the gatekeepers are willing to provide advice,
review designs, and review code contributions just as we do for everyone
else.

Jeffrey Altman



--------------enig31E03B62FAF9D62145BDDA61
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)

iQEcBAEBAgAGBQJQTXYKAAoJENxm1CNJffh4TmgH/1wFLXAH/MN88Ne5m/R2+kuP
PPIuv2RwfAWg2dDNatZkxuQgfNhEAZRqFhdlGFYoonytFKyyER1bjqS33C27pzik
XiA7sbPYgE67DJQ/bltHrbXlBK56XRaF7uTj/uS/piSiYyV+ZksuAexO9XQWqmyx
pvw+/Mf3PcepzF2eDCLITI7UHHkxrOa0paLEWmEdZ134b4IeEWk6otQaTQ7RlrOt
czvekCA7XPNpHbyJQB+tlcppZBl3+1ynBNRkBQIfzgyWkmeciQQ9YqotFmpBGd/a
TqpBL4LpR4C5ZGk01VHDCRIT96dBN6Z5nTCbJ2vrYmBiX3XT0MstRJfKUbbpEOY=
=nq4s
-----END PGP SIGNATURE-----

--------------enig31E03B62FAF9D62145BDDA61--