Minutes of the P2PSIP WG meeting at IETF 84
TUESDAY, July 31, 2012
1520-1650 (Afternoon Session II), Regency F

Version: 0.2 (revised by Marc Petit-Huguenin)

Note taker: Miguel A. Garcia 
Jabber scribe: Roland Bless

15:20 Administrativia ........................................ 10 min
       B. Rosen, CJ. Bernardos

15:30 WG Status Update ....................................... 15 min
       B. Rosen, CJ. Bernardos

The chairs go through the WG status slides.

Discussion about draft-ietf-p2psip-sip. Cullen indicates that not much
is missing, but it is taking more than anticipated. He estimates that
it will take 10 hours of work.

Brian asks for a volunteer to edit the draft. Thomas Schmidt volunteers 
to edit this draft.

Discussion about draft-peng-p2psip-snmp-04. Brian indicates it is
probably a good document to have, but it requires a lot of work.
There was a call for adoption, but the chairs concluded that we are
not ready yet: the document needs further work. Brian asks people who
commented to clearly indicate what needs to be done to get a good
document. The authors need guidance to make a good document out of
it. The chairs call for reading the document and comment for how to
improve the document.

DISCO and SHARE documents. These will be adopted as WG items.


Diagnostics draft. Is there an objection to schedule WGLC last in the
rotation of WGLC? Seems not. It will go to WGLC in about 2 or 3 months
from now.

Cullen asks for re-question again the Diagnostics after his presentation.


15:45 RELOAD update .......................................... 15 min
   o draft-ietf-p2psip-base-22
     * Presenter: C. Jennings

Cullen goes through his slides. Cullen discusses the updates to -22,
indicating that HTTP errors are not used, but instead, application
errors. Marc Petit-H asks why 401 cannot be used. Cullen indicates
that this is not the HTTP request, but the application. Marc asks why
don't we use HTTP basic and HTTP digest? Cullen indicates that some of
these errors are not related to authentication, so we still would need
them. EKR indicates that Basic is wrong, the way to do it correctly is
in the body. Cullen: on most web platforms you might not be able to
access the password servlet. EKR suggests to take a hmmm. Brian
presents the two alternatives:
a) As per slide. Username password passed in the form request
b) Passing username password as HTTP Basic and Digest.
Hmm reveals support por a)

Cullen asks if he has missed an error code. Marc Petit-H indicated
that he has recently implemented this, and he hasn't missed any error
code.

slide: Adrian Farrel discuss
Brian as an individual asks how a user detect faults and repairs
them. Cullen indicates that half of the draft is about that
question. Brian asks why hasn't this been answered to the discuss. EKR
indicates that this group does not have enough operational knowledge
to address these issues.
Brian asks if people agree with the issues raised in the discuss. Marc
says that Diagnostics is a good tool to discover faults. SNMP also is
a good tool.
Brian indicates that pushing Diagnostic might be something to
consider.
Gonzalo indicates that the next step is to talk to Adrian.

Marc indicates that he has compiled a list of pending issues in the
base draft. Marc presents his slides.

Slide 2: EKR disagrees with the proposed solution. The text will be clarified
to say that the hash does contain the length prefix.
Slide 3: agreed to make everything based on 1. 3rd paragraph will be changed
to "search through the finger table entries from i=1 and ..."
Slide 4: It is agreed to delete multicast. Everything related to multicast,
anycast and broadcast will be removed from the draft, so there is no need for
naked Ping.
(there is no Slide 5)
Slide 6: Cullen thinks there is nothing to update in the draft.That needs to be
checked with Michael.
Slide 7: Agreed. "Connection table" to be replaced with "Neighbor table" in
10.7.4.1.
Slide 8: The text will change to say that the configuration file will
always be signed. EKR: Read the configuration file, get the key out of
it, ... Agreement to update the draft. Configuration files retrieved from the
configuration server are *always* signed.
Slide 9: Cullen highlights the problems of implementing the
propsal. EKR indicates that the server does not need a dB of listing
all the nodes with the certificates; can use a hash with a
secret. Cullen: the enrollment server SHOULD return the same Node-ID
it had. No clear conclusion, it needs to be taken to the list.
Slide 10: Marc proposes to remove anycast. DNS is needed. If someone
needs anycast, it can be written in a different draft.
Slide 11: Cullen. We may need something more specific:
"reload-config". Agreed. The agreed path is "/.well-known/reload-config"
Slide 12: EKR agrees with the proposal.


16:00 One Hop Lookups Algorithm Plugin for RELOAD ............ 10 min
   o draft-peng-p2psip-one-hop-plugin-02
     * Presenter: Peng Jin

Peng Jin goes through the slides.

Brian asks who has read the document. Cullen has read it and finds it
valuable. Cullen indicates that this algorithm will be much easier to
debug than Chord. Brian encourages to continue the work. Right now the
WG has to focus on closing documents rather than getting new ones.


16:10 Next steps ............................................. 15 min
       B. Rosen, CJ. Bernardos

Consider bringing Diagnostics earlier.

16:25 AOB