IETF 79  GROW WG  Friday Nov 12 9:00AM

Chair Report - Peter Schoenmaker -
  2 documents in IESG Evaluation - Graceful Shutdown and MRT
  new documents - unique origin ASN's - Danny McPherson
  waiting new reviews - virtual aggr.

Pierre Francois -
  presented latest BGP Graceful Shutdown draft
  draft-ietf-grow-gshut-02
  http://www.ietf.org/proceedings/79/slides/grow-4.pdf
  no questions

Elisa Jasinska -
  presented on draft-jasinka-ix-bgp-route-server-01
  http://www.ietf.org/proceedings/79/slides/grow-3.pdf
  debate on splitting document into two due to BGP
  protocol changes -- vote 8 for two documents versus 8 against
  John Scudder - AD's will meet on splitting document and come to compromise
  no other questions

Terry Manderson -
  presented on draft-ietf-grow-geomrt-00
  http://www.ietf.org/proceedings/79/slides/grow-1.pdf
  presented change involving creating a new dump subtype
  no objections to issuing a new -01 draft based on presentation

Danny McPherson -
  presented on draft-mcpherson-unique-origin-as-00
  no slides
  documenting operational practice
  using same origin AS for anycast has issues when
  detecting route leaks.  No descrimator to determine source.
  using a unique origin AS enables detecting route
  leaks, and RPKI work.   Requires a unique origin
  for each anycast instance.  32-bit AS space should
  be sufficient.  RIR's will need to determine if justified
  as critical infrastructure.  Biggest issue is conservation
  of AS space.   Example, detecting root nameserver prefix
  leaks.   Several comments on ML.  Would like to issue
  a new draft and publish as operational routing practice.
  Comments -
  Wes George - would be helpful putting together a diagram showing
   fundamental issue.
  Danny - will put slides together to illustrate and send to ML.
  Chairs - confirmed interest in WG and will accept document.

Tony Li - 
  presenting draft for Enke Chen
  draft-chen-ebgp-error-handling-00
  http://www.ietf.org/proceedings/79/slides/grow-2.pdf
  currently - all errors reset BGP connection
  Danny McPherson - one question - can you summarize IDR feedback/pushback?
  Tony Li - lots of pushback - may not be noticed by operators, blackholes,
   silent loss, is TRAP sufficient notification
  Danny McPherson - there are protocol lessons to be learned
  Jake Khoun - could be signaled through negotiations so that it
    could be known how errors will be treated.  Signal through
    capability negotiation
  Chris Liljenstolpe - SNMP traps - should be generic - informational level
    alert - could be syslog - however router vendor normally signals and
    use whatever provider has wired for notification.
  Ruediger Volk - Good stuff, go forward  - trap needs to be done.  relies
    on good diagnostic info - may not fit entirely in a trap message.
  Peter S.- how would you fix an error?  operator manually reset connection?
  Tony Li - one way - get a refresh, but could get same error. Perhaps,
    partial refresh of bad.  Real fix -call upstream to fix code!
  John Scudder - Do what you do when you have gratuitous session resets.
    Value is panic according to own schedule and call vendor.
    Call for WG adoption on IDR - have comments - see IDR list.

  Chen Li? from jabber - how bout notifying the BGP speaker of bad update.
  John Scudder - Another draft - debug notification draft by Raszuk.
    In current draft, the sender get's no notification on error.