[08:33:00] --- benno@nlnetlabs.nl has joined [08:33:05] --- benno@nlnetlabs.nl has left [09:01:15] --- wouter has joined [09:03:06] --- Benno Overeinder has joined [09:23:18] --- Benno Overeinder has left [10:03:39] --- iljitsch has joined [10:04:12] --- dthaler has joined [10:05:08] --- matthijs has joined [10:06:11] agenda bashing [10:06:18] Marcelo is scribe [10:08:10] sound is pretty bad... [10:08:17] --- tskj has joined [10:08:39] but, I can hear Geoff [10:08:43] first up is Kurtis on Shim6 doc status [10:09:00] --- arifumi has joined [10:09:02] hba-05 in RFC editor queue [10:09:12] failure-detection-11 in AD followup [10:09:52] there are still pending DISCUSS'es [10:10:28] about firewalls, and pair probing techniques [10:11:21] the mikes are OK now [10:11:53] Jari: 3 discusses: 1 from Lars on timer details which should be resolved in latest version, 1 from Cullen on experimental vs PS, and 1 from Dave Ward on justifying why not using BFD [10:12:31] --- John Ronan has joined [10:12:55] --- internetplumber has joined [10:13:12] Jari needs to handle the last one [10:18:03] --- matthijs has joined [10:20:09] --- kquinn has joined [10:24:29] --- dthaler has joined [10:24:37] room seems to be back up now [10:24:59] now on "Current WG Drafts" slide [10:26:04] applicability, locator pair selection, and api docuiments being worked on [10:26:07] --- john.zhao has joined [10:26:11] next up: ESD [10:26:33] http://www3.ietf.org/proceedings/08mar/slides/shim6-6.pdf [10:27:29] about having identifiers that are not locators [10:28:03] slide 4 [10:29:09] would require a lookup before sending anything, rather than deferred context establishment [10:29:24] --- bnsmith has joined [10:29:30] slide 5 [10:30:04] --- John Ronan has joined [10:30:20] could use ULA space with HBA/CGA in bottom 64 bits [10:31:01] looking up locators for an unreachable ULID is non-trivial [10:31:08] (slide 6) [10:31:58] slide 7: e.g. put records in the reverse tree (ip6.arpa) in DNS [10:32:11] AAAA or SRV [10:32:43] --- dthaler has left [10:32:58] --- dthaler has joined [10:33:39] now on slide 10 [10:34:01] slide 11 [10:35:10] Kurtis: locators could be IPv4 addresses, so could also work with NAT-PT [10:35:40] if a proxy can use prefix [10:37:15] traffic engineering... no way for an end node to know what weight values to use [10:37:25] could define a DHCPv6 option [10:37:39] slide 12 [10:39:21] slide 13 [10:40:10] shim6 allows (by accident) routers to rewrite source locators in packets that have the Payload Extension header [10:40:40] --- kurtis@jabber.psg.com has joined [10:40:55] slide 14 [10:41:53] some issues if probe says A1 works and A2 fails but routers rewrite source to be A2 [10:42:19] Iljitsch: just means other side would ignore the source address [10:43:27] Erik: but rewrite is a strong hint that one should send stuff back to that address [10:44:40] with CGAs (but not HBAs) can learn new locators based on the rewriting [10:46:25] slide 15 [10:46:29] --- matthijs has joined [10:46:53] IPv4 as locator works if ULID is IPv6 CGA [10:47:58] not interesting in the host case, but interesting in the proxy case if the proxy has a public IPv4 address [10:48:41] questions? [10:49:04] Kurtis: v4 locator discussion may be relevant to v6ops discussion [10:49:16] Erik: but shim6 proxies assume shim6 is at both ends [10:50:38] um.. does everyone else still have audio? [10:50:47] no audio right now [10:50:52] er no sound in room right now [10:50:55] ah ok.. [10:50:59] they're switching speakers, laptops, ... [10:51:05] ah ok.. [10:51:05] next up is [10:51:05] http://www3.ietf.org/proceedings/08mar/slides/shim6-5.pdf [10:51:41] Marcelo, on proxy-shim6 [10:51:53] audio should be back now [10:52:17] slide 2 [10:52:18] yup [10:52:38] --- tskj has joined [11:12:12] --- XMMP has joined [11:12:21] --- XMMP has left [11:12:30] --- matthijs has joined [11:14:09] --- jpc has joined [11:14:59] --- dudi has joined [11:15:06] --- dudi has left [11:31:04] --- dthaler has joined [11:31:27] jabber seems to be going up and down [11:33:41] next up http://www3.ietf.org/proceedings/08mar/slides/shim6-3.pdf [11:34:16] --- wouter has joined [11:34:43] thanks dthaler [11:34:48] email from Alexa Morris 10 mins ago "We are aware that there is a problem with the jabber server and that it keeps crashing. We do not know why the crashes are occurring, but we are working to address this issue right now. I apologize for the disruption and any inconvenience this may be causing you." [11:35:35] slide 2 [11:36:16] slide 3 [11:36:47] slide 4 [11:37:16] --- jronan has joined [11:37:41] --- matthijs has joined [11:37:46] (presentation topic is on draft-bonaventure-informed-path-selection-00.txt and draft-saucez-idips-00.txt) [11:37:56] --- kquinn has joined [11:37:57] slide 5 [11:38:20] thanks.. jabber seems to be 'seizing' now and again. [11:38:42] primary/backup and load sharing are possible with RFC3484 but some challenges [11:39:13] --- jpc has joined [11:39:15] preferences are static not dynamic [11:40:08] (BTW, as noted above, email from Alexa Morris says "We are aware that there is a problem with the jabber server and that it keeps crashing. We do not know why the crashes are occurring, but we are working to address this issue right now. I apologize for the disruption and any inconvenience this may be causing you.") [11:40:40] now on slide 12 [11:41:02] (which says slide 8 in the bottom right corner, but is 12/19 in the PDF) [11:42:01] suggest supporting TE by having a service to identify path weights and change locators even if there's no failure [11:43:36] slide 13 (labelled 9) [11:44:13] Dow Street: would the service check paths among multiple ISPs (answer: yes) [11:44:18] --- kquinn has left [11:44:48] Tim Sheppard: how do you discover the service (answer: anycast) [11:45:35] Tim Sheppard: so if you're a host multihomed to two networks with different services? [11:47:03] (I didn't hear any real answer to Tim's question...) [11:47:12] agreed [11:47:29] other than the comment "that's a hard question" :) [11:47:46] thanks for doing the scribe btw much appreciated [11:48:05] yw [11:48:08] --- m_ersue has joined [11:48:19] --- m_ersue has left [11:48:32] Erik Nordmark: if both ends do this, the two might get different lists, what happens? [11:48:51] (again no answer...) [11:49:16] next up Iljistch on future work... [11:49:40] http://www3.ietf.org/proceedings/08mar/slides/shim6-2.pdf [11:49:48] Tim Sheppard: anyone have running code? [11:50:03] for shim6? [11:50:08] Geoff Huston: yes, there are 4 implementations listed on web site [11:50:13] (right for shim6) [11:50:13] yeas UCL has running code. [11:50:33] --- john.zhao has joined [11:50:46] http://inl.info.ucl.ac.be/LinShim6 [11:50:47] now on slide 2 of Iljitsch's slides [11:50:58] slide 3 [11:51:02] partially done I guess [11:51:15] slide 4: ingress filtering [11:51:26] who's speaking.. can't hear him [11:51:31] Iljitsch is presenting [11:51:35] iljitsch [11:51:53] can you hear him? [11:52:01] very very very low [11:52:05] very low volume [11:52:07] volume is up to the last here [11:52:16] slide 5 [11:52:44] --- juampe has joined [11:52:53] --- juampe has left [11:52:55] Jari: 6man is addressing ingress filtering [11:53:47] Iljitsch is fixing mike, can you hear now? [11:53:52] yes [11:53:57] yes.. [11:54:05] now on slide 6 [11:54:11] ("initial failures") [11:54:52] slide 7, skipping... [11:55:01] slide 8... [11:55:09] slide 9... rewriting in routers [11:55:17] --- dvijay has joined [11:55:53] if app didn't specify source address, local app doesn't care about rewriting [11:56:04] so can be changed as long as nothing has looked at it [11:56:22] (sounds like schroedingers cat...) [11:56:34] slide 10: NAT [11:57:04] 1-to-1 nat is bad but less bad than regular nat [11:57:15] slide 11: NAT (2) [11:57:39] can solve checksum issue, but not ipsec, referrals, path changes, ... [11:58:20] Tim Sheppard: even ESP doesn't work, right? [11:58:43] probably right [11:59:09] slide 12: NAT (3) [12:00:25] Erik Nordmark, on slide 11: IKE itself isn't that hard since it works with NAT-T (RFC 3947) [12:01:24] Erik: some referrals work as long as the referrals don't cross the NAT boundary... in-to-in and out-to-out work, just not the ones that cross the boundary [12:01:50] Iljitsch: could ask someone outside what outside address is to use in a referral [12:02:07] Christian Vogt: referrals in app payloads still break, esp. if encrypted [12:03:43] --- dvijay has left [12:04:21] any opinions on what new work to take on? [12:05:09] Erik: maybe we could make proxies be useful using DNS based prefixes in reverse tree to get a set of locator prefixes [12:07:11] --- end of meeting --- [12:07:41] --- matthijs has left [12:07:45] Thanking you. [12:08:01] --- jronan has left [12:08:03] --- john.zhao has left [12:08:10] --- wouter has left [12:09:15] --- dthaler has left [12:18:40] --- Inca has joined [12:19:00] --- juampe has joined [12:21:00] --- jpc has left [12:21:17] --- juampe has left [13:09:50] --- becarpenter has joined [13:11:25] --- becarpenter has left [15:31:08] --- badra has joined [15:31:12] --- badra has left [15:31:21] --- badra has joined [15:31:23] --- badra has left [15:34:26] --- Inca has joined [20:59:36] --- Inca has left [21:07:40] --- Inca has joined