IETF
QUIC
quic@jabber.ietf.org
Thursday, September 20, 2018< ^ >
janaiyengar has set the subject to: Really Draining
Room Configuration
Room Occupants

GMT+0
[00:08:15] Sean Turner joins the room
[00:13:24] ekr joins the room
[00:31:59] ekr leaves the room
[00:55:59] Sean Turner leaves the room
[01:36:32] Eric Kinnear joins the room
[01:58:39] Eric Kinnear leaves the room: Stream reset by peer
[03:08:27] mnot joins the room
[03:08:27] mnot leaves the room: Disconnected: closed
[04:40:31] janaiyengar joins the room
[05:04:05] janaiyengar leaves the room: Stream closed by us: Replaced by new connection (conflict)
[05:04:06] janaiyengar joins the room
[05:09:35] janaiyengar leaves the room: Stream reset by peer
[05:25:35] Jari Arkko joins the room
[06:16:17] Jari Arkko leaves the room
[06:17:38] Jari Arkko joins the room
[06:18:14] Jari Arkko leaves the room
[06:18:15] Jari Arkko joins the room
[06:18:50] Jari Arkko leaves the room
[07:11:25] ekr joins the room
[07:44:30] ekr leaves the room
[10:08:16] Martin Thomson joins the room
[10:15:56] Martin Thomson leaves the room
[10:53:56] Lars Eggert joins the room
[11:13:50] masaori@xmpp.jp joins the room
[12:23:41] Lars Eggert leaves the room
[12:38:19] Lars Eggert joins the room
[12:56:33] Mirja Kuehlewind joins the room
[13:01:34] Martin Thomson joins the room
[13:01:36] masaori@xmpp.jp leaves the room: Connection failed: connection closed
[13:02:52] masaori@xmpp.jp joins the room
[13:13:22] masaori@xmpp.jp leaves the room: Stream reset by peer
[13:18:11] chi.jiun.su joins the room
[13:18:20] masaori@xmpp.jp joins the room
[13:25:10] Eric Kinnear joins the room
[13:27:04] ted.h joins the room
[13:28:09] Igor Lubashev joins the room
[13:28:37] brian@trammell.ch joins the room
[13:31:13] <ted.h> No sound from the room, though we are getting Lars' visualization.
[13:33:04] Sean Turner joins the room
[13:33:31] afrind joins the room
[13:34:18] janaiyengar joins the room
[13:36:36] afrind leaves the room: Stream closed by us: Replaced by new connection (conflict)
[13:36:37] afrind joins the room
[13:37:41] <brian@trammell.ch> minutes  = https://docs.google.com/document/d/1qdZu4mksN8yOoIZPUKXC5GyzuDyEt3DWgjx-ZZOm8gM/edit
[13:37:50] <Sean Turner> there it is
[13:37:59] mnot joins the room
[13:38:12] Christian joins the room
[13:38:29] subodh joins the room
[13:39:07] <Mirja Kuehlewind> there is a lot of background noise today
[13:39:15] <Mirja Kuehlewind> is someone typing next to the mic?
[13:39:23] ekr joins the room
[13:47:35] <Mirja Kuehlewind> ACK loss is critical if that could lead to concealment of congestion indication
[13:48:16] <Mirja Kuehlewind> there should always be at least reliably an indication that there was some congestion
[13:50:26] <Mirja Kuehlewind> if you send an ack for each X packets, it will always be the same size
[13:51:04] <ted.h> Are folks seeing the Webex Chat?
[13:51:11] <Mirja Kuehlewind> no idea…
[13:51:13] <ekr> No
[13:51:14] <ted.h> UoA Video has a question there
[13:51:20] <mnot> We don’t use webbed chat
[13:51:22] <ted.h> Pasted:
[13:51:24] <ted.h> If the bit vector /replaces/ the counters, do we still know the culumulative totals in a way that is tollerant to loss?
[13:51:25] <mnot> Tell them to use the jabber?
[13:52:05] <janaiyengar> The bit vector is exact in that it says exactly what mark was on what packet.
[13:52:35] <janaiyengar> this bit vector can be repeated as often as you want in subsequent acks, and truncated when a size limit is exceeded
[13:53:12] <Mirja Kuehlewind> but with the vector is it possible to loose information when truncated
[13:53:34] <brian@trammell.ch> or when the ack is lost
[13:53:39] <ted.h> @mnot They are moving to jabber.
[13:54:09] <janaiyengar> @mirja: yes, truncation can cause info loss
[13:57:34] gorryf joins the room
[13:58:27] <Mirja Kuehlewind> Can we maybe add a field to the regular ACK frame in case all ECN marks are the same or otherwise indicating that there is an ECN frame coming on…?
[13:58:58] Martin Thomson leaves the room
[13:59:10] <brian@trammell.ch> mic?
[13:59:34] <Mirja Kuehlewind> no sure, just another idea...
[14:00:32] <gorryf> Gorry-> Mirja - is this to provide simple feedback when there is no marking active?
[14:01:02] <brian@trammell.ch> i think that's a way to implement the make-ecn-not-look-exotic point. maybe add that as a comment to the issue?
[14:02:41] <gorryf> I really suggest we don’t try to eliminate the ECT(0) or ECT(1) count, we really should know whether marks were changed.
[14:03:06] <gorryf> It’s only a counter... and the path can change, we should know.
[14:03:11] <Mirja Kuehlewind> gorry: if there is no marking at all or a long period with no congestion
[14:03:12] <brian@trammell.ch> mt is talking about optimizing by counding unmarked, as opposed to marked
[14:03:54] <gorryf> Mirja: OK, but a cumulative count is also a constant?
[14:04:19] dragana joins the room
[14:04:25] <Mirja Kuehlewind> yes but more bits
[14:04:36] <gorryf> a variant?
[14:04:46] <gorryf> a /var/int/
[14:05:05] <Mirja Kuehlewind> and not a separate frame to add
[14:08:22] <gorryf> Gorry: I really think we need to be able to detect the presence of BOTH marks - but we do not need accurate counting of both.
[14:08:26] Christian leaves the room
[14:09:45] <gorryf> The reason is: - There is an existing RFC that allows alternate marking when you set a DSCP, and that creates some interaction we can not legislate against if that was deployed.
[14:10:56] <gorryf> Saving the varint in the record seems an unreasonable optimisation.
[14:12:06] Martin Thomson joins the room
[14:12:48] <janaiyengar> @gorryf: fair point. We're retaining both counts.
[14:13:00] <gorryf> And that is good with me:-)
[14:14:25] janaiyengar leaves the room: Stream reset by peer
[14:15:40] Christian joins the room
[14:16:59] ekr leaves the room
[14:20:56] ekr joins the room
[14:21:03] Sean Turner leaves the room: Replaced by new connection
[14:21:07] Sean Turner joins the room
[14:21:16] <brian@trammell.ch> (break is over)
[14:21:39] Sean Turner leaves the room
[14:22:37] subodh leaves the room: Stream closed by us: Replaced by new connection (conflict)
[14:22:39] subodh joins the room
[14:23:06] janaiyengar joins the room
[14:24:16] Sean Turner joins the room
[14:31:47] Mirja Kuehlewind leaves the room
[14:33:53] Mirja Kuehlewind joins the room
[14:35:11] Mirja Kuehlewind leaves the room
[14:44:48] Martin Thomson leaves the room
[14:45:39] subodh leaves the room: Stream reset by peer
[14:45:45] subodh joins the room
[14:46:24] g.e.montenegro joins the room
[14:52:07] Martin Thomson joins the room
[14:58:13] <ted.h> The siren is louder than Jana
[14:58:27] <mnot> That is true.
[14:58:42] Christian has set the subject to: Siren Retry
[15:01:02] <spencerdawkins> Could Brian repeat where he says "you have to do it this way" should go? I missed it both times he said it ...
[15:03:26] <Sean Turner> somebody needs to mute :)
[15:03:46] <brian@trammell.ch> ah sorry, missed this
[15:23:00] <ted.h> Couldn't hear Mike at all, fyi.
[15:29:19] ekr leaves the room
[15:29:26] Igor Lubashev leaves the room: Disconnected: closed
[15:29:26] Martin Thomson leaves the room
[15:29:27] Eric Kinnear leaves the room: Stream reset by peer
[15:29:29] brian@trammell.ch leaves the room
[15:29:37] janaiyengar leaves the room: Stream reset by peer
[15:29:44] Lars Eggert leaves the room
[15:30:56] ekr joins the room
[15:31:10] ted.h leaves the room
[15:31:17] ekr leaves the room
[15:33:59] ekr joins the room
[15:34:59] janaiyengar joins the room
[15:39:22] janaiyengar leaves the room: Stream reset by peer
[15:40:17] ekr leaves the room
[15:45:26] ekr joins the room
[15:45:35] subodh leaves the room: Connection failed: connection closed
[15:47:06] gorryf leaves the room
[15:49:14] Christian leaves the room
[15:49:37] Sean Turner leaves the room: Replaced by new connection
[15:49:40] Sean Turner joins the room
[15:49:50] Sean Turner leaves the room
[15:50:05] Sean Turner joins the room
[16:12:15] Sean Turner leaves the room
[16:15:21] Sean Turner joins the room
[16:19:56] Sean Turner leaves the room
[16:20:18] Sean Turner joins the room
[16:20:48] Sean Turner leaves the room
[16:21:06] Sean Turner joins the room
[16:37:24] Sean Turner leaves the room
[16:42:43] subodh joins the room
[16:43:37] ekr leaves the room
[16:48:25] Sean Turner joins the room
[16:48:38] Sean Turner leaves the room
[16:48:56] Sean Turner joins the room
[16:49:10] Sean Turner leaves the room
[16:56:03] Sean Turner joins the room
[16:56:13] Sean Turner leaves the room
[16:57:20] Sean Turner joins the room
[16:57:49] Sean Turner leaves the room
[17:00:31] Sean Turner joins the room
[17:01:09] ekr joins the room
[17:03:56] Eric Kinnear joins the room
[17:04:55] <mnot> https://docs.google.com/document/d/1qdZu4mksN8yOoIZPUKXC5GyzuDyEt3DWgjx-ZZOm8gM/edit
[17:07:20] brian@trammell.ch joins the room
[17:13:19] subodh leaves the room: Stream reset by peer
[17:13:33] subodh joins the room
[17:15:06] Martin Thomson joins the room
[17:16:52] janaiyengar joins the room
[17:19:12] Christian joins the room
[17:20:06] Sean Turner leaves the room: Replaced by new connection
[17:20:08] Sean Turner joins the room
[17:20:37] Sean Turner leaves the room
[17:21:40] Sean Turner joins the room
[17:25:18] gorryf joins the room
[17:30:22] <gorryf> So, I think we are discussing if QUIC will be 20x more aggressive than a TCP session under severe congestion overload?
[17:38:18] ted.h joins the room
[17:39:36] <brian@trammell.ch> no
[17:40:19] <brian@trammell.ch> we're talking about a cwin replacement for use until path validation is complete during the handshake for amplification protection
[17:54:36] <ted.h> Are you getting a fire alarm test in the room?
[17:56:46] <mnot> apparently.
[17:57:07] <brian@trammell.ch> *please disregard the bytes versus packets disussion*
[18:03:06] <spencerdawkins> +q
[18:07:09] <mnot> Our network:
[18:07:10] <mnot> http://n2.netalyzr.icsi.berkeley.edu/summary/id=36a470be-9443-00dd5981-ee2e-4911-a9df
[18:07:22] <mnot> Note the proxy.
[18:15:35] brian@trammell.ch leaves the room: Disconnected: closed
[18:21:26] brian@trammell.ch joins the room
[18:35:01] ekr leaves the room
[18:39:20] ekr joins the room
[18:43:02] Lars Eggert joins the room
[18:47:50] ekr leaves the room
[18:47:51] brian@trammell.ch leaves the room: Disconnected: closed
[18:47:51] janaiyengar leaves the room: Stream reset by peer
[18:47:57] masaori@xmpp.jp leaves the room
[18:48:05] Martin Thomson leaves the room
[18:55:28] Christian leaves the room
[19:00:53] Martin Thomson joins the room
[19:00:53] brian@trammell.ch joins the room
[19:01:50] ekr joins the room
[19:03:04] janaiyengar joins the room
[19:10:40] <ted.h> That's funny
[19:26:49] <spencerdawkins> I look forward ro the discussions about next steps, noting that another AD will be cashing any check I sign ...
[19:26:51] Christian joins the room
[19:30:56] <ted.h> q+
[19:38:32] <spencerdawkins> +1 Jana on experimental versions
[19:38:36] <ted.h> The use cases folks have in mind might turn out to be a very small subset of partial reliability.  If we're fine with that being an extension or version from the experimental space, without providing full consideration of partial reliability, then we can communicate that and go on.
[19:39:05] afrind leaves the room: Stream reset by peer
[19:39:15] <ted.h> If we want a full version of partial reliability that includes their uses cases and energy, though, pushing this off seems likely to slow us down more than the "curing" phase alone would.
[19:39:20] <spencerdawkins> @ted.h, thanks for the background.
[19:40:25] <ted.h> @ekr  Not all the use cases I am hearing about are RTCWEB related.  Unfortunately, I cannot say much more than that.
[19:41:51] <ted.h> Roni's question is a usefully abstract version of the point I was raising.
[19:42:58] <spencerdawkins> +1 Martin Thomson
[19:44:23] ekr leaves the room
[19:45:41] Eric Kinnear leaves the room: Stream reset by peer
[19:46:01] Eric Kinnear joins the room
[19:50:00] ekr joins the room
[20:16:43] brian@trammell.ch leaves the room
[20:16:52] Sean Turner leaves the room
[20:17:06] ekr leaves the room
[20:18:08] Lars Eggert leaves the room
[20:18:12] Eric Kinnear leaves the room: Stream reset by peer
[20:18:49] chi.jiun.su leaves the room: offline
[20:18:56] mnot leaves the room
[20:18:57] Christian leaves the room
[20:19:00] ted.h leaves the room
[20:19:03] g.e.montenegro leaves the room
[20:19:41] janaiyengar leaves the room: Stream reset by peer
[20:27:50] Martin Thomson leaves the room
[20:35:37] subodh leaves the room: Connection failed: connection closed
[20:46:49] mnot joins the room
[20:47:47] Eric Kinnear joins the room
[20:47:54] mnot leaves the room
[21:01:32] gorryf leaves the room
[21:42:18] subodh joins the room
[21:47:04] subodh leaves the room
[22:00:11] Eric Kinnear leaves the room: Stream reset by peer
[22:20:15] Eric Kinnear joins the room
[22:32:29] afrind joins the room
[22:42:43] Eric Kinnear leaves the room: Connection failed: connection timed out
[22:50:23] ekr joins the room
[23:05:37] afrind leaves the room: Stream reset by peer
[23:06:19] Martin Thomson joins the room
[23:06:42] Martin Thomson leaves the room: Stream closed by us: Replaced by new connection (conflict)
[23:06:43] Martin Thomson joins the room
[23:11:30] ekr leaves the room
[23:22:04] dragana leaves the room: Disconnected: closed
[23:32:56] Martin Thomson leaves the room
[23:56:53] ekr joins the room
Powered by ejabberd - robust, scalable and extensible XMPP server Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!