Displaying #traffic-server/2015-12-10.log:

Thu Dec 10 00:07:18 2015  _klk_:Joined the channel
Thu Dec 10 00:12:35 2015  gancho:Joined the channel
Thu Dec 10 00:15:09 2015  masaori:Joined the channel
Thu Dec 10 00:15:33 2015  danielxu:Joined the channel
Thu Dec 10 00:15:38 2015  masaori:Joined the channel
Thu Dec 10 00:23:48 2015  masaori:Joined the channel
Thu Dec 10 00:29:28 2015  gancho:Joined the channel
Thu Dec 10 00:31:31 2015  masaori:Joined the channel
Thu Dec 10 00:52:17 2015  zwoop:masaori masaori rokubo It's actually failing the regressions in HPack as well
Thu Dec 10 00:52:18 2015  zwoop:RPRINT HPACK_Encode: encoded value was invalid
Thu Dec 10 00:52:19 2015  zwoop: REGRESSION_RESULT HPACK_Encode: FAILED
Thu Dec 10 00:52:46 2015  masaori:zwoop: that is TS-4063
Thu Dec 10 00:53:04 2015  masaori:and maskit is going to send a patch
Thu Dec 10 00:53:11 2015  zwoop:ok, cool
Thu Dec 10 00:53:32 2015  zwoop:if this is going to take a while, we might want to either turn off that test for now, or put something temporary in there
Thu Dec 10 00:53:40 2015  zwoop:since it's failing all the CI, we are sort of flying blind right now
Thu Dec 10 00:54:16 2015  masaori:it will be fixed in today
Thu Dec 10 01:18:55 2015  masaori:oh, I mistook little bit. "git pull https://github.com/maskit/trafficserver ts4063" made merge commit, and I amended only it.
Thu Dec 10 01:21:08 2015  masaori:zwoop: using "--force" is allowed?
Thu Dec 10 01:21:54 2015  masaori:I mean "force push"
Thu Dec 10 01:24:04 2015  _klk_:Joined the channel
Thu Dec 10 01:41:44 2015  _klk_:Joined the channel
Thu Dec 10 02:58:39 2015  briang:bcall, that init_cwnd stuff is broken for spdy / http2
Thu Dec 10 02:58:44 2015  briang:it's only set for HTTPClientSessions
Thu Dec 10 02:58:50 2015  briang:not SpdyClientSessions or H2 client sessions
Thu Dec 10 03:12:25 2015  danielxu:Joined the channel
Thu Dec 10 03:29:22 2015  gancho:Joined the channel
Thu Dec 10 03:58:47 2015  ibrezac:Joined the channel
Thu Dec 10 04:21:20 2015  biilmann:Joined the channel
Thu Dec 10 04:51:15 2015  biilmann:Joined the channel
Thu Dec 10 05:28:18 2015  biilmann:Joined the channel
Thu Dec 10 06:02:47 2015  _klk_:Joined the channel
Thu Dec 10 06:04:14 2015  danielxu:Joined the channel
Thu Dec 10 07:24:46 2015  biilmann:Joined the channel
Thu Dec 10 07:47:01 2015  slx86:Joined the channel
Thu Dec 10 08:01:03 2015  psp:Joined the channel
Thu Dec 10 08:02:47 2015  psp1:Joined the channel
Thu Dec 10 08:05:18 2015  ke4qqq:Joined the channel
Thu Dec 10 08:12:26 2015  masaori:Joined the channel
Thu Dec 10 08:15:49 2015  danielxu:Joined the channel
Thu Dec 10 08:22:55 2015  mkitajou:Joined the channel
Thu Dec 10 08:41:13 2015  danielxu:Joined the channel
Thu Dec 10 09:17:52 2015  Lethalman:Joined the channel
Thu Dec 10 10:35:26 2015  mturk:Joined the channel
Thu Dec 10 10:35:26 2015  mturk:Joined the channel
Thu Dec 10 11:36:44 2015  mturk:Joined the channel
Thu Dec 10 11:36:45 2015  mturk:Joined the channel
Thu Dec 10 12:52:25 2015  JSeymour:Joined the channel
Thu Dec 10 14:11:51 2015  reveller1:Joined the channel
Thu Dec 10 14:21:27 2015  ibrezac:Joined the channel
Thu Dec 10 14:27:48 2015  esproul:Joined the channel
Thu Dec 10 14:28:18 2015  swoc:Joined the channel
Thu Dec 10 14:31:57 2015  yamc:Joined the channel
Thu Dec 10 14:44:37 2015  bahumbug:Joined the channel
Thu Dec 10 14:44:46 2015  niq:Joined the channel
Thu Dec 10 14:52:52 2015  mturk:Joined the channel
Thu Dec 10 14:52:52 2015  mturk:Joined the channel
Thu Dec 10 16:00:37 2015  jumby:Joined the channel
Thu Dec 10 16:20:24 2015  igalic:i just realized that the most solid code we have in trafficserver is completely redundant in a modern system: traffic_cop
Thu Dec 10 16:28:28 2015  gancho:Joined the channel
Thu Dec 10 16:32:56 2015  PSUdaemon:totally redundant
Thu Dec 10 16:33:01 2015  PSUdaemon:but i dunno if i would call it solid :)
Thu Dec 10 16:33:13 2015  PSUdaemon:also, probably just mostly redundant
Thu Dec 10 16:33:15 2015  PSUdaemon:not totally
Thu Dec 10 16:33:23 2015  gancho:Joined the channel
Thu Dec 10 17:10:23 2015  biilmann:Joined the channel
Thu Dec 10 17:10:55 2015  bcall:briang: I am not sure, but I would guess init_cwnd wouldn't work with spdy and http/2
Thu Dec 10 17:11:22 2015  bahumbug:Joined the channel
Thu Dec 10 17:11:31 2015  niq:Joined the channel
Thu Dec 10 17:13:40 2015  jpeach:isn't it that init_cwnd isn't as important for http/2 because the session is longer?
Thu Dec 10 17:18:28 2015  shinrich1:Joined the channel
Thu Dec 10 17:19:43 2015  blattj:Joined the channel
Thu Dec 10 17:27:41 2015  davet:Joined the channel
Thu Dec 10 17:35:41 2015  blattj:Joined the channel
Thu Dec 10 17:36:18 2015  biilmann:Joined the channel
Thu Dec 10 18:06:59 2015  gancho:Joined the channel
Thu Dec 10 18:15:43 2015  zwoop:jpeach probably only if you also disable net.ipv4.tcp_slow_start_after_idle
Thu Dec 10 18:15:58 2015  yamc:Joined the channel
Thu Dec 10 18:16:46 2015  jpeach:yeh good point, though is init_cwnd applied after idle? or just on new stuff?
Thu Dec 10 18:17:39 2015  zwoop:I'd *assume* it's the starting point, so after idle as well, but not 100% sure
Thu Dec 10 18:17:50 2015  zwoop:seems crazy if it'd allow it to go down to < init_cwnd no ?
Thu Dec 10 18:18:26 2015  blattj:Joined the channel
Thu Dec 10 18:24:35 2015  es:Joined the channel
Thu Dec 10 18:27:36 2015  es:Joined the channel
Thu Dec 10 18:40:46 2015  shinrich1:Joined the channel
Thu Dec 10 18:42:38 2015  biilmann:Joined the channel
Thu Dec 10 19:00:21 2015  gancho:Joined the channel
Thu Dec 10 19:11:59 2015  gancho:Joined the channel
Thu Dec 10 19:12:01 2015  reveller2:Joined the channel
Thu Dec 10 19:16:18 2015  PSUdaemon:init_cwnd happens on idle as well, but you can change that
Thu Dec 10 19:17:06 2015  PSUdaemon:with the setting that zwoop pointed out
Thu Dec 10 19:24:30 2015  zwoop:PSUdaemon right, that's the point of the setting. I thought jpeach asked if it'd go down to init_cwnd on idle, or to something less (like 3 or 4) ?
Thu Dec 10 19:25:06 2015  shinrich1:Joined the channel
Thu Dec 10 19:27:18 2015  blattj:Joined the channel
Thu Dec 10 19:27:43 2015  jpeach:ah that seems sensible
Thu Dec 10 19:28:20 2015  _klk_:Joined the channel
Thu Dec 10 19:29:12 2015  PSUdaemon:back to initial
Thu Dec 10 19:31:44 2015  jpeach:zwoop: do you know whether it is safe to install openssl and libressl side-by side?
Thu Dec 10 19:59:17 2015  shinrich1:Joined the channel
Thu Dec 10 20:03:25 2015  _klk_:Joined the channel
Thu Dec 10 20:06:19 2015  shinrich1:Joined the channel
Thu Dec 10 20:16:12 2015  bahumbug:Joined the channel
Thu Dec 10 20:16:20 2015  niq:Joined the channel
Thu Dec 10 20:43:06 2015  biilmann:Joined the channel
Thu Dec 10 20:44:22 2015  gancho:Joined the channel
Thu Dec 10 20:45:36 2015  shinrich1:Joined the channel
Thu Dec 10 21:04:08 2015  _klk_:Joined the channel
Thu Dec 10 21:13:24 2015  biilmann:Joined the channel
Thu Dec 10 21:32:18 2015  davet:Joined the channel
Thu Dec 10 21:57:30 2015  biilmann:Joined the channel
Thu Dec 10 22:04:08 2015  shinrich1:Joined the channel
Thu Dec 10 22:04:25 2015  davet:Joined the channel
Thu Dec 10 22:07:48 2015  yamc:Joined the channel
Thu Dec 10 22:14:16 2015  jacksontj:Joined the channel
Thu Dec 10 22:22:48 2015  _klk_:Joined the channel
Thu Dec 10 22:26:47 2015  biilmann:Joined the channel
Thu Dec 10 22:36:08 2015  yamc:Joined the channel
Thu Dec 10 22:38:31 2015  swoc:Joined the channel
Thu Dec 10 22:58:53 2015  _klk_:Joined the channel
Thu Dec 10 23:03:49 2015  shinrich1:Joined the channel
Thu Dec 10 23:07:54 2015  shinrich2:Joined the channel
Thu Dec 10 23:38:10 2015  blattj:Joined the channel
Thu Dec 10 23:46:44 2015  _klk_:Joined the channel
Thu Dec 10 23:53:09 2015  _klk_:Joined the channel

Comments