#jython IRC Log

Index

IRC Log for 2016-08-04

Timestamps are in GMT/BST.

[1:03] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:a506:6d83:952f:2ffc) Quit (Remote host closed the connection)
[1:25] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[2:26] * eatkin (~eatkin@166.70.212.121) Quit (Quit: ZNC 1.6.2 - http://znc.in)
[2:27] * eatkin (~eatkin@166.70.212.121) has joined #jython
[2:27] * eatkin (~eatkin@166.70.212.121) Quit (Remote host closed the connection)
[2:28] * eatkin (~eatkin@166.70.212.121) has joined #jython
[2:28] * eatkin (~eatkin@166.70.212.121) Quit (Remote host closed the connection)
[2:29] * eatkin (~eatkin@166.70.212.121) has joined #jython
[2:50] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Remote host closed the connection)
[3:22] * nickmbailey (~nickmbail@cpe-70-117-83-204.austin.res.rr.com) has joined #jython
[3:27] * nickmbailey (~nickmbail@cpe-70-117-83-204.austin.res.rr.com) Quit (Ping timeout: 276 seconds)
[4:23] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[4:27] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[4:42] * Einherjer (~einherjer@69.64.40.177) has joined #jython
[5:24] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[5:30] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[7:11] * fwierzbicki1 (~Adium@99-106-169-5.lightspeed.sntcca.sbcglobal.net) has joined #jython
[7:12] * fwierzbicki (~Adium@99-106-169-5.lightspeed.sntcca.sbcglobal.net) Quit (Ping timeout: 244 seconds)
[7:27] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[7:31] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[8:27] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[8:32] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[9:28] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[9:32] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[10:29] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[10:33] * jimbaker (~jbaker@python/psf/jimbaker) Quit (Ping timeout: 250 seconds)
[10:33] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[10:37] * jimbaker (~jbaker@8.44.156.98) has joined #jython
[10:37] * jimbaker (~jbaker@8.44.156.98) Quit (Changing host)
[10:37] * jimbaker (~jbaker@python/psf/jimbaker) has joined #jython
[10:37] * ChanServ sets mode +o jimbaker
[11:30] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[11:34] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[12:10] * TomA (~TomA@c-68-32-46-223.hsd1.mi.comcast.net) has joined #jython
[12:30] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[12:35] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[13:31] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[13:35] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Ping timeout: 250 seconds)
[13:39] * maucar (~maurizioc@c-24-62-42-67.hsd1.ma.comcast.net) has joined #jython
[14:20] * xemdetia (xemdetia@nat/ibm/x-ywracefzxxftdxfh) has joined #jython
[14:22] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) has joined #jython
[14:59] * pizzaman (~pm@mary008-0301-dhcp-170.bu.edu) has joined #jython
[16:22] <nickmbailey> jimbaker: so we are still seeing some potential issues even with the fix on that ticket
[16:22] <nickmbailey> specifically right now we're only seeing a problem on centos boxes
[16:23] <nickmbailey> i wonder if maybe bytesBeforeUnwritable() isn't accurate in certain situations
[16:57] * TomA (~TomA@c-68-32-46-223.hsd1.mi.comcast.net) Quit (Remote host closed the connection)
[16:58] * TomA (~TomA@c-68-32-46-223.hsd1.mi.comcast.net) has joined #jython
[17:11] <nickmbailey> actually it isn't just centos
[17:11] <nickmbailey> somehow it still seems like things are always guaranteed to be flushed
[17:41] * Einherjer (~einherjer@69.64.40.177) Quit (K-Lined)
[17:58] * pizzaman (~pm@mary008-0301-dhcp-170.bu.edu) Quit (Ping timeout: 258 seconds)
[18:16] * pizzaman (~pm@mary008-0301-dhcp-153.bu.edu) has joined #jython
[18:34] <jimbaker> nickmbailey, not following the specifics here
[18:37] * AndyBotwin (~Gustavo@unaffiliated/andybotwin) has joined #jython
[18:37] * AndyBotwin (~Gustavo@unaffiliated/andybotwin) Quit (Max SendQ exceeded)
[19:21] <nickmbailey> jimbaker: we're still seeing the same issue we were seeing before. just somewhat less frequently
[19:22] <nickmbailey> the issue being that the server closes it's connection too early before the client has read everything
[19:54] * nickmbailey (~nickmbail@2605:6000:e8ce:3100:8488:90cb:3b79:c007) Quit (Remote host closed the connection)
[19:58] * nickmbailey (~nickmbail@cpe-70-117-83-204.austin.res.rr.com) has joined #jython
[20:08] <topi`> nickmbailey: maybe something about the TCP protocol semantics?
[20:08] <topi`> but, maybe not, the server can close the socket immediately after the socket has been established
[20:09] <topi`> then you'd see a FIN_WAIT state with netstat, for a while
[20:12] * maucar (~maurizioc@c-24-62-42-67.hsd1.ma.comcast.net) Quit (Quit: Leaving)
[20:36] * TomA (~TomA@c-68-32-46-223.hsd1.mi.comcast.net) Quit (Ping timeout: 265 seconds)
[20:36] <nickmbailey> well the problem is in netty itself i think. we're closing the netty channel before netty has flushed everything to the underlying socket
[20:36] * TomA (~TomA@c-68-32-46-223.hsd1.mi.comcast.net) has joined #jython
[20:37] <nickmbailey> if i limit the buffer of any call to send() to 8192 bytes things seem to work...
[21:18] <jimbaker> nickmbailey, well that's a possible workaround
[21:19] <jimbaker> and compatible with send semantics guarantees
[21:31] <nickmbailey> yeah i'm slightly worried about performance implications
[21:31] <nickmbailey> but maybe that would work for a short term fix
[21:33] <jimbaker> nickmbailey, in my experience it's sending 2 bytes at a time that tends to be a problem, not 8192 bytes :)
[21:33] <nickmbailey> haha
[21:35] * TomA (~TomA@c-68-32-46-223.hsd1.mi.comcast.net) Quit (Remote host closed the connection)
[22:00] <nickmbailey> well this is the workaround that seems to be doing slightly better https://github.com/riptano/jython/pull/8/commits/b914e4d47169a99c59393e85490c7d850e879e07
[22:03] <nickmbailey> i came up with the 8192 number because that's how much data our client's we're trying to read and failing
[22:03] <nickmbailey> and also it's a default at least a few places in the netty codebase
[22:04] <nickmbailey> so that's about a half step past "pulled it out of a magic hat"
[22:13] <jimbaker> nickmbailey, yeah, it's a very reasonable number
[22:13] <jimbaker> and an easy workaround if it works out well in your testing
[22:13] <nickmbailey> yeah its actually the default page size in an internal buffer allocator i guess? http://netty.io/4.1/xref/io/netty/buffer/PooledByteBufAllocator.html
[22:14] <nickmbailey> so perhaps there is some sort of race with flushing whole pages of buffers or something
[23:00] * pizzaman (~pm@mary008-0301-dhcp-153.bu.edu) Quit (Ping timeout: 252 seconds)
[23:38] * xemdetia (xemdetia@nat/ibm/x-ywracefzxxftdxfh) Quit (Ping timeout: 244 seconds)

Index

These logs were automatically created by JythonLogBot on irc.freenode.net using a slightly modified version of the Java IRC LogBot.