Differences

This shows you the differences between two versions of the page.

Link to this comparison view

irc:1441317600 [2017/05/27 13:44]
irc:1441317600 [2017/05/27 13:44] (current)
Line 1: Line 1:
 +[00:20:53] <​AlexLehm>​ unless its documented as async, it's probably blocking
 +
 +[00:22:26] <​s0va>​ BadApe: if it does not return future as result of method call, or method call doesn'​t take some sort of handler/​callback as an argument, it's blocking
 +
 +[00:22:29] <​s0va>​ :)
 +
 +[00:24:50] <​BadApe>​ i had a feeling that was the case
 +
 +[00:26:32] <​s0va>​ well, you can still use blocking libraries in worker-style verticles, or you can use vertx.executeBlocking() in eventloop
 +
 +[00:27:11] <​s0va>​ you can also execute blocking code in your own custom executor (wrapped in callables), so it's okay
 +
 +[00:27:14] <​s0va>​ :)
 +
 +[02:56:02] *** ChanServ sets mode: +o temporalfox
 +
 +[08:58:03] *** ChanServ sets mode: +o purplefox
 +
 +[09:16:06] <​marcels>​ Hi, we are seeing some random NPE's during deploy of modules since 2.1.6. Restarting the same module solves the issue. Any ideas on what could cause this ?
 +
 +[09:17:53] <​marcels>​ https://​gist.github.com/​msoute/​7c88d7a34f06e936ff0b
 +
 +[09:17:59] <​marcels>​ and a gist with stack :-)
 +
 +[10:01:12] <​purplefox>​ cescoffier: pmlopes temporalfox , coming to the meeting now
 +
 +[10:01:18] <​temporalfox>​ already there!
 +
 +[11:28:10] <​guest1_>​ i involved hbase-client and vertx 3.0.0 in the same project, both libs use differect version netty, so i thought netty conflict occurs.
 +
 +[11:28:39] <​guest1_>​ what can i do to solve this problem?
 +
 +[11:36:18] <​pmlopes>​ guest1_: that is tricky which version does hbase use?
 +
 +[11:40:37] <​guest1_>​ vertx3.0.0 -> io.netty:​netty-common:​4.0.28.Final, ​ hbase-client:​1.1.0.1->​io.netty:​netty-all:​4.0.23.Final
 +
 +[11:42:08] <​guest1_>​ so i got this error "​java.lang.IncompatibleClassChangeError:​ Class io.netty.handler.codec.http.DefaultCookie does not implement the requested interface io.netty.handler.codec.http.cookie.Cookie"​
 +
 +[11:47:53] <​s0va>​ guest1_: i force netty-all 4.0.30 in my pom.xml as separate dependency, then i add <​exclusions>​ element for io.netty.* group for vertx-core
 +
 +[11:48:27] <​s0va>​ plus i add <​exclusions>​ for all dependencies that require netty.
 +
 +[11:48:54] <​s0va>​ run mvn dependency:​tree to check for artifacts that require netty
 +
 +[11:49:37] <​purplefox>​ guest1_: i recommend excluding the 4.0.23 dependency so only the 4.0.28 should be used
 +
 +[11:49:47] <​purplefox>​ 4.0.28 should be backwards compat with 4.0.23 anyway
 +
 +[11:50:46] <​s0va>​ purplefox: 4.0.30 works okay for me :)
 +
 +[11:51:32] <​purplefox>​ it's probably ok, but we didn't test against that version so we can't guarantee. it's probably safer to go with the one vert.x uses
 +
 +[11:51:37] <​s0va>​ purplefox: while i'm at it, httpserver'​s idleTimeout()... is that suppose to work?
 +
 +[11:51:52] <​s0va>​ because if i set it on httpserveroptions
 +
 +[11:51:56] <​purplefox>​ features are generally supposed to work, yes ;)
 +
 +[11:53:32] <​s0va>​ then connect to server using telnet and do nothing, connection doesn'​t get dropped after specified number of millis; in fact it doesn'​t get dropped at all (3.0.0)
 +
 +[11:53:50] <​purplefox>​ s0va: do you have a reproducer?
 +
 +[11:54:24] <​s0va>​ pastebin?
 +
 +[11:54:46] <​s0va>​ just a sec.
 +
 +[11:54:50] <​s0va>​ i'll write one
 +
 +[11:55:06] <​s0va>​ bugs are still on eclipse bugtracker, aren't they?
 +
 +[11:56:00] <​purplefox>​ s0va: btw, idle timeout is in seconds, not millis
 +
 +[11:56:24] <​purplefox>​ this might be your problem
 +
 +[11:56:40] <​purplefox>​ if you set 30000 intending 30 seconds you will be waiting a long time ;)
 +
 +[11:59:36] <​guest1_>​ thank you, i will try to remove 4.0.23... do you know how to exclude for gradle.... ^^;;
 +
 +[12:00:18] <​guest1_>​ ok google it .. lol
 +
 +[12:03:56] <​purplefox>​ guest1_: +1 ;)
 +
 +[12:05:30] <​s0va>​ purplefox: http://​pastebin.com/​5fdpcRj8
 +
 +[12:05:51] <​s0va>​ so if you run this, then connect to server using telnet, do nothing, connection is not dropped.
 +
 +[12:06:00] <​s0va>​ purplefox: should i submit a bug report?
 +
 +[12:08:15] <​s0va>​ doesn'​t matter, i'll submit a bug report
 +
 +[12:08:50] <​purplefox>​ s0va: your reproducer works as expected for me
 +
 +[12:09:06] <​purplefox>​ I connect using telnet, and after 5 seconds the connection is closed:
 +
 +[12:09:27] <​purplefox>​ [email protected] ~ $ telnet localhost 8080
 +
 +[12:09:27] <​purplefox>​ Trying 127.0.0.1...
 +
 +[12:09:27] <​purplefox>​ Connected to localhost.
 +
 +[12:09:27] <​purplefox>​ Escape character is '​^]'​.
 +
 +[12:09:27] <​purplefox>​ Connection closed by foreign host.
 +
 +[12:09:28] <​purplefox>​ [email protected] ~ $
 +
 +[12:11:26] <​s0va>​ dafuq?
 +
 +[12:11:47] <​s0va>​ i'm using linux, oracle jdk 8u60.
 +
 +[12:12:28] <​purplefox>​ s0va: how, exactly, are you running this?
 +
 +[12:12:40] <​s0va>​ from idea
 +
 +[12:13:00] <​purplefox>​ same here
 +
 +[12:13:04] <​purplefox>​ also using linux
 +
 +[12:13:21] <​purplefox>​ what, exactly are you doing on the client side?
 +
 +[12:13:57] <​s0va>​ omg...
 +
 +[12:14:06] <​s0va>​ this timeout is in seconds, not milliseconds
 +
 +[12:14:30] <​purplefox>​ s0va: yes, that's what i said a few minutes ago ;)
 +
 +[12:14:52] <​s0va>​ sorry for false alert
 +
 +[12:14:59] <​purplefox>​ np
 +
 +[14:08:38] <​purplefox>​ temporalfox:​ pmlopes : there are 4 vertx-core PRs submitted by me (the top 4) waiting for review (they are all quite small) any chance you could take a look?
 +
 +[14:08:48] <​pmlopes>​ sure
 +
 +[14:11:28] <​temporalfox>​ ok
 +
 +[14:49:45] <​Krzysztof_>​ I guess that if (pipelining || keepAlive) in ConnectionManager could be simply if(keepAlive) unless it is done this way for readability
 +
 +[15:20:01] <​purplefox>​ Krzysztof_: can you make comments on the PR please? :)
 +
 +[15:29:30] <​Krzysztof_>​ I am testing it
 +
 +[15:29:45] <​Krzysztof_>​ small tests went fine, now running stress tests on real code
 +
 +[15:37:10] <​Krzysztof_>​ pipelining works fine
 +
 +[15:55:19] <​Krzysztof_>​ everything works fine, I've got expected performance,​ thanks
 +
 +[17:36:32] <​Narigo>​ purplefox, temporalfox is clement unavailable today? I saw a new commit in mysql postgresql which gives me a new kind of error now.... :)
 +
 +[17:51:53] <​purplefox>​ Narigo: he was around earlier
 +
 +[17:52:10] <​purplefox>​ it's almost 6pm in france now so probably he has finished for the day
 +
 +[19:31:54] <​Ashe`>​ hello; small questions (I'm 100% new to vert.x): I'd like to use vert.x as an API gateway, where I basically give vert.x a bunch of "​api"​ jars (they are JAX-RS interfaces) and want to go through a single proxy from which I'll forward the requests to the actual implementations
 +
 +[19:32:08] <​Ashe`>​ 1) does vert.x play nicely with JAX-RS and CXF?
 +
 +[19:32:40] <​Ashe`>​ 2) does the maven service factory allow completely unloading/​reloading of jars (like, to upgrade versions of the same classes)
 +
 +[21:21:37] *** ChanServ sets mode: +o purplefox
 +
 +[22:43:23] *** ChanServ sets mode: +o purplefox