Differences

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

Link to this comparison view

irc:1433196000 [2017/05/27 13:44] (current)
Line 1: Line 1:
 +[12:50:22] <​AlexLehm>​ there was an issue in the jenkins that caused the projects to build repeatedly (which Julien has fixed now)
 +
 +[12:51:00] <​AlexLehm>​ this showed occasional test errors in my project, which probably means that there is some timeing issue in the tests, but I have no idea how to analyse that
 +
 +[13:05:02] <​AlexLehm>​ or it could an issue with vertx context (haven'​t quite gotten then hang of this)
 +
 +[14:02:28] <​temporalfox>​ AlexLehm try to repeat the test with the @Repeat rule
 +
 +[14:02:32] <​temporalfox>​ like 1000 times
 +
 +[14:02:35] <​temporalfox>​ and see how it goes
 +
 +[14:02:44] <​AlexLehm>​ the problem, its not always the same test
 +
 +[14:03:02] <​AlexLehm>​ but i can do that
 +
 +[14:08:56] <​AlexLehm>​ I wonder if the tests run longer than the test method, sometimes I get RejectedExecutionException:​ event executor terminated
 +
 +[15:16:10] <​temporal_>​ AlexLehm such tests are problematic :-)
 +
 +[16:03:03] <​gigo1980>​ hi, i had the discussion. that events have no persitense when one node is going down.
 +
 +[16:03:22] <​gigo1980>​ is there a persistence layer for the events that are in the queue
 +
 +[16:24:24] <​AlexLehm>​ temporal_, I am not sure how to improve that, the actual tests are just using a vertx instance and are closing the resources when finished
 +
 +[16:25:12] <​temporal_>​ AlexLehm it should be handled I think and investigated
 +
 +[16:25:26] <​temporal_>​ AlexLehm there is no magic recipe to investigate such bugs
 +
 +[16:25:30] <​temporal_>​ and no generic recipe
 +
 +[16:27:58] <​AlexLehm>​ gigo1980, you are constantly joining and quitting with gigo1980_
 +
 +[16:29:05] <​gigo1980>​ this might be a computer at home. but why he is doing that
 +
 +[16:29:34] <​gigo1980>​ he has never do this bevore
 +
 +[16:32:12] <​AlexLehm>​ probably trying to cope with your nick already logged it
 +
 +[16:32:24] <​AlexLehm>​ well, doesn'​t matter
 +
 +[16:39:27] <​AlexLehm>​ ok, i will continue with some normal programmings and just run the build in background repeatedly
 +
 +[16:39:32] <​AlexLehm>​ lets see what that turns up
 +
 +[17:17:28] <​temporal_>​ this constant connect/​disconnect from gigo1980_ ​ is annoying
 +
 +[17:51:52] <​AlexLehm>​ you can disable join/leave messages in the irc client usually
 +
 +[17:52:48] <​AlexLehm>​ on a completely unrelated subject, listening to Jean-Michel Jarre while programming is great
 +
 +[17:55:35] <​temporal_>​ AlexLehm thanks I just disabled it
 +
 +[18:23:52] <​andyhedges>​ I have a problem with the mongo vertx client
 +
 +[18:24:03] <​andyhedges>​ the client works perfectly in a not replicated set
 +
 +[18:24:20] <​andyhedges>​ but when you run a command against a replicated set the response includes a timestamp
 +
 +[18:25:00] <​andyhedges>​ unfortunately vertx' bson implementation can parse a timestamp and just throws an unsupported method exception
 +
 +[18:25:25] <​andyhedges>​ can't
 +
 +[18:25:44] <​andyhedges>​ that should read '​can'​t parse a timestamp'​
 +
 +[18:26:51] <​andyhedges>​ java.lang.UnsupportedOperationException:​ null
 +
 +[18:26:52] <​andyhedges> ​ at io.vertx.ext.mongo.impl.codec.json.AbstractJsonCodec.readTimeStamp(AbstractJsonCodec.java:​387) ~[[redacted]-2.1.0-ALPHA-fat.jar:​na]
 +
 +[18:26:54] <​andyhedges> ​ at io.vertx.ext.mongo.impl.codec.json.AbstractJsonCodec.readValue(AbstractJsonCodec.java:​71) ~[[redacted]-2.1.0-ALPHA-fat.jar:​na]
 +
 +[18:26:56] <​andyhedges> ​ at io.vertx.ext.mongo.impl.codec.json.AbstractJsonCodec.readDocument(AbstractJsonCodec.java:​230) ~[[redacted]-2.1.0-ALPHA-fat.jar:​na]
 +
 +[21:17:24] <​gigo1980>​ hi together
 +
 +[21:17:42] <​gigo1980>​ the messeging from the eventbus is it done completly be hazlecast ?
 +
 +[21:30:00] <​gigo1980>​ http://​docs.hazelcast.org/​docs/​3.4/​manual/​html/​map-persistence.html ​ <-- is something like this posible to implement this in the event architecture ?
 +
 +[21:30:05] <​gigo1980>​ for some event types ?
 +
 +[21:32:51] <​gigo1980>​ ok is it posible to access the embedded hazelcast from vertx 3 directly to store own objects ?
 +
 +[21:34:29] <​AlexLehm>​ i think hazelcast is used in the background but it not directly accessible from vertx
 +
 +[21:40:10] <​gigo1980>​ so if i want to use hazzlecast directly i have to implement it right ?
 +
 +[21:40:47] <​gigo1980>​ i need an in memory object storage with persitence feature
 +
 +[21:43:15] <​AlexLehm>​ I think so (but I have basically no experience with hazelcast)
 +
 +[23:02:50] <​andyhedges>​ Anyone know what the drainHandler does
 +
 +[23:02:52] <​andyhedges>​ ?
 +
 +[23:03:48] <​andyhedges>​ I've got a situation where my event loop gets blocked with one for a long time, currently up to 5 mins
 +
 +[23:03:53] <​andyhedges>​ without sign of stop
 +
 +[23:05:20] <​AlexLehm>​ assume you are writing data to a stream (e.g. a socket) and the stream buffer is full, it will drop the rest of the chunks.
 +
 +[23:05:39] <​andyhedges>​ Yeah, ending a http response
 +
 +[23:05:43] <​andyhedges>​ so what would cause that?
 +
 +[23:05:58] <​AlexLehm>​ drainHandler waits until there is space left again in the buffer and then you can start to write again
 +
 +[23:06:23] <​AlexLehm>​ are you using this or is this in a library method?
 +
 +[23:06:42] <​andyhedges>​ Lib
 +
 +[23:06:52] <​andyhedges>​ I'm just calling end() on a response
 +
 +[23:06:57] <​AlexLehm>​ i would expect that this shouldn'​t block
 +
 +[23:07:27] <​AlexLehm>​ even if it cannot write the data to the socket, it still shouldn'​t block the event loop
 +
 +[23:08:30] <​andyhedges>​ Yeah, I'm a bit at a loss tbh
 +
 +[23:09:14] <​AlexLehm>​ I'm pretty sure Tim would say in this situation, "do you have a reproducer?"​
 +
 +[23:10:14] <​andyhedges>​ Yeah, I know, hard because it only happens on EC2
 +
 +[23:10:22] <​andyhedges>​ can't make it happen on my laptop
 +
 +[23:10:32] <​andyhedges>​ Is Tim away btw?
 +
 +[23:11:58] <​AlexLehm>​ i think he said something about appointments this week
 +
 +[23:13:33] <​AlexLehm>​ he has answered some mail questions though
 +
 +[23:14:51] <​andyhedges>​ OK, I think I'll turn my other questions into issues then
 +
 +[23:15:04] <​andyhedges>​ before I forget them :)
 +
 +[23:15:30] <​AlexLehm>​ I can recomment either Evernote or Trello for that
 +
 +[23:15:52] <​andyhedges>​ ha ha, thanks
 +
 +[23:23:39] <​AlexLehm>​ i usually forget which tool i used to note some items
 +
 +[23:25:20] <​andyhedges>​ Yeah, I often can't remember if I taken some notes electronically or in my notebook