2018-11-19 10:49:18 UTC - Franck NEGRELLO: @Franck NEGRELLO has joined the channel ---- 2018-11-19 12:36:49 UTC - Ganga Lakshmanasamy: We have a program which would create a producer instance for each user who connects a particular app. So how do we handle the situation when there are 200,000 users? This would create the same number of producers which will have individual threads running inside each one of them. Is there any other way we can handle this? ---- 2018-11-19 12:38:39 UTC - Ganga Lakshmanasamy: Here is another question, I am using 2.1.1 version and I see that MessageBuilder is deprecated in this. Is there any documentation on what function to use to send out a message to topic. I am using java apis. ---- 2018-11-19 12:42:27 UTC - Samuel Sun: hi , is there anyway to know the time broker write the data to the bookeeper ? not the client sent time ---- 2018-11-19 12:55:48 UTC - Ivan Kelly: @Ganga Lakshmanasamy each user has their own topic? ---- 2018-11-19 12:56:16 UTC - Ivan Kelly: @Ganga Lakshmanasamy MessageBuilder has been deprecated in favour of TypedMessageBuilder ---- 2018-11-19 12:57:13 UTC - Ivan Kelly: it's to do with the new schema stuff in 2.1 ---- 2018-11-19 12:58:20 UTC - Ivan Kelly: @Samuel Sun no such timestamp is recorded. What usecase do you have that would need it? ---- 2018-11-19 16:24:09 UTC - Emma Pollum: Hello! Can someone help me with a pulsar->zookeeper issue? I'm trying to launch my nodes but they keep failing with the following exception ---- 2018-11-19 16:25:03 UTC - Emma Pollum: ---- 2018-11-19 16:40:38 UTC - Emma Pollum: I have tried allowing the timeout to expire, and deleting the znode manually. Even if I load a brand new node into the cluster it still gives the same exception. ---- 2018-11-19 16:49:56 UTC - Emma Pollum: Update: I commented out my settings for google offloader and I can now connect. Does anyone know why this would cause the problem stated before? ---- 2018-11-19 17:00:52 UTC - Sijie Guo: @Emma Pollum It can be you misconfigure your google offloader and cause it throws any exception and crash broker startup. so when it reattempts, the broker’s znode is still not expired yet. ---- 2018-11-19 17:01:42 UTC - Emma Pollum: Hmm. I'll look into my configuration. Thanks. ---- 2018-11-19 20:03:36 UTC - Ivan Kelly: @Sijie Guo I'm not sure offloader will mess up the broker boot. it should just default to using the null offloader ---- 2018-11-19 20:04:15 UTC - Ivan Kelly: but requick reboots does look like part of the root cause ---- 2018-11-19 20:54:26 UTC - Emma Pollum: It was totally my configuration! ---- 2018-11-19 20:55:05 UTC - Emma Pollum: So next question for you guys, are there any rules or guidelines you use when deciding if and how to partion a topic? Should it be based on message size, publish rate, etc? ---- 2018-11-19 20:56:11 UTC - Matteo Merli: One partition is being served from 1 broker. So if the load on a partition exceeds (or gets close to) the capacity of 1 node, that should be a partitioned topic ---- 2018-11-19 20:56:41 UTC - Matteo Merli: and by load, you should consider max(cpu, traffic-in, traffic-out) ---- 2018-11-19 20:57:08 UTC - Matteo Merli: that’s from Pulsar’s perspective ---- 2018-11-19 20:58:10 UTC - Matteo Merli: If you use partitions as a way to distribute load to consumers while retaining per-partition message ordering. then the number of partition might be dictated by the number of consumers (but that’s kind of a different story) ---- 2018-11-19 20:59:40 UTC - Emma Pollum: :thumbsup: Thank you ---- 2018-11-20 02:28:51 UTC - 陈梓立: @陈梓立 has joined the channel ---- 2018-11-20 06:48:26 UTC - Liyong Wang: @Liyong Wang has joined the channel ---- 2018-11-20 07:19:56 UTC - huanjun: @huanjun has joined the channel ----
