You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@pulsar.apache.org by Apache Pulsar Slack <ap...@gmail.com> on 2019/08/26 09:11:03 UTC
Slack digest for #general - 2019-08-26
2019-08-25 09:14:04 UTC - WingerPun: @WingerPun has joined the channel
----
2019-08-25 16:48:16 UTC - Chris Bartholomew: That doc refers to synchronous replication. This medium article explains the difference between asynchronous and synchronous replication: <https://medium.com/@pckeyan/apache-pulsar-geo-replication-ad4f0ca3224b>
----
2019-08-25 22:19:23 UTC - Igor Zubchenok: up
----
2019-08-25 22:20:24 UTC - Igor Zubchenok: I implemented small ping service that measures message time of delivery by creating both producer and consumer for a topic. And send a message every 10 ms.
----
2019-08-25 22:37:07 UTC - Igor Zubchenok: What I see strange is these peaks.
The same peak pattern at the same moment happens at both servers if I run test from two different location.
It happens every 5-6.5 seconds
----
2019-08-25 22:38:54 UTC - Igor Zubchenok:
----
2019-08-25 22:58:49 UTC - Igor Zubchenok: Another strange point is increasing time of delivery from 2ms to 13ms between peaks. If I zoom it looks like:
----
2019-08-25 22:59:14 UTC - Igor Zubchenok:
----
2019-08-26 01:39:53 UTC - Igor Zubchenok: I measured CPU very often and found that broker process CPU increases during peaks.
----
2019-08-26 02:48:58 UTC - Luke Lu: Hey guys, has anyone here seen the “head-scratching” subscription stats, where no messages are coming in for a while and there is a constant backlog on a connected consumer waiting for data. Shouldn’t the subscription backlog go to zero in this case? This is using the websocket API:
```
"SomeSubscription" : {
"msgRateOut" : 0.0,
"msgThroughputOut" : 0.0,
"msgRateRedeliver" : 0.0,
"msgBacklog" : 377,
"blockedSubscriptionOnUnackedMsgs" : false,
"msgDelayed" : 0,
"unackedMessages" : 0,
"type" : "Exclusive",
"activeConsumerName" : "de307",
"msgRateExpired" : 0.0,
"consumers" : [ {
"msgRateOut" : 0.0,
"msgThroughputOut" : 0.0,
"msgRateRedeliver" : 0.0,
"consumerName" : "de307",
"availablePermits" : 635,
"unackedMessages" : 0,
"blockedConsumerOnUnackedMsgs" : false,
"metadata" : { },
"connectedSince" : "2019-08-25T15:38:05.655Z",
"clientVersion" : "2.4.0",
"address" : "/10.36.33.96:49970"
} ],
"isReplicated" : false
}
```
----
2019-08-26 07:15:09 UTC - xue: I modify broker.conf, functionsWorkerEnabled = false, all clusters can be started successfully, function-worker can be started separately, through bin/pulsar-admin namespaces-clusters--cluster <existing-clusters>, <new-cluster> public/functions commands to solve another problem, when doing geo replication test. Wait, is the pulsarFunctionsCluster attribute in the functions_worker.yml file configured to be the name of all clusters, or the name of the local cluster?
----
2019-08-26 07:18:06 UTC - xue: pulsarFunctionsCluster: pulsar-cluster,or pulsarFunctionsCluster: pulsar-cluster,pulsar-cluster2?
----
2019-08-26 08:17:18 UTC - Emanuel: @Emanuel has joined the channel
----
2019-08-26 08:20:27 UTC - Emanuel: Hey guys, is it possible to use pulsar SQL from pulsar functions or alternatively from a flink processor?
----