Slack digest for #general - 2019-12-20

[Apache Pulsar Slack <ap...@gmail.com>]

2019-12-19 09:59:44 UTC - Pradeep Mishra: Hi, i need help figuring out how to increase consuming throughput in pulsar queue.
transaction per second for consuming is same whether i use 1 consumer or 4 consumers
anybody knows how to increase tps by increasing consumers in same queue?
+1 : Gautam Lodhiya
-1 : Vladimir Shchur
----
2019-12-19 11:43:06 UTC - Roman Popenov: @Vladimir Shchur, try looking for `LogLevel` in the chart and see how they set the log level for bookkeeper
----
2019-12-19 11:43:22 UTC - Roman Popenov: And then try that additional parameter with broker
----
2019-12-19 11:46:09 UTC - vikash: thanks it  works
----
2019-12-19 11:51:46 UTC - Vladimir Shchur: I've checked many times.. For bookeeper there is only memory logging configured: -XX:G1LogLevel=finest\"", looks like it's not useful
----
2019-12-19 12:32:00 UTC - Roman Popenov: What about something like `-Dlog4j.logLevel=DEBUG`
----
2019-12-19 12:32:21 UTC - Roman Popenov: I haven’t looked at what logger that is used for the project, but I think there something similar to that you can find
----
2019-12-19 12:32:39 UTC - Roman Popenov: Or maybe a config file
```-Dlog4j.configurationFile=```

----
2019-12-19 12:33:58 UTC - Roman Popenov: There is also
```-Droot.log.level=```

----
2019-12-19 14:08:22 UTC - Vladimir Shchur: @Roman Popenov I wonder where to include those flags :slightly_smiling_face:
----
2019-12-19 14:13:52 UTC - Roman Popenov: in the jvm xargs
----
2019-12-19 14:20:33 UTC - Vladimir Shchur: Looks like there is only one place to put them. Still this setup didn't change the level

`PULSAR_MEM: "\"-Xms128m -Xmx256m -XX:MaxDirectMemorySize=128m -Dio.netty.leakDetectionLevel=disabled -Dio.netty.recycler.linkCapacity=1024 -Dlog4j.logLevel=DEBUG -Droot.log.level=DEBUG -XX:+ParallelRefProcEnabled -XX:+UnlockExperimentalVMOptions -XX:+AggressiveOpts -XX:+DoEscapeAnalysis -XX:ParallelGCThreads=32 -XX:ConcGCThreads=32 -XX:G1NewSizePercent=50 -XX:+DisableExplicitGC -XX:-ResizePLAB -XX:+ExitOnOutOfMemoryError -XX:+PerfDisableSharedMem\""`
----
2019-12-19 14:22:06 UTC - Vladimir Shchur: <https://github.com/apache/pulsar/blob/master/deployment/kubernetes/helm/pulsar/values-mini.yaml#L192>
----
2019-12-19 14:25:12 UTC - Roman Popenov: Have you tried loading the SLF4J config file ?
----
2019-12-19 14:26:10 UTC - Roman Popenov: `-Djava.util.logging.config.file=logging.properties`
----
2019-12-19 14:27:12 UTC - Roman Popenov: ```#logging.properties file contents

#Define handlers
handlers=java.util.logging.ConsoleHandler

#Set handler log level
java.util.logging.ConsoleHandler.level=FINE

#Define your logger level
com.company.application.package.package.level=FINE

#Assign your handler to your logger
com.company.application.package.package.handlers=java.util.logging.ConsoleHandle```
----
2019-12-19 14:27:16 UTC - Roman Popenov: something along those lines
----
2019-12-19 14:28:08 UTC - Vladimir Shchur: No, I can't do that, since there is no way to restart the pod with new configuration except deleting it and it will delete my file as well :disappointed:
----
2019-12-19 14:30:26 UTC - Roman Popenov: ```-Dorg.slf4j.simpleLogger.defaultLogLevel=debug```

----
2019-12-19 14:30:46 UTC - Roman Popenov: <http://www.slf4j.org/api/org/slf4j/impl/SimpleLogger.html>
----
2019-12-19 14:31:11 UTC - Roman Popenov: I think though
```-Dorg.slf4j.simpleLogger.defaultLogLevel=debug```
should work as system property
----
2019-12-19 14:51:23 UTC - Vladimir Shchur: Looks like I've found correct flag `-Dpulsar.log.level=info` , still it goes after helm parameters, so I've no idea for now how to overwrite it
+1 : Roman Popenov
----
2019-12-19 14:54:11 UTC - Roman Popenov: Good to know!
----
2019-12-19 15:39:04 UTC - Roman Popenov: Are there any instrcutions on how to expose the pulsar-proxy through an ingress or load balancer to the outside world?
----
2019-12-19 15:51:48 UTC - Vladimir Shchur: I don't think there is anything specific to pulsar here, any generic kubernetes solutions will work
----
2019-12-19 15:58:06 UTC - David Kjerrumgaard: <https://cloud.google.com/kubernetes-engine/docs/tutorials/http-balancer> shows how to setup a simple LB
----
2019-12-19 18:24:54 UTC - Rodrigo Batista: @Rodrigo Batista has joined the channel
----
2019-12-19 18:32:09 UTC - Charlie Walker: @Charlie Walker has joined the channel
----
2019-12-19 19:15:37 UTC - Armin Woworsky: @Armin Woworsky has joined the channel
----
2019-12-19 19:17:12 UTC - Charlie Walker: Hello,
I am trying to delete a persistent topic which has no active producers or consumers, but does have a subscription registered. To delete the topic I need to unsubscribe, but my unsubscribe attempts fail with the error `org.apache.bookkeeper.mledger.ManagedLedgerException: ManagedCursor not found:` logged at the broker. When I run `stats-internal` I don’t see a cursor for my subscription. I have tried to run `reset-cursor` but that fails as well. How do I regenerate a cursor for my subscription so I can unsubscribe and then delete this topic? Or is there something else I am missing here?
----
2019-12-19 20:01:47 UTC - Charlie Walker: I was able to solve this but it wasn’t very pretty.
• I re-connected my consumer and that re-generated the cursor
• I then disconnected the consumer and ran `unsubscribe`. The command timed out at the admin client, but at the broker there was a unsubscribe successful log along with a bookie error log. After this `stats` still showed the subscription registered.
• I then restarted my consumer which complained the topic was fenced.
• I restarted the broker and the subscription was gone and I was able to delete the topic
----
2019-12-20 08:23:40 UTC - Sijie Guo: @Sijie Guo set the channel topic: - First Pulsar Summit is planned for April 2020 in San Francisco. CFP and pre-registration are open: <https://pulsar.apache.org/blog/2019/12/18/Pulsar-summit-cfp/>
- Pullsar release 2.4.2 released!
<http://pulsar.apache.org/release-notes/#2.4.2>
----