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 2018/10/29 09:11:05 UTC
Slack digest for #general - 2018-10-29
2018-10-28 16:59:42 UTC - William Fry: I’m building off of the `apachepulsar/pulsar:2.2.0` Docker image, trying to set up a docker-compose file to create a function in standalone mode. If I run with `create` (in cluster mode), I get `Reason: javax.ws.rs.ProcessingException: java.net.ConnectException: Connection refused (Connection refused)`. If I run with `localrun`, I get `AttributeError: 'ConsumerConfiguration' object has no attribute 'read_compacted'`. It sounds like I want to do the latter for just running a local docker container, but I can’t find anything in the docs that explains the `read_compacted` attribute?
----
2018-10-28 17:29:21 UTC - William Fry: I see it defined here: <https://github.com/apache/pulsar/blob/9a9378944c8a6c3cd9124c888ad58f67e127ceb0/pulsar-client-cpp/python/src/config.cc>
----
2018-10-28 17:31:46 UTC - Sijie Guo: In localrun, did you update your python client library? In cluster mode, did you point the admin CLI to the right http endpoint? (Did you expose the http port to be accessed out side of docker?)
----
2018-10-28 17:32:33 UTC - William Fry: In localrun, I’m using the client library that is packaged in that Docker image.
----
2018-10-28 17:33:22 UTC - William Fry: I’d expect mismatched client versions or HTTP access, if I were running in localrun mode from outside of Docker, but that’s not what I’m doing as of now
----
2018-10-28 17:44:42 UTC - William Fry: Essentially, I get a container running Pulsar in standalone mode and then use `docker exec -it <container> /bin/bash`
----
2018-10-28 17:45:03 UTC - William Fry: From there, I run the pulsar-admin functions command for `localrun` which leads to the `read_compacted` error.
----
2018-10-28 17:54:26 UTC - William Fry: Ok, I got it to work when using `create` after the container is up. Originally I had two commands in the docker-compose.yml, one for starting the cluster and one for creating the function. I imagine the cluster wasn’t fully up when the command for creating the function ran.
----
2018-10-28 17:54:48 UTC - William Fry: Now it creates successfully, but if I try to trigger it with the admin API, I get `Reason: HTTP 408 Request Timeout`
----
2018-10-28 17:54:54 UTC - William Fry: Any idea what could cause this, @Sijie Guo
----
2018-10-28 18:08:35 UTC - William Fry: Also, are there still issues with the `pip` package for the > 1.9 python client?
----
2018-10-28 23:36:30 UTC - Yuya Ishida: @Yuya Ishida has joined the channel
----
2018-10-29 03:11:11 UTC - Benjamin: @Benjamin has joined the channel
----
2018-10-29 05:08:42 UTC - Baoyi Chen: @Baoyi Chen has joined the channel
----
2018-10-29 07:44:15 UTC - dba: Hi guys
I am working on the .NET/C# client. So far I can produce and consume. I noticed two things:
1. Even though I set protocol version to v7 and send messages with magicNumber+Checksum, when I was consuming the messages I got the package without magicNumber+checksum. I had to set protocol version to v11 before the server would add magicNumber+checksum.
2. When sending a message I forgot to set producer_name on the metadata. The server accepted the message (got a SendReceipt), but it was lost. I could see in the logs that the server got an error trying to deserialize the metadata.
Could anyone explain the difference between the Unsubscribe and CloseConsumer commands?
----
2018-10-29 08:10:54 UTC - Ali Ahmed: @dba thank you for your effort are trying to build a native C# client or wrapping over the C++ library ?
----
2018-10-29 08:11:32 UTC - dba: @Ali Ahmed It is native.
----
2018-10-29 08:12:31 UTC - Ali Ahmed: @Matteo Merli @Ivan Kelly probably know moe about the wire protocol, do you have preview code that you can share ?
----
2018-10-29 08:13:23 UTC - dba: @Ali Ahmed I don't have a preview just yet. I hope to have something before 2019.
----