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 2018/05/17 09:11:02 UTC
Slack digest for #general - 2018-05-17
2018-05-16 10:35:35 UTC - Byron: Got the same message. However I commented out the Boost Python3 bits to use Python3 and that worked (althought maybe the boost env variables needs to be set expliclt? <https://stackoverflow.com/questions/46374747/cmake-3-9-3-cannot-find-boost1-65-1-boost-python>) .. but then I ran into a `make pulsarShared` error about `Undefined symbols for architecture x86_64`
----
2018-05-16 17:25:25 UTC - Matteo Merli: @Byron Ok, I think it should be working now. I have updated the formula and fixed the cmake list as well. Please update the formula from gist.
For python3 , I’ve added the option: `brew install ./libpulsar.rb --with-python3`. After that I was able to install successfully on a MacOS-10.11 with Python3 installed
----
2018-05-16 17:35:39 UTC - Byron: :raised_hands:
----
2018-05-16 17:40:47 UTC - Byron: I was able to install and then build the examples
----
2018-05-16 17:41:10 UTC - Matteo Merli: Great!
----
2018-05-16 18:00:30 UTC - Byron: Examples are working..
----
2018-05-16 18:10:11 UTC - Byron: I think I came across this question before.. but is there a way to set the initial message for new subscriptions? If I recall this will be in 2.0?
----
2018-05-16 18:10:44 UTC - Matteo Merli: It is, though it’s not exposed in C++ yet
----
2018-05-16 18:10:47 UTC - Byron: And the admin API can reset the position if I recall
----
2018-05-16 18:10:50 UTC - Byron: Ok
----
2018-05-16 18:11:44 UTC - Matteo Merli: Yes, that works, though in Java there’s the option to create a subscription on earliest or latest message, if it’s not there yet
----
2018-05-16 18:12:52 UTC - Matteo Merli: we have an handful of features to add to C++. Tracking at <https://github.com/apache/incubator-pulsar/projects/7>
----
2018-05-16 18:17:12 UTC - Byron: Great thanks. I am super excited there is a Go client
----
2018-05-16 18:17:45 UTC - Matteo Merli: thanks for all the inputs on the API!
----
2018-05-16 18:18:10 UTC - Byron: Of course! Any time
----
2018-05-16 19:30:04 UTC - Ali Ahmed: I still get an error
----
2018-05-16 19:30:10 UTC - Ali Ahmed: @Ali Ahmed uploaded a file: <https://apache-pulsar.slack.com/files/U6EHQ91KM/FAQCAS3RN/01.cmake|01.cmake>
----
2018-05-16 20:52:35 UTC - Matteo Merli: @Ali Ahmed Fixed the recipe. It should work without gtest & gmock now
----
2018-05-16 20:57:39 UTC - Ali Ahmed: everything looks good
```
brew install <https://gist.githubusercontent.com/merlimat/1b4925e01bc7fcfb04b9a6193026a020/raw/50ef3692fb340842501401e22a52c1b3ffc3c9dc/libpulsar.rb>
######################################################################## 100.0%
==> Downloading <https://s3-us-west-2.amazonaws.com/pulsar-preview/apache-pulsar-2.1.0-incubating-SNAPSHOT-src.tar.gz>
######################################################################## 100.0%
==> cmake . -DBUILD_TESTS=OFF
==> make pulsarShared pulsarStatic
🍺 /usr/local/Cellar/libpulsar/2.1.0-incubating-SNAPSHOT: 45 files, 10MB, built in 1 minute 41 seconds
```
----
2018-05-16 21:03:44 UTC - Igor Zubchenok: Hello
What could be the reason of the following exceptions? I actually expect pulsar client to handle such issues internally (automatically reconnect to other available brokers).
```1. org.apache.pulsar.client.api.PulsarClientException: Not connected to broker. State: Connecting
2. org.apache.pulsar.client.api.PulsarClientException: java.nio.channels.ClosedChannelException
3. org.apache.pulsar.client.api.PulsarClientException: Disconnected from server at 192.168.1.10/192.168.1.10:6650```
----
2018-05-16 21:04:49 UTC - Matteo Merli: Is that in a consumer calling `acknowledge()` ?
----
2018-05-16 21:09:30 UTC - Igor Zubchenok: Yes! Right!
----
2018-05-16 21:12:16 UTC - Matteo Merli: Yes, acks are not retried because they are best-effort by nature (there’s no ack on the ack). The result is that when an ack fail (eg: for being temporarily disconnected, the messages are replayed by broker).
In 2.0, we have improved that part by doing transparent batching and keeping track of acks from client side. There you won’t get exception when disconnected.
In 1.22 the recommendation is to use `asyncAcknowledge()`
----
2018-05-16 21:13:39 UTC - Igor Zubchenok: I actually use `acknowledgeCumulativeAsync` and `acknowledgeAsync` and have these exceptions
----
2018-05-16 21:14:06 UTC - Matteo Merli: yes, the sync version is based on the async. I was meaning to ignore exception
----
2018-05-16 21:14:50 UTC - Matteo Merli: there’s not much to do in any case, message will be replayed and consumer will get the chance to ack again
----
2018-05-16 21:16:57 UTC - Igor Zubchenok: Thanks, I'll try to find out how to live with this tomorrow. :wink:
+1 : Sijie Guo, Ali Ahmed
----
2018-05-16 21:19:45 UTC - Matteo Merli: :slightly_smiling_face:
----