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 2019/08/08 09:11:09 UTC
Slack digest for #general - 2019-08-08
2019-08-07 11:27:56 UTC - boldrick: @Matteo Merli thanks
----
2019-08-07 12:19:39 UTC - Kim Christian Gaarder: Think I found some bugs in very basic Reader functionality, see <https://github.com/kimcs/pulsar-readerbug> for reproduction code. The code found bugs when using Reader functionality including startMessageIdInclusive configuration, the new 2.4.0 hasMessageAvailable method and the readNext method.
----
2019-08-07 12:23:49 UTC - Kim Christian Gaarder: Also, while working on reproducing those bugs, I saw another problem. It appeared that I was denied to subscribe on topics due to “incompatible schema”. e.g. If I switch running the tests with java 8 and java 11 compiler and runtime, then I get the following in my log (even when using default byte[] as schema):
14:22:02.934 [pulsar-client-io-6-1] WARN org.apache.pulsar.client.impl.PulsarClientImpl
- [public/default/test-start-message-id-inclusive-magic-m6km32j] Failed to get create topic reader
java.util.concurrent.CompletionException: org.apache.pulsar.client.api.PulsarClientException$IncompatibleSchemaException: Trying to subscribe with incompatible schema
----
2019-08-07 12:25:52 UTC - Kim Christian Gaarder: @Matteo Merli @Sijie Guo Well, that didn’t go so well, I hit what I think are some bugs. See my new post in general for details on this.
----
2019-08-07 12:29:17 UTC - Kim Christian Gaarder: This is particularly disturbing as I actually delete the topic using the java admin client before attempting to create the reader on that topic … So very mysterious
----
2019-08-07 12:45:35 UTC - Sijie Guo: were you using a different schema before?
----
2019-08-07 13:07:06 UTC - jgil: @jgil has joined the channel
----
2019-08-07 13:46:06 UTC - Chris Miller: Have a look at <https://github.com/apache/pulsar/pull/3836/> and <https://github.com/apache/pulsar/issues/3837>
----
2019-08-07 14:03:08 UTC - Aaron: I noticed that when cumulative acking a batched message, two odd things happen:
1. Unless the batched message is the last in the batch, Pulsar sends an acknowledgment for the previous batched message.
2. After cumulative acknowledging a batched message in the middle of a batch, the ledger cursor goes to the next batch of messages instead of going to the next message in the current batch.
Are these behaviors intended and accurate?
----
2019-08-07 17:40:56 UTC - Chris Bartholomew: @Sijie Guo Will do
----
2019-08-07 21:21:16 UTC - Devin G. Bost: <https://www.meetup.com/utah-data-engineering-meetup/events/263761070/>
+1 : Ali Ahmed, Jerry Peng, Jon Bock, Karthik Ramasamy
----
2019-08-07 22:47:42 UTC - Grant Wu: How does tiered storage interact with retention?
----
2019-08-07 22:49:40 UTC - Grant Wu: i.e. what happens if you set 7 day message retention and you have 2 week old messages in s3?
----
2019-08-08 01:39:32 UTC - Jianfeng Qiao: Hi, while using cpp client, anyone run into this issue that consumer always stops receiving messages after running for a while?
----
2019-08-08 01:40:25 UTC - Jianfeng Qiao: I looked into the code, it seems that the work posted by ExecutorService is not processed.
----
2019-08-08 01:40:43 UTC - Jianfeng Qiao: The version of cpp client is 2.3.1
----
2019-08-08 01:48:01 UTC - Sijie Guo: have you checked `topic stats` and `topic stats-internal`?
----
2019-08-08 01:51:30 UTC - Jianfeng Qiao: Yes
----
2019-08-08 01:52:09 UTC - Jianfeng Qiao: The messages are not processed and reside in backlog
----
2019-08-08 01:53:45 UTC - Sijie Guo: is it blocked due to unacked or permits are not sent back to the broker?
----
2019-08-08 01:54:59 UTC - Jianfeng Qiao: I see the source code and find the reason that messages are not processed is that the task posted by listenerExecutor_->postWork() is not processed
----
2019-08-08 02:00:03 UTC - Sijie Guo: ok can you reproduce the bug using a test?
----
2019-08-08 06:23:26 UTC - divyasree: <!here> Do pulsar support Hollow Netflix OSS?
----
2019-08-08 06:24:38 UTC - Ali Ahmed: please don’t use “here”.
hollow can be easily integrated with pulsar , I don’t think it’s been done yet.
----
2019-08-08 06:44:27 UTC - Luo Tian: report a bug on documentation: <https://pulsar.apache.org/docs/en/functions-api/>
----
2019-08-08 06:44:58 UTC - Luo Tian: code snippet: `import java.util.Function;` should be `import java.util.function.Function`
----
2019-08-08 06:45:36 UTC - Sijie Guo: Do you mind filing a github issue for it?
<https://github.com/apache/pulsar>
----
2019-08-08 06:45:39 UTC - divyasree: Is there any documentation for hollow integration?
----
2019-08-08 06:46:19 UTC - Yi Tang: a bit issue about pulsar dashboard doc, i was misled that thought no ready docker image and did build a image from docker file
----
2019-08-08 06:46:53 UTC - Luo Tian: sure.
----
2019-08-08 06:47:58 UTC - Ali Ahmed: not currently it’s been while since I took but essentially you will need to implement a custom HollowIncrementalProducer service that reads from pulsar.
----
2019-08-08 06:49:24 UTC - Sijie Guo: Do you mind filing a github issue for it?
<https://github.com/apache/pulsar>
----
2019-08-08 06:50:03 UTC - divyasree: oh ok
----
2019-08-08 06:53:33 UTC - Yi Tang: sure, a little question more, this dashboard only provide view of the cluster , no admin action could be done on it?
----
2019-08-08 06:59:01 UTC - Yi Tang: <https://github.com/apache/pulsar/issues/4913>
----
2019-08-08 08:38:55 UTC - Yuvaraj Loganathan: @divyasree Please don't use @ here
----