pulsar

sorry for late response, now pulsar client timeout has a bug in timeout handle, the problem has been found. I am testing the functionality of transaction, I believe many problems can be fixed in 2.8.0.

I am also facing the same issue while upload schema.
HTTP 500 Internal Server <http://Errorjava.io|Errorjava.io>.ByteArrayInputStream@2299c582
Can anyone please provide solution on this issue?

The pulsar vesion is 2.7.0 and I run this using this docker-compose:

still getting the issue:
```2021-03-29 15:59:21.684 INFO  [140544168470336] ConnectionPool:85 | Created connection for <pulsar://localhost:6650>
2021-03-29 15:59:21.685 INFO  [140544142661184] ClientConnection:356 | [[::1]:51970 -&gt; [::1]:6650] Connected to broker
2021-03-29 15:59:21.782 INFO  [140544142661184] HandlerBase:54 | [<persistent://public/default/some-random-topic>, ] Getting connection from pool
2021-03-29 15:59:21.800 INFO  [140544142661184] ConnectionPool:85 | Created connection for <pulsar://pulsar-broker-0.pulsar-broker.platform.svc.cluster.local:6650>
2021-03-29 15:59:21.814 ERROR [140544142661184] ClientConnection:482 | [&lt;none&gt; -&gt; <pulsar://pulsar-broker-0.pulsar-broker.platform.svc.cluster.local:6650>] Resolve error: asio.netdb:1 : Host not found (authoritative)
2021-03-29 15:59:21.814 INFO  [140544142661184] ClientConnection:1428 | [&lt;none&gt; -&gt; <pulsar://pulsar-broker-0.pulsar-broker.platform.svc.cluster.local:6650>] Connection closed```
I used the command:
```helm install --set initialize=true pulsar --values values-minikube.yaml -n platform apache/pulsar```
and I used the values-minikube.yaml file attached:

Thanks - yeah the problem I am seeing right now is under heavy system load my clients are giving the following
```Failed to create producer: 48695 lookup request timedout after ms 5000"```
This seems to correlate at moments of high storage write latency

Not sure if it helps with diagnostics but our backlog is pretty huge - so not sure if reading from the backlog has other caches we can tune to help offset disk reads?

Seems to have solved the issue for a short amount of time but has now reappeared. I've attached a snippet of the logs, seems to happen every few minutes (ie the "Removed Consumer" message in the logs.

This file contains the broker stats captured after the consumers were stuck.  Of particular interest is the partition-2, which is using ledger 10830, which shows 5 messages in backlog but only 4 unacked/delivered:

This file contains some client logs I hacked into dotPulsar recently.  It shows a lot of setup and the sends, but near the bottom it also shows the messages coming in from the wire, including referencing messages 10830:0 through 10830:4, but conspicuously missing is 10830:3:

listenerName has been supported in C++ client, but this feature is going to be released in v2.8.0. So, the v2.7.1 libpulsar.so doesn’t support listenerName. If I want to use listenerName in C++, the only way is building my libpulsar.so? (building C++ is so complicated) Is there any official libpulsar.so that newer than v2.7.1 and I can download it?

So when v2.8.0 will be released? it has past due for  7 days …

2.7.1..
Here I am attaching the url for ur reference..
<https://pulsar.apache.org/docs/en/cookbooks-deduplication/#:~:text=Configure%20message%20deduplication&amp;text=Sets%20the%20default%20behavior%20for,level%20or%20the%20topic%20level>.

<@U6E5YMRKR> i just added the setting in broker section of values.yaml and I used below command to install "helm install --values examples/values-minikube.yaml pulsar-max apache/pulsar --set initialize=true ", but there is no change in broker.conf for that parameter

I think the description of the Subscription Auth Mode is clear enough.

Here are the thread stacks (thread 2 is mine, it's a stuckness reporter):

example of a Async Profiler 2.0 JFR trace analysed in Java Mission Control 8.0 UI

I did not use CustomPartition yet, but the doc makes me think that maybe the Java driver should be preferred :

An error occurred while testing function mesh，kubectl apply -f crd.yaml

this book suggests that Pulsar does attach a version number to each message