Kafka configuration reported as unknown config

[itatdcer]

Describe the bug

I have configured a property for a kafka incoming topic which quarkus report as unknown:
mp.messaging.incoming.topic-test1-in.pause-if-no-requests=false

The documentation has this propertties available to use but I get a warning on startup:
2022-09-13 17:34:03,125 WARN [org.apa.kaf.cli.con.ConsumerConfig] (main) The configuration 'pause-if-no-requests' was supplied but isn't a known config.

In DEV mode, sometimes, it also issues a warning for the property:
mp.messaging.incoming.topic-test1-in.group.id=my-consumer-group
The warning is issues on the AdminClientConfig class:
2022-09-13 17:38:29,082 WARN [org.apa.kaf.cli.adm.AdminClientConfig] (executor-thread-0) The configuration 'group.id' was supplied but isn't a known config

Expected behavior

No warning issued if these property is present

Actual behavior

No response

How to Reproduce?

No response

Output of uname -a or ver

No response

Output of java -version

No response

GraalVM version (if different from Java)

No response

Quarkus version or git rev

2.12.1.Final

Build tool (ie. output of mvnw --version or gradlew --version)

No response

Additional information

No response

[quarkus-bot]

/cc @alesj, @cescoffier, @ozangunalp

[ozangunalp]

The pause-if-no-requests log is already fixed upstream, and will soon land in Quarkus.
The group-id log comes from the global kafka broker health check. We pass the default kafka configuration directly to the admin client we use to check the cluster.
We cannot filter for only admin config keys, as there can be security related config as well, but we can at least filter out the group.id.

[cescoffier]

I would add that these warnings are totally normal. If you write a serializer/deserializer/partitioner/interceptor.... anything that receives a config will display these warnings.

We can clean up the configuration with what we know is unrelated, but imagine that your thing requires the group.id config. It won't receive it. So, we need to clean up per target (consumer/producer/admin) and make sure we are not removing anything that could be used. It's an open set, so impossible to say exactly.