Matt Howlett
Matt Howlett
I think there's a race condition when producing a message to `p.ProduceChannel()`, then immediately calling Flush. I think the following sequence is possible: message take from `produceChannel` here: https://github.com/confluentinc/confluent-kafka-go/blob/master/kafka/producer.go#L556 length...
on my local machine: ``` % dotnet run -- -b localhost:9092 -m throughput ``` gives: ``` rdkafka#producer-1 producing on dotnet-benchmark [Task] Produced 5000000 messages in 192010ms 26k msg/s rdkafka#producer-2 producing...
this test hangs in v1.8.2, as indicated.
On OSX, calling `Library.Load(path)` with an absolute path ignores the absolute path and loads the `librdkafka.dylib` included in the librdkafka.redist package referenced by `Confluent.Kafka`. This is the case even if...
we should look at adding automated style checking using `autopep8` or similar.
It has been reported that: ``` AdminClient({...}).delete_topics( topics=topics, operation_timeout=30.0 ) ``` gives the exception: ``` cimpl.KafkaException: KafkaError{code=_DESTROY,val=-197,str="Handle is terminating: Success"} ``` whereas: ``` client: AdminClient = AdminClient({...}) client.delete_topics( topics=topics, operation_timeout=30.0...
Dependency: https://github.com/confluentinc/rest-utils/pull/267
an idea for how to set up config for your new EoS example @emasab
This test was never completed. The case is covered by the ProduceConsumeExternalRefProtobuf Serde integration test. So we can just delete this.