Re: Race condition with stream use of Global KTable

For 3), my reasoning is that: since topic-2 is derived from topic1, and hence as long as they are co-partitioned (meaning that the record from topic-1 resulting in the record(s) in topic-2 are guaranteed to be sent to the same client) there's no race condition since the record to query will always

Re: Using processor API via DSL

Hi Alessandro, the `TransformSupplier` is internally wrapped with a `ProcessorSupplier`, so the statement `transform` is essentially equivalent to adding the Transformer via Topology#addProcessor() to your processor topology is correct. If you do not change the key, you should definitely use on

Kafka client keeps fetch metadata for deleted topic, which leads to UNKNOWN_TOPIC_EXCEPTION

Recently we used Apache Flink to consume kafka topics with a regex pattern. It is found that when we deleted some unused topics, the logs will keep flushing UNKNOWN_TOPIC_EXCEPTION. I looked up the source code of kafka client and found that for consumer, topicExpiry is disable in Metadata, whic