Hi Team Recently, I had seen strange behavior in kafka-connect. We have source connector with single task only, which reads from S3 bucket and copy to kafka topic.We have two worker node in a cluster, so at any point of time a task can be assigned to single worker node.
I saw in logs that both the worker node were processing/ reading the data from S3 bucket, which should be impossible since we have configured that a single task should be created and read the data. Is it possible in any scenario that due to worker process restarting multiple times or registering/ de-registering the connector, a task can be left assigned in both the worker node. Note : I have seen this only one time, after that it was never reproduced. Regards and Thanks Deepak Raghav