kettle-beam icon indicating copy to clipboard operation
kettle-beam copied to clipboard

Beam Kafka Consumer not committing offset to the configured consumer group

Open rhaces opened this issue 5 years ago • 2 comments

When running a pipeline with a Beam Kafka consumer, the kettle-beam job will not commit the offset to kafka and therefore if the process is restarted it will duplicate messages (using auto.offset.reset set to earliest)

This has been tested with Neo4j as the output of the job.

rhaces avatar Feb 10 '20 18:02 rhaces

As usual for parallel environments, the answer is complicated...

https://stackoverflow.com/questions/48406321/kafkaio-checkpoint-how-to-commit-offsets-to-kafka

mattcasters avatar Feb 11 '20 12:02 mattcasters

I made the option "Allow offsets to be committed back" the default.

mattcasters avatar Feb 12 '20 16:02 mattcasters