Hi Yuval, If your pipeline can implement an exactly-once delivery guarantee depends on your pipeline. Usually Flink’s fault tolerance mechanism is built around periodically snapshots of intermediate states called checkpoints. As long as checkpointing is enabled and all the operators you are using in your pipeline can reply the snapshotted data on recovery your sink should only emit the same element once.
In case you already have pipeline running you can also tell us about the topology so we can explain it with your example. Best, Fabian
