Does Spark Structured streaming query require write-ahead logs for at least once delivery with ForeachBatch output sink?

[DimaP]

Is it sufficient to use the checkpoint directory with write-ahead logs?

BTW. I use Kafka connector to read data from EventHub