The KafkaOffsetUDR is used to determine an offset which is not at the beginning or the end of collection, and only applies if you have selected Start at requested in the Kafka Collection agent configuration. The Kafka Collection agent waits for the KafkaOffsetUDR and does not consume before the offset information. Messages can be persisted to a database or aggregation agent.


If you send a KafkaOffsetUDR from initialize without any content, messages are read from the beginning (from the first offset).

The following fields are included in the KafkaOffsetUDR:


offsets (map<int,long>)

This field is populated with offset information. As offsets in Kafka are unique per partition, this maps partition identifiers (int) to an offset (long).
offsetsAsString (string)

This field contains a comma separated key-value list. It is available as a convenience for populating with offset information, as it provides a simple way to store the map with partitions and offset as a string. It reads and writes the same underlying data as the offsets (map<int,long>) field, so that changing the value of offsetsAsString (string) changes the value of offsets (map<int,long>) and vice versa.


An example of two partitions as string type: