Hi! We're using flink and its elasticsearch (ES) sink to process content changes at wikimedia. The connector uses ES' bulk API but is rather strict when it comes to interpreting a response: It fails if a single action of the bulk has failed. This behavior is not configurable at the moment so I would like to propose an improvement, to overcome this limitation:
Ticket: https://issues.apache.org/jira/browse/FLINK-32028 Pull-Request: https://github.com/apache/flink-connector-elasticsearch/pull/83 I'm looking forward to getting feedback on this proposal. Peter