Hi! We encountered a bug in elasticsearch recently that prevents us from using the flink connectors for opensearch|elaticsearch: size estimations for bulk requests ignore potentially large parameter values, see [1]. As a workaround, we patched the connector to expose the flush method of the underlying BulkProcessor, as suggested in FLINK-33857, see [2], [3]. So far that's working reliably by allowing us to write our own flush trigger and I would appreciate feedback on our proposed workaround.
[1] https://github.com/elastic/elasticsearch/issues/103406 [2] https://issues.apache.org/jira/browse/FLINK-33857 [3] https://github.com/apache/flink-connector-elasticsearch/pull/85 Best regards, Peter
