Question about concurrentRequests in ElasticsearchWriter

[장석현 <tj...@gmail.com>]

Hi,

I'm currently working with the ElasticsearchSink class in the Flink Elasticsearch connector. I noticed that in the createBulkProcessor method, setConcurrentRequests(0) is used, which makes the flush() operation blocking.

From my understanding, it seems that even if we set setConcurrentRequests to a non-zero value, the use of pendingActions should prevent any data loss. Could someone please explain the rationale behind making setConcurrentRequests blocking by setting it to 0? Why is this necessary, and what are the benefits compared to using a non-zero value?

Thank you in advance for your help!