Nick Knize

I opened an issue to move this forward: https://github.com/opensearch-project/opensearch-hadoop/issues/304

> In the streaming mode, we would want to combine the action, metadata and the optional source into a single structure...However, it would add a breaking change to the bulk...

@reta do we want to create a meta issue and start documenting the client breaking changes to help drive a bwc / 3.0 release discussion for supporting this feature?

> Why does this feature entail a breaking change? There will need to be HTTP2 streaming OpenSearch APIs added the clients in order to leverage the streaming index API. This...

I also want to point out that we intend for this to be the transport replacement for both `_bulk` as well as the standard [`Put` document DSL API](https://opensearch.org/docs/1.2/opensearch/popular-api/#index-a-document-with-a-random-id) and Update....

won't be actively working unless needed

I suspect the surface area of this will be quite large. I'd be okay targeting 3.0.0?

Super excited about this!!! It's non-invasive which is fantastic! Enables us to further refactor the transport layer for extensions to support this. This has my full endorsement! Nice work!

> I am reading this proposal as 1) make transport pluggable _today_ in OpenSearch, Transport [is already pluggable today](https://github.com/opensearch-project/OpenSearch/blob/main/server/src/main/java/org/opensearch/plugins/NetworkPlugin.java#L58) (e.g., [plugins/transport-nio](https://github.com/opensearch-project/OpenSearch/tree/main/plugins/transport-nio)) > ...and 2) implement gRPC as an option. Sort...

Does the current implementation add thread pool tracking to the `_cat/thread_pool` API?