Andriy Redko

@DanielThomas could you please share if you tune the`messageTimeout` setting in your application [1] or the default value (1s) is being used? Thank you. [1] https://github.com/openzipkin/zipkin-reporter-java?tab=readme-ov-file#tuning

Yeah, I am looking into the way to address the root cause, meantime (as I workaround, not a solution), could you try please to lower the `messageTimeout` so the lock...

> There may be other examples in code we don't control too. Given that, I'm rather weary of side effects of changing to using a `ReentrantLock` for this code. Thanks...

> So I think unless someone has a proposal for concrete changes to consider to avoid the potential deadlock prior to Java 24 when using virtual threads, we can close...

@snuyanzin whenever you have time to look, please

> I left some comments Thanks a lot, @snuyanzin ! Addressed all except `spotless` version bump (I think we should do it in flink-connector-parent), thank you!

@snuyanzin if you don't mind to glance through, thank you

@andrross thanks a mill for approving, conflicts resolved + more tests added, could you please reapprove? thank you!

Thanks @shwetathareja , I would side with @Bukhtawar here and the (early) conclusions from https://github.com/opensearch-project/OpenSearch/issues/3035 as @andrross mentioned. We already use some versioning as part of content type (fe `application/vnd.opensearch+json;compatible-with=7`),...

> While header solves the problem of not modifying the request/response structure (directly), it is not very apparent from the user perspective. Yes, but this is the approach OS (and...