Samuel Cantero

Hello @dtheodor ! > If I am reading things correctly, when using FutureProducer::send, queue full errors will result in re-trying to enqueue, but the original order of sending is lost...

I was not able to reproduce this issue. FWIW I created a consumer that subscribed into a non-existing topic and then a producer instance wrote a single message to that...

> then will start a rebalance process for Consumer (this process will take almost 5 minutes), ah ok, so I'll need to try to repro this with a consumer group...

fwiw we're getting the same issue but only when fetch-from-replicas is enabled on rdkafka 1.8.2

@edenhill @emasab are there any chances to get feedback on this PR?

Hi @emasab! Thanks for checking this. We need a way to compute the assignment logic on the client side and as far as I understood KIP-848 supports that. > The...

Please let us know if you need further information. If you can also provide the estimate you have came across that would useful for us, thanks again @emasab.

@connormullett The example actually works and finishes as expected though it's indeed confusing we're getting bootstrap connection refused failures being logged (that might be a bug on librdkafka). But if...

> Does is make sense here to have a message to check configuration in case of these types of errors? mm I need to check but I think the latest...

Thanks for reporting it, we'll investigate further. `0.35.0` does not contain the event api change if you wanna give it a try. Can you provide some snippet of your code?...