Daniel Shmuglin

@oleksiyk So I [commented out](https://github.com/oleksiyk/kafka/blob/master/lib/connection.js#L201) your printout, plus added `data` argument to it. Obviously I get a huge blob of binary data, which doesn't look nice in log. However I...

Regardless of my findings to come, if any, WDYT about wrapping [`this.queue[correlationId].resolve(new Buffer(data.slice(4, length + 4)));`]( https://github.com/oleksiyk/kafka/blob/master/lib/connection.js#L205) with `try...catch...report` meanwhile to keep node process from dying disgracefully?

@oleksiyk here you go. Recorded four messages. [kafka_messages.tar.gz](https://github.com/oleksiyk/kafka/files/1280062/kafka_messages.tar.gz)

@oleksiyk did you have a chance to take a look into my dumps? Anything interesting? I didn't have much progress. My plan is to add additional copy of `queue` with...

1. I don't have logs enabled (or maybe I've missed it) 2. Pretty easy. So I have a few instances of exactly the same micro-service (same groupId, same consumer). Once...

Those are docker. `start` and `stop` will do. I think I've seen it both in those that did the restart and those that didn't move.

Three brokers in kafka cluster. Single topic that is

@oleksiyk so I added another queue to connection to try and catch duplicate responses for the same correlation id (see [here](https://github.com/hugebdu/kafka/commit/998310c1438df9b610924b42b8b33588f179a686)) [Here](https://gist.github.com/hugebdu/9e0564e684c2c8603b5db917049df80f) is the stderr.log Look for `Wrong correlationId received`...

Nope

As I mentioned before, we run our node processes in docker and we do suspect some networking problems there (planning to upgrade to latest docker - there were many network...