David Butler

Interesting... I just tried this test under JRuby and got ```E, [2017-05-13T15:12:57.713000 #17281] ERROR -- : [LogStashLogger::Device::UDP] SystemCallError - Unknown error - No message available```. So it's not caught as...

I can't reproduce this behavior in MRI 2.3.3. I suspect the behavior might be platform specific. What OS are you using? I test on a Macbook Pro.

The test above writes multiple large messages. If you just write one message, you'll see that it fails and never retries again. The `unrecoverable error, aborting write` warning is the...

I was not able to reproduce this behavior locally. I'm running logstash 1.50. What version of logstash are you running and what is your listener configuration?

My reading of the [LogStash TCP input source code](https://github.com/logstash-plugins/logstash-input-tcp/blob/fe0d590f7e729654c1e6782c1ecbf92f1778a314/lib/logstash/inputs/tcp.rb#L210-L212) indicates that it will read up to 16384 bytes. So if you're using the latest version of logstash, it probably isn't...

[caller_locations](http://www.rubydoc.info/stdlib/core/2.0.0/Kernel%3Acaller_locations) looks like a good place to start. I don't imagine this information would be useful to add to every log message by default. I can only imagine it being...

Hi, Currently there is no HTTP device for LogStashLogger. If you want to write one, I would accept a pull request.

@brycemcd Thanks! I opened an issue with Ruby-Kafka to find out why they are so restrictive with the Ruby version.

Creating a new type of device is a good approach, since the configuration interface is different and Ruby-Kafka also doesn't support older versions of Ruby. The Poseidon device can be...

Also, I would prefer to name the new device `kafka` and the old one `poseidon` for clarity. It's a breaking change but easily documented in the readme.