Mitsunori Komatsu

I see. I got the point. Yes. It sounds a good improvement. I would try it later.

Sorry, I may have read your comment wrongly. Setting BufferChunkRetentionSize to less than the buffer limit of Fluentd / Fluent Bit should work. I'll update Fluency README to mention the...

Ah, I remember that indeed there is room to improve to make it more strict. Let me try.

@frosiere Even with `setRecordFormatter()`, it looks still a bit complicated to inject Jackson modules. And I have another concern that a wrong RecordFormatter can be passed (e.g. passing JsonlRecordFormatter to...

Yeah, `setRecordFormatter()` is more powerful. I'll go with it. I'm a bit busy these days and it would take some time until the feature is released.

Hi @sokomishalov, Could you give me an instruction including configurations to reproduce this issue?

I tried to reproduce this issue, but didn't. I changed Fluent Bit's buffer size down to 4KB since no error happened with the default 32KB. Fluency test code ``` FluencyBuilderForFluentd...

Looks like the server side (Fluentd) sent a TCP RST packet. I recommend you to monitor TCP packets between Fluentd and Fluency to isolate the cause.

That looks a good workaround. The difficult point to take care of this issue is we need to build Fluency code with JDK 8 to create jar files to avoid...

@tmeinhart How about trying Android's desugaring feature https://developer.android.com/studio/write/java8-support#library-desugaring ?