We use beats to send most of our logs to logstash with the following parameters:
- bulk_size: 2048
- load_balanced: yes
- workers: number of logstashes
- pipelining: 2
In logstash we have 8 worker threads with 2 cores and 4GB HEAP. The question I have is does the internal batch sizing in logstash need to correlate with the bulk_size a beat might be sending? That is, does it need to be larger?