Graylog Performance

[cs.xdumas (kaiser)]

Hi,

I have graylog 2.3 installed on red hat
I have elasticsearch 4 shards no replicas 2.0
I have mongodb 3.0

My machine: 64Go RAM
12 processors CPU

Elasticsearch heap: 24go
Java heap: 15Go

Everything is installed on a single machine.

My problem is that all memory is not consumed and there is always 47 Go memory unused

I can only get 200msg/s before graylog buffers the messages

Could you tell me if there is something I could do to make graylog use all memory?

Regards,

Kaiser

[jochen]

FWIW, it’s good that the Java processes of Graylog and Elasticsearch are not using all available memory. The operating system is supposed to use “free” memory for the disk buffers and caches (leading to higher I/O throughput).

4 GB of heap memory for Graylog should be sufficient.