How to speed up search in elasticsearch through kibana 4?

Question

elasticsearch

Set up ELK.
A month has passed and kibana began to update the data on duty, sometimes it doesn’t update at all, it transfers to the status page where it shows that the status of elasticsearch is red.
I understood and could not understand what was the matter. I gave service elasticsearch restart and immediately everything ran fine. Where to look for a problem?
Before the restart, there was a little less cpu idle than now, and LA was increased, but everything was within the limits, LA was around 2 with 16 cores.

Answer 1

Elastic is a Java process, how much RAM is allocated to it personally?
status elasticsearch red - What is written in elasticsearch logs?
Kibana is just a webmorda, logstash is just an aggregator. All data is in Elastica, here are its settings and parameters and you need to look, dig to understand.
TOTAL:
Increase elastic memory.

The question is, why is there enough memory at first, and then not anymore, where does it go?
Can you tell me how to squander this parameter (I mean the amount of allocated memory), or give a keyword for Google?

Well, at first, you have little data, so there is enough memory. Then, when there are more of them, it seems to stop being enough))))
It's easy to waste, but it depends slightly on the operating system.
On CentOS, you need to fix the config file /etc/sysconfig/elasticsearch
ES_HEAP_SIZE parameter

# Heap Size (defaults to 256m min, 1g max)
ES_HEAP_SIZE=30g

In other Axes by analogy.
Give it 50% of RAM, i.e. in this case - 16G