Thanks for the clarification. I was handed this configuration originally where page cache was a fraction of the heaps. False assumption on my part that page cache was a subset of heap size. I really appreciate this help. Testing these updates.
So, the page cache can be larger than initial/max heap. The total of the two should be less than the total available RAM on the system? Is that an accurate statement?
There is nothing else running on the servers except for some small logging processe...
I could not find a good explaination of what page cache size should be to be honest. We recently scaled our neo4j hardware and have been tinkering with some of these settings to try to get better performance.
# du -hc *graph.db* | tail
28K graph.db/s...
We are still on 3.4 b/c we haven't been able to focus on upgrading yet; however, i'm hoping to do that soonish. I would like to go to 3.5 first perhaps, but i'm not sure on which upgrade paths are best supported from our current version.
I did read t...
Thanks @dana.canzano. We are on AWS with an m4.4xlarge EC2 instance type with 64 GB of RAM.
Docker Compose details:
# /usr/local/bin/docker-compose -f /etc/neo4j/neo4j.yaml config
services:
neo4j-server:
container_name: neo4j
environment:
...
