based on those stats I would recommend the following:
- Run one master/data node per host. It seems you are currently 2 nodes per host.
- Assuming nothing else is running on the Elasticsearch hosts, set heap to 50% of the available RAM.
- You currently appear to have very small shards. If this is volumes you expect going forward, make sure to reduce the number of primary shards per index and consider switching from daily to e.g. weekly or monthly indices. See this blog post for further guidance.