Hi ES community,
Trying to setup a cluster with about 10 heterogenous blades at the moment, more to be added in the form of VMs and other blades. Have had non-cluster experience with 1.5.4, but mostly to setup proof of concepts for small scale demos.
Now the requirement is to use logstash-forwarder from 10 remote sites to redis+LS+ES. 20 people need Grafana and Kibana based dashboards.
Some clarifications:
- What should be the ideal characteristics of a
- Dedicated Master Node: (eg: high CPU, small disk size, high RAM)
- Dedicated Data Node: (eg: high speed disk, high RAM, low CPU)
- Dedicated Client Node: (eg: High CPU, RAM etc)
-
discovery.zen.ping.unicast.hosts: ["IP0", "IP1"]
Should this list contain the list of all IPs in the cluster? What happens when new hosts are added? Do all the configs need to be updated? -
ES2.0 yml config does not seem to have the config for shards and replicas. Is it not required?
index.number_of_shards: 5
index.number_of_replicas: 2 -
Is a replica only for fault tolerance, or does it speed up querying when many people are using the front end dashboards for customer queries?
Thanks in advance!