Preparing the Search Cluster

Info

Always use a fresh server. New Elasticsearch install if performed on a server that has old Elasticsearch installations, does not update all the config files (https://caringo.atlassian.net/wiki/spaces/public/pages/2443809957/Configuring+Elasticsearch#Scripted-Configuration). Hence, some leftover config files remain not updated with the new Elasticsearch install.

Perform the following steps to prepare the search servers for Elasticsearch:

Verify the servers against the https://perifery.atlassian.net/wiki/spaces/public/pages/2443809601.
Appropriately cable the servers to your network infrastructure so they are reachable from the Swarm nodes.

Important

Make sure that Elasticsearch nodes are not accessible externally; only the Swarm nodes and Gateway servers should be able to access Elasticsearch.

Install RHEL/CentOS 7 Linux and apply any required updates. Contact DataCore Support for questions about enabling or disabling IPv6.

Best Practice

Use RHEL/CentOS minimal server with compatibility libraries which is standard for Swarm development and testing. RHEL/CentOS desktop consumes extra resources that Elasticsearch uses and alters the OS configuration to emphasize user interface vs server performance. It also requires additional updating and security maintenance.

Configure the servers with static IP addresses.
Configure DNS, if desired.
Adjust the server firewall rules. See firewalld.org.
Adjust the rules to permit the following ports if on CentOS 7 install and run iptables:
- Allow public access on these ports:
  firewall-cmd --permanent --zone=public --add-port=9200/tcp firewall-cmd --permanent --zone=public --add-port=9300/tcp firewall-cmd --reload
- Search the Support portal for SwarmNFS 2.x - Access to Elasticsearch (IPTables) for SwarmFS access.