Elasticsearch-triple

Elasticsearch is an open-source distributed search and analytics engine used for log analytics. This package deploys a three-node Elasticsearch cluster on Kubernetes.

elasticsearch-triple is a high availability setup of elasticsearch, that sets up a 3-node cluster of elasticsearch for a robust and reliable setup.

Requirements

• Kubernetes >= 1.20.0
• Kustomize >= v3.3.X
• prometheus-operator

Prometheus Operator is necessary since we configure a ServiceMonitor to make some metrics available from elasticsearch on prometheus. Please refer, `elasticsearch-single to read about the available Prometheus rules.

Image repository and tag

• Elasticsearch image: elasticsearch/elasticsearch:7.16.2
• Elasticsearch repo: Elasticsearch on Github
• Elasticsearch documentation: Elasticsearch Homepage

Configuration

Fury distribution Elasticsearch Triple is deployed with the following configuration:

• Elasticsearch cluster with 3 nodes
• Listens on port 9200 for client connections
• Listens on port 9300 for node-to-node connections
• Uses default unicast Zen Discovery module to discover Elasticsearch nodes
• Resource limits are 2000m for CPU and 4G for memory
• Requires 30Gi storage
• Only a single Elasticsearch node can be deployed on each node of the cluster
• Prometheus exporter to expose Elasticsearch metrics
• Metrics are scraped by Prometheus every 30s

Deployment

You can deploy Elasticsearch Triple by running the following command in the root of the project:

kustomize build | kubectl apply -f -

To learn how to customize compute resources for Elasticsearch please follow the example.