Overview of OpenSearch

OpenSearch is an open-source, distributed search and analytics suite derived from a fork of Elasticsearch. It supports use cases such as full-text search, log and infrastructure monitoring, security analytics, and operational observability. The project is maintained under the Apache License 2.0 and hosted publicly on GitHub, with active stewardship from AWS.

Pulling Official Images

# Retrieve the latest OpenSearch server image
docker pull opensearchproject/opensearch:latest

# Retrieve the corresponding OpenSearch Dashboards image
docker pull opensearchproject/opensearch-dashboards:latest

Launching a Standalone Node

For local development or evaluation, a single-node cluster can be started with minimal configuration:

docker run \
  --name os-standalone \
  -d \
  -p 9200:9200 \
  -p 9600:9600 \
  -e "discovery.type=single-node" \
  -e "network.host=0.0.0.0" \
  -e "plugins.security.disabled=false" \
  opensearchproject/opensearch:latest

Key environment variables:

• discovery.type=single-node: Enables single-node mode without cluster coordination.
• network.host=0.0.0.0: Binds to all interfaces for external acces.
• plugins.security.disabled=false: Activates the Security plugin (default credentials: admin:admin).

Verifying Connectivity

After startup, confirm the node is responsive using TLS-authenticated curl:

curl -k -u 'admin:admin' https://localhost:9200

The -k flag bypasses certificate validation — appropriate to self-signed dev certificates.

Running OpenSearch Dashboards

Dashboards connect to the OpenSearch backend via HTTPS or HTTP depending on security settings:

# With security enabled (HTTPS + auth)
docker run \
  --name os-dash \
  -d \
  -p 5601:5601 \
  -e "server.host=0.0.0.0" \
  -e 'OPENSEARCH_HOSTS=["https://host.docker.internal:9200"]' \
  opensearchproject/opensearch-dashboards:latest

# With security disabled (HTTP only, no auth)
docker run \
  --name os-dash-insecure \
  -d \
  -p 5601:5601 \
  -e "server.host=0.0.0.0" \
  -e 'OPENSEARCH_HOSTS=["http://host.docker.internal:9200"]' \
  -e "OPENSEARCH_DASHBOARDS_SECURITY_ENABLED=false" \
  opensearchproject/opensearch-dashboards:latest

Note: Use host.docker.internal for host resolution from containers on Docker Desktop. On Linux, replace it with the host’s actual IP or configure DNS accordingly.

Multi-Node Cluster Using Docker Compose

A minimal two-node cluster with Dashboards can be orchestrated via docker-compose.yml:

version: '3.8'

services:
  opensearch-01:
    image: opensearchproject/opensearch:latest
    container_name: opensearch-01
    environment:
      - cluster.name=search-cluster
      - node.name=os-node-1
      - discovery.seed_hosts=opensearch-01,opensearch-02
      - cluster.initial_master_nodes=opensearch-01,opensearch-02
      - bootstrap.memory_lock=true
      - "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m"
    ulimits:
      memlock:
        soft: -1
        hard: -1
    ports:
      - "9200:9200"
      - "9600:9600"

  opensearch-02:
    image: opensearchproject/opensearch:latest
    container_name: opensearch-02
    environment:
      - cluster.name=search-cluster
      - node.name=os-node-2
      - discovery.seed_hosts=opensearch-01,opensearch-02
      - cluster.initial_master_nodes=opensearch-01,opensearch-02
      - bootstrap.memory_lock=true
      - "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m"
    ulimits:
      memlock:
        soft: -1
        hard: -1

  dashboards:
    image: opensearchproject/opensearch-dashboards:latest
    container_name: os-dashboards
    ports:
      - "5601:5601"
    environment:
      - "server.host=0.0.0.0"
      - 'OPENSEARCH_HOSTS=["https://opensearch-01:9200"]'
      - "OPENSEARCH_USERNAME=admin"
      - "OPENSEARCH_PASSWORD=admin"
    depends_on:
      - opensearch-01
      - opensearch-02

This configuration uses internal service names (opensearch-01, opensearch-02) for inter-container communication and assumes TLS is configured or disabled appropriately in production environments.

Observability Considerations

Monitoring OpenSearch containers involves tracking JVM metrics (heap usage, GC frequency), OS-level resource consumption (CPU, memory), and OpenSearch-specific health endpoints (/_cluster/health, /_nodes/stats). For Kubernetes deployments, leverage Prometheus exporters like the OpenSearch Prometheus Exporter alongside standard tooling such as kube-state-metrics and cAdvisor.