Search code examples
dockerelasticsearchhealth-check

ElasticSearch Healthcheck on docker-compose failing

Elasticsearch healthcheck on docker-compose stops any dependent services because the container is always unhealthy. I see this when I run docker ps -a --format "table {{.Names}}\t{{.Image}}\t{{.Status}}"

NAMES           IMAGE                  STATUS
elasticsearch   elasticsearch:7.12.1   Up 26 seconds (unhealthy)

I am attempting to start metricbeat such that elasticsearch, kibana and logstash are started:

metricbeat:
  image: elastic/metricbeat:7.12.1
  user: root
  depends_on:
    elasticsearch:
      condition: service_healthy
    kibana:
      condition: service_healthy
    logstash:
      condition: service_healthy
    redis:
      condition: service_healthy

How can I ensure that the elasticsearch (and other containers are healthy) and allow for metricbeat to start with all the resources available?

I would avoid creating a Docker image for any of these unless absolutely required.

My docker-compose configuration looks like this:

version: '3.7'
services:
  elasticsearch:
    # specifying discovery.type='single-node' bypasses bootstrapping checks.
    image: elasticsearch:7.12.1
    container_name: elasticsearch
    healthcheck:
      test: [ "CMD", "curl",  "--fail" , "http://elasticsearch:9200/_cluster/health?wait_for_status=green&timeout=1s", "||", "exit", "1" ]
      interval: 5s
      timeout: 3s
          
    volumes:
      - elasticsearch-data:/usr/share/elasticsearch/data
    networks:
      - elastic
    ports:
      - 9200:9200
      - 9300:9300
    labels:
      co.elastic.metrics/module: "elasticsearch"
      co.elastic.metrics/hosts: "http://elasticsearch:9200"
      co.elastic.metrics/metricsets: "node_stats,node"
      co.elastic.metrics/xpack.enabled: "true"
    environment:
      - node.name=elasticsearch
      - cluster.name=cluster-7
      - discovery.type=single-node
      - 'ES_JAVA_OPTS=-Xms512m -Xmx512m'
      - xpack.monitoring.enabled=true
      - xpack.monitoring.elasticsearch.collection.enabled=true
      
    ulimits:
      memlock:
        soft: -1
        hard: -1
      nofile:
        soft: 65536
        hard: 65536
    cap_add:
      - IPC_LOCK
Solution

  • The approach can be different depending on elastic version. Also you need to use something else if you use OpenSearch they have bit different output for health (but just for info)

    I`m adding what I use in my docker-compose

    healthcheck:
  interval: 10s
  retries: 80
  test: curl --write-out 'HTTP %{http_code}' --fail --silent --output /dev/null http://localhost:8200/

    or

    healthcheck:
   test: curl -s http://elasticsearch01:9200 >/dev/null || exit 1
   interval: 30s
   timeout: 10s
   retries: 50