Docker-compose#

This guide explains how to run the Argilla server with Elasticsearch using docker-compose.

Launching Argilla and Elasticsearch with docker-compose#

For this method, you first need to install Docker Compose.

Then, create a folder:

mkdir argilla && cd argilla

and launch the docker-contained web app with the following command, after having set a value for the environment variable ARGILLA_AUTH_SECRET_KEY, which can be generated with openssl rand -hex 32:

wget -O docker-compose.yaml https://raw.githubusercontent.com/argilla-io/argilla/main/docker/docker-compose.yaml && docker-compose up -d

Warning

The latest versions of docker should be executed without the dash ‘-’, e.g.:

docker compose up -d

This is a convenient way because it automatically includes an Elasticsearch instance, Argilla’s main persistent layer.

Warning

Keep in mind that if you execute

docker-compose down

you will lose all your datasets in Argilla!

Tip

Instead, execute

docker-compose stop

Note

By default, telemetry is enabled. This helps us to improve our product. For more info about the metrics and disabling them check telemetry.

Persisting ElasticSearch data#

To avoid losing all the data when the docker-compose/server goes down, you can add some persistence by mounting a volume in the docker-compose.

To this end, modify the elasticsearch service and create a new volume in the docker-compose.yaml file:

services:
  elasticsearch:
    image: docker.elastic.co/elasticsearch/elasticsearch:8.5.3
    container_name: elasticsearch
    environment:
      - node.name=elasticsearch
      - cluster.name=es-argilla-local
      - discovery.type=single-node
      - "ES_JAVA_OPTS=-Xms512m -Xmx512m"
    ulimits:
      memlock:
        soft: -1
        hard: -1
    networks:
      - argilla
    # Add the volume to the elasticsearch service
    volumes:
      - elasticdata:/usr/share/elasticsearch/data
  argilla:
    # ... here goes the rest of the docker-compose.yaml

# ...

# At the end of the file create a volume for ElasticSearch
volumes:
  elasticdata:

Then, even if the ElasticSearch service goes down the data will be persisted in the elasticdata volume. To check it you can execute the command:

docker volume ls

Note that if you want to apply these changes, and you already have a previous docker-compose instance running, you need to execute the up command again:

docker-compose up -d