1. Documentation
2. 2025.0 (Latest) 2024.1 2024.0 10.3 10.2 10.1 10.0 9.10 9.9 9.8 9.7 9.6 9.5 9.4 9.3 9.2 9.1 9.0 8.2 8.1 8.0 7.6
3. Deployment
4. Docker-compose templates

Docker-compose templates

To deploy one of our docker-compose templates, click Download to download the template files, or select a link for additional details.

• [Download] for development and testing purposes, and for getting started quickly
• [Download] for generic production purposes
• [Download] for Deep Learning
• [Download] for production deployments that need easy Hadoop connectivity
• [Download] for production deployments that need low-latency scoring capability

Each template provides two files:

• The environment file .env
• The definition file docker-compose.yml

You can mix and match these templates if you need to produce a custom deployment, e.g. if you need a production deployment that has both Real-Time Scoring and Hadoop Connectivity capabilities.

See also: Services and locations provided by these templates.

Instructions for docker-compose deployments

To deploy one of these templates, take the following steps.

1. If you have not yet done so, install Docker on Linux / on Windows.

2. Choose one of the templates from the list above, and click Download to fetch the ZIP file. Unzip and examine the following two files:

   • .env (note that because of the preceding dot, this file is usually hidden)
   • docker-compose.yml
3. As discussed in step (7), set the variables PUBLIC_URL and SSO_PUBLIC_URL in the .env file.
4. Transfer these two files to a folder on the server host, the machine where you installed Docker.

5. Connect to the server host (ssh), and change directory (cd) to the folder containing those two files.

6. The template refers to an external Docker network (jupyterhub-user-net-$JUPYTER_STACK_NAME), that should be created before starting the deployment (if your planned deployment contains JupyterHub, which is true by default). The default stack name is default, so if it is not changed, then the network should be created using the following command:

    docker network create jupyterhub-user-net-default
   

7. The deployed stack needs to have a valid public URL setting, that will be used to connect to it using the external clients (like RapidMiner Studio and a browser) and also for internal communication. This URL should be set before the first startup to a valid HTTP URL using the PUBLIC_URL and SSO_PUBLIC_URL environment variables in the .env file.

   • Using http://localhost or http://127.0.0.1 is not supported, because this URL will be used also for internal container-to-container communication between our services.
   • If deploying on a single host, we prefer to use at least the host's public IP address, like http://192.168.1.101 or a publicly resolvable hostname that resolves to this IP address, like http://platform.rapidminer.com.

   • If the deployment cannot listen on the default HTTP and HTTPS ports (80 and 443), then

     • the port number should be also provided in the PUBLIC_URL and SSO_PUBLIC_URL variables, like http://platform.rapidminer.com:8080
     • the external port mapping should be set in the docker-compose.yml file at the rm-proxy-svc service definition, like 8080:80
     • the external ports should be set using the PROXY_HTTP_PORT and PROXY_HTTPS_PORT variables for the rm-server-svc in the docker-compose.yml file, like PROXY_HTTP_PORT=8080 and PROXY_HTTPS_PORT=8443. (If it is missed during the first startup, or needs to be changed, then later it should be configured in the /rapidminer-home/configuration/standalone.xml file at the proxy settings and requires restarting the service.)

   • It is highly preferred to use HTTPS for the connection. In this case the PUBLIC_URL and SSO_PUBLIC_URL variables should be configured using the https:// prefix and the certificate chain and private key files should be provided in PEM format in the ssl sub-folder using the filenames certificate.crt and private.key. The default filenames can be changed using the environment variables in the Proxy section of the .env file.

8. Set additional frequently used configuration values

   • The initial admin password can be set using the KEYCLOAK_PASSWORD variable
   • The AUTH_SECRET value is used as internal authentication encryption key. We propose to change the default value to any base64 encoded string.

9. If SSO configuration is not disabled (this is the case by default), then the platform deployment needs to be initialized before the first startup. This may take up to 1-2 minutes. Run the command:

    docker-compose up -d rm-init-svc
   

10. Finally, start the stack by running the command:

    docker-compose up -d
    

If the Docker images are not available on the host, they will be automatically downloaded from the Docker Hub.

Good to know

• Additional docker-compose commands for common deployment administration tasks are described in the technology overview.

• To scale up or down the number of RapidMiner Job Agents, you can use the Docker Deployment Manager or the following docker-compose command:

  docker-compose up --scale rm-server-job-agent-svc=5 -d
  

Services and locations

Once the deployment is running, the configured reverse proxy listens on the standard HTTP (80) port by default and if a HTTPS certificate is configured then on the HTTPS (443) port also. The following locations are available on the deployment public URL (depending on the list of the deployed services).

The initial login credentials are set in the .env file (KEYCLOAK_USER and KEYCLOAK_PASSWORD variables). By default you can login using the username "admin" and password: "changeit".

To learn more about the used technologies and how to operate and administer your platform deployment, see our technology overview page.