Install Inmanta with Docker¶

This page explains how to setup an orchestration server using docker. This guide assumes you already have docker and docker-compose installed on your machine.

Pull the image¶

Use docker pull to get the desired image:

docker pull ghcr.io/inmanta/orchestrator:latest

This command will pull the latest version of the Inmanta OSS Orchestrator image.

Start the server with docker-compose¶

Here is a minimalistic docker-compose file content that can be used to deploy the server on your machine.

version: '3'
services:
    postgres:
        container_name: inmanta_db
        image: postgres:13
        environment:
            POSTGRES_USER: inmanta
            POSTGRES_PASSWORD: inmanta
            PGDATA: /var/lib/postgresql/data/pgdata
        networks:
            inm_net:
                ipv4_address: 172.30.0.2
        volumes:
            - type: volume
              source: pgdata
              target: /var/lib/postgresql/data
    inmanta-server:
        container_name: inmanta_orchestrator
        image: ghcr.io/inmanta/orchestrator:latest
        ports:
            - 8888:8888
        networks:
            inm_net:
                ipv4_address: 172.30.0.3
        depends_on:
            - "postgres"
        command: "server --wait-for-host inmanta_db --wait-for-port 5432"

networks:
    inm_net:
        ipam:
            driver: default
            config:
                - subnet: 172.30.0.0/16
volumes:
    pgdata:

docker-compose up

You should be able to reach the orchestrator to this address: http://172.30.0.3:8888.

The PostgreSQL server started by the above-mentioned docker-compose file has a named volume pgdata attached. This means that no data will be lost when the PostgreSQL container restarts. Pass the -v option to the docker-compose down to remove the volume.

The default server config included in the container images assumes that the orchestrator can reach a database server with hostname inmanta_db and that it can authenticate to it using the username inmanta and password inmanta. When using a different setup than the one mentioned above, you should overwrite the server config with one matching your needs. You can find more instructions for overwriting the config in a following section, here.

Warning

We don’t recommend using the setup described above as a production environment. Hosting a database in a container as shown here is not ideal in term of performance, reliability and raises some serious data persistence concerns.

Overwrite default server configuration¶

By default the server will use the file located in the image at /etc/inmanta/inmanta.cfg. If you want to change it, you can copy this file, edit it, then mount it in the container, where the original file was located.

If you use docker-compose, you can simply update this section of the example above:

inmanta-server:
    container_name: inmanta_orchestrator
    image: ghcr.io/inmanta/orchestrator:latest
    ports:
        - 8888:8888
    volumes:
        - ./resources/my-server-conf.cfg:/etc/inmanta/inmanta.cfg

Starting the ssh server¶

By default, no ssh server is running in the container. You don’t need it to have a functional orchestrator. If you want to enable ssh anyway, for easy access to the orchestrator, you can overwrite the startup command of the container with the following:

server-with-ssh

If you use docker-compose, it should look like:

inmanta-server:
    container_name: inmanta_orchestrator
    ...
    command: "server-with-ssh"

Warning

By default, the inmanta user doesn’t have any password, if you want to ssh into the container, you also need to set the authorized_keys file for the inmanta user. You can do so by mounting your public key to the following path in the container: /var/lib/inmanta/.ssh/authorized_keys. When starting, the container will make sure that the file has the correct ownership and permissions.

Waiting for the database¶

Depending on you setup, you might want your container to wait for the database to be ready to accept connections before starting the server (as this one would fail, trying to reach the db). You can do this by adding the following arguments to the startup command of the container:

server --wait-for-host <your-db-host> --wait-for-port <your-db-port>

If you use docker-compose, it should look like:

inmanta-server:
    container_name: inmanta_orchestrator
    ...
    command: "server --wait-for-host <your-db-host> --wait-for-port <your-db-port>"

Setting environment variables¶

You might want your inmanta server to be able to reach some environment variables. There are two ways you can achieve this:

Set the environment variables with docker, either using the --env argument or in your docker-compose file. Those variables will be accessible to the inmanta server and any agent it starts, but not to any other process running in the container (if you for example login via ssh to the container and try to install a project again).

(Recommended) Set the environment variables in a file and mount it to the following path in the container: /etc/inmanta/env. This file will be loaded when starting the server and for every session that the inmanta user starts in the container.

inmanta-server:
    container_name: inmanta_orchestrator
    image: ghcr.io/inmanta/orchestrator:latest
    ports:
        - 8888:8888
    volumes:
        - ./resources/my-server-conf.cfg:/etc/inmanta/inmanta.cfg
        - ./resources/my-env-file:/etc/inmanta/env

Changing inmanta user/group id¶

If you mount a folder of your host in the container, and expect the inmanta user to interact with it, you might face the issue that the inmanta user inside the container doesn’t have ownership of the files. You could fix this by changing the ownership in the container, but this change would also be reflected on the host, meaning that you would lose the ownership of you files. This is a very uncomfortable situation. While Podman has been offering the possibility to do a mapping of a user id in the container to a user id on the host at runtime, which would solve our problem here, Docker still doesn’t offer this functionality. The inmanta container allows you to change the user and group id of the inmanta user inside the container when starting the container to match the user on the host, getting rid that way of any conflict in the files ownership.

This can be done easily by simply setting those environment variables:

INMANTA_UID: Will change, when starting the container, the id of the inmanta user.
INMANTA_GID: Will change, when starting the container, the id of the inmanta group.

If you use docker-compose, you can simply update this section of the example above:

inmanta-server:
    container_name: inmanta_orchestrator
    ...
    environment:
        INMANTA_UID: 1000
        INMANTA_GID: 1000

Log rotation¶

By default, the container won’t do any log rotation, to let you the choice of dealing with the logs according to your own preferences. We recommend that you do so by mounting a folder inside of the container at the following path: /var/log. This path contains all the logs of inmanta (unless you specified a different path in the config of the server) and the logs of the SSH server.