An example Docker Compose project to run FlowForge

FlowForge uses Docker Compose to install and manager the required components. Instructions on how to install Docker Compose on your system can be found here:

https://docs.docker.com/compose/install/

These instructions assume you are running Docker on a Linux or MacOS host system.

To access the Projects created you will need to set up a wildcard DNS entry that points to the domain entered in the etc/flowforge.yml file.

e.g. assuming that Docker is running on a machine with IP address 192.168.0.8 then an A record point to *.example.com.

This will mean that any host at example.com will resolve to the 192.168.0.8.

Note When testing locally you can not use the loopback address 127.0.0.1 for this, e.g. in the /etc/hosts file, as this will resolve to the TCP/IP stack inside each container.

To build the 2 required containers simply run ./build-containers.sh.

This will build and tag both flowforge/forge-docker and flowforge/node-red and flowforge/file-server.

This container holds the FlowForge App and the Docker Driver.

This is a basic Node-RED image with the FlowForge Launcher and the required Node-RED plugins to talk to the FlowForge Platform.

This is the container you can customize for your deployment.

This holds the Object Store used to allow persistent file storage for Projects running on Docker

Configuration details are stored in the etc/flowforge.yml file which is mapped into the flowforge/forge-docker container. You will need to edit this file to update the domain and base_url entries to match the DNS settings.

You also need to update the VIRTUAL_HOST entry in the docker-compose.yml file to use the same domain as in the etc/flowforge.yml file.

You should also update the email section to point to a local SMTP server so you can invite users to join teams.

Once the container have been built you can start the FlowForge by running:

docker-compose -p flowforge up -d

This will also create a directory called db to hold the database files used to store project instance and user information.

If upgrading from version before version 1.2.0 you will need to manually create the database for the persistent context store.

To do this you will need to run the following command after starting:

docker exec -it flowforge_postgres_1 /docker-entrypoint-initdb.d/setup-context-db.sh

This is experimental

If you are actively developing FlowForge, the following instructions can be used to run it with the Docker driver using a locally mounted source tree.

1. Ensure that you have all of the FlowForge source repositories checked out next to each other - including this repository.

2. Run npm install in each repository that has a package.json file.

3. In the flowforge repo, run npm run dev:local to setup proper dev symlinks between the repos.

4. Follow the instructions above to setup DNS.

5. Edit the etc/flowforge.yml file in the flowforge repository to use the docker driver:

port: 3000
host: 0.0.0.0
domain: example.com
base_url: http://forge.example.com
api_url: http://forge:3000

driver:
  type: docker
  options:
    socket: /tmp/docker.sock

6. Depending on what OS you are running on, the core project has one binary dependency that needs to be rebuilt for it to work inside Docker - bcrypt. The super hacky way to get that to work is to edit flowforge/package.json and modify the serve task to first reinstall that module:

"serve": "npm uninstall bcrypt && npm install bcrypt && npm-run-all --parallel build-watch start-watch"

You only need to do this the first time you run under docker - you can then revert that change for the subsequent runs.

Note: You will need to reinstall the module when you go back to running outside of docker.

7. Start the platform with: docker-compose -f docker-compose-local-dev.yml up --build

That will start the standard environment, but the forge container will have the local source tree mounted, and use npm run serve to start the code. This means it will automatically rebuild/restart whenever source code changes are made.