Skip to content

Latest commit

 

History

History
223 lines (156 loc) · 10.3 KB

README.md

File metadata and controls

223 lines (156 loc) · 10.3 KB

Official VOLTTRON docker image

Introduction

This image provides a reproducible way to install VOLTTRON within a docker container. By using a volume mount of the VOLTTRON_HOME directory, runtime changes made by the platform are visible on the host and are preserved across instances of the container. Similarly, changes made from the host are reflected in the container. The image uses the utility gosu, which allows the non-root user executing the volttron platform inside the container to have the same UID as the host user running the container on the host system. In conjunction with volume mounting of the directory, this ensures that file ownership and permissions in VOLTTRON_HOME match the host user, avoiding cases where root in the container leaves files inaccessible to a user without sudo permissions on the host.

Prerequisites

  • Docker ^20.10.8
  • Docker-compose ^3.4

If you need to install docker and/or docker-compose AND you are running this image on an Ubuntu machine, you can use the script in this repo. From the root level, execute the following command:

$ ./docker_install_ubuntu.sh

Quickstart:

To create the container and start using the platform on the container, run the following commands from the command line. Ensure that you are in the root level of the directory.

# Creates Volttron instance with ZMQ message bus
$ docker-compose up

# To ssh into the container
$ docker exec -itu volttron volttron1 bash

# To stop the container
$ docker-compose stop

# To start the container after it's been stopped
$ docker-compose start

# To get a list of all containers created from docker-compose
$ docker-compose ps

# To stop and remove the container
$ docker-compose down

For Volttron instances using ZMQ message bus:

  • Set the master username and password on the Volttron Central Admin page at https://0.0.0.0:8443/index.html
  • To log in to Volttron Central, open a browser and login to the Volttron web interface: https://0.0.0.0:8443/vc/index.html

Platform Initialization

The VOLTTRON container when created is just a blank container with no agents. Now there is an initialization routine available within the docker container to allow the installation of agents before launching of the instance. To do this one will mount a platform_config.yml file to /platform_config.yml within the container. One is also likely to need to mount agent configurations (specified in the platform_config.yml file), into the container. The recommended way to do this is through a docker-compose.yml file. An example of this is included in this repository, based on the one in the volttron-fuel-cells repo.

The platform_config.yml file has two sections: config, which configures the main instance and populate's the main config file ($VOLTTRON_HOME/config), and agents, which contains a list of agents with references to configurations for them (note the frequent use of environment variables in this section).

Main Configuration

The main instance configuration is composed of key value pairs under a "config" key in the platform_config.yml file. For example, the vip-address and bind-web-address would be populated using the following partial file:

# Properties to be added to the root config file:
# - the properties should be ingestable for volttron
# - the values will be presented in the config file
#   as key=value
config:
  vip-address: tcp://0.0.0.0:22916
  bind-web-address: https://0.0.0.0:8443
  # volttron-central-address: a different address
  # volttron-central-serverkey: a different key

  ...

Agent Configuration

The agent configuration section is under a top-level key called "agents". This top-level key contains several layers of nested key-value mappings. The top level of the section is keyed with the names of the desired agents, which are used as the identity of those agents within the platform. For each agent key, there is a further mapping which must contain a source key and may contain either or both a config and/or config_store key; the values are strings representing resolvable paths. An example follows at the end of this section.

Note that the agent section does not contain the detailed configuration of the agents; for each agent it gives a path to a dedicated configuration file for that agent. As with the platform_config.yaml file, it is generally desirable to mount a local directory containing the configurations into the container, again using a docker-compose.yaml file.

...

# Agents dictionary to install.  The key must be a valid
# identity for the agent to be installed correctly.
agents:

  # Each agent identity.config file should be in the configs
  # directory and will be used to install the agent.
  listener:
    source: $VOLTTRON_ROOT/examples/ListenerAgent
    config: $CONFIG/listener.config

  platform.actuator:
    source: $VOLTTRON_ROOT/services/core/ActuatorAgent

  historian:
    source: $VOLTTRON_ROOT/services/core/SQLHistorian
    config: $CONFIG/historian.config

  weather:
    source: $VOLTTRON_ROOT/examples/DataPublisher
    config: $CONFIG/weather.config

  price:
    source: $VOLTTRON_ROOT/examples/DataPublisher
    config: $CONFIG/price.config

  platform.driver:
    source: $VOLTTRON_ROOT/services/core/MasterDriverAgent
    config_store:
      fake.csv:
        file: $VOLTTRON_ROOT/examples/configurations/drivers/fake.csv
        type: --csv
      devices/fake-campus/fake-building/fake-device:
        file: $VOLTTRON_ROOT/examples/configurations/drivers/fake.config

Other Notes

Agents within the platform_config.yml file are created sequentially, it can take several seconds for each to spin up and be visible via vctl commands.

Development

If you plan on extending or developing the "platform_config.yml", "configs/", or the setup scripts in "core/", build the Docker image, "Dockerfile-dev", only once using docker-compose -f docker-compose-dev.yml build --no-cache volttron1. Then start the container using docker-compose -f docker-compose-dev.yml up. When you want to make changes to "platform_config.yml", "configs/", or "core/", simply make the changes and then rerun your container. You do not have to rebuild the image every time you make changes to those aforementioned files and folders because they are mounted into the container. The only time you should rebuild the image is when you make changes to the "volttron" source code since that is not mounted to the container but rather baked into the image during the image build. Once you are satisfied your changes, update 'Dockerfile' with the changes you used in 'Dockerfile-dev' and submit a PR.

To setup your environment for development, do the following:

  1. Give execute permissions for ./core/*
chmod a+x core/*
  1. Pull in volttron from the official volttron repo using the following git command:
# Clones https://github.com/VOLTTRON/volttron.git into the 'volttron' directory
git submodule update --init --recursive

Why are we doing this? This repo has a directory called 'volttron', which contains the volttron codebase. In other words, this repo contains another repo in a subfolder. When you initially clone this repo, the 'volttron' directory is empty. This directory contains the volttron codebase used to create the volttron platform.

OPTIONAL: This repo uses a specific version of volttron based on the commit in the 'volttron' submodule. If you want to use the latest volttron from the develop branch from the volttron repo, execute the following command (NOTE: this is not required):

# Ensure that you are in the `volttron` folder
git pull origin develop
  1. Build the image locally:
  • Using docker-compose (preferred)
docker-compose -f docker-compose-dev.yml build --no-cache --force-rm
  1. Run the container:
  • Using docker-compose (preferred)
docker-compose -f docker-compose-dev.yml up

Testing

If you want to work on improving/developing the Dockerfile, you can locally run a test script to check whether the image works as expected. To run the test, see the following:

# run the test (rebuilds and tests the most current image)
$ ./run-test-docker-image.sh

# You can also run the test but skip rebuilding the image
$ ./run-test-docker-image.sh -s

Note: If you want to use a different image name and/or tag, you must ensure that the image name in docker-compose.yml matches the image name given to the integration test script. For example, if you run the integration tests with the following options ./run-test-docker-image.sh -g test -t integ, then the 'image' key in docker-compose.yml must be set to 'volttron/test:integ'.

Updating the image on Dockerhub

If you are not part of the Volttron Core development team, you can skip this section.

See: https://confluence.pnnl.gov/confluence/display/VNATION/Docker+Image+Publishing+Procedures

Advanced Usage

In order for volttron to keep its state between runs, the state must be stored on the host. We have attempted to make this as painless as possible, by using gosu to map the hosts UID onto the containers volttron user. The following will create a directory to be written to during VOLTTRON execution.

  1. Create a directory (eg mkdir -p ~/vhome). This is where the VOLTTRON_HOME inside the container will be created on the host.
  2. Start the docker container with a volume mount point and pass a LOCAL_USER_ID environmental variable.
    docker run -e LOCAL_USER_ID=$UID -v /home/user/vhome:/home/volttron/.volttron -it volttron/volttron

In order to allow an external instance connect to the running volttron container one must add the -p : (e.g. 22916:22916)

Troubleshooting

My VC Platform agent can't connect to the Volttron Central address. I see volttron.platform.vip.agent.subsystems.auth ERROR: Couldn't connect to https://localhost:8443 or incorrect response returned response was None in the logs

This most likely occurs if you are deploying this container behind a proxy. Ensure that your ~/.docker/config.json has no "proxies" configuration.

My Forwarder shows a BAD status when I run vctl status

Ensure that the configuration for your forwarder is using the same volttron-central-address property in volttron config, which is set in your platform_config.yml file.