Options

Configuration of the server can be done either through the command line or by creating a custom YAML configuration file. Both options will be detailed below.

🧙 Configure using the Wizard

The most straight forward way of creating a new server configuration is using the command vnode new which allows you to configure the most basic settings.

By default the node configuration file is stored at user level, which isolates this configuration from other users. In case you want this configuration to be available for all users, simply add the --system flag (make sure you have sufficient rights to do this).

👩🔬 Configure using a custom YAML file

The configuration wizard outputs a YAML file which can be loaded into VANTAGE6. It is also possible to create this file yourself, to see Configuration File Structure or an example on our github page. This file can be stored (and referred to) at any location at the machine, although it recommended to use either the VANTAGE6 system or user folder. These folders are different per operating system.

🗃 Configuration File Structure

Similar to the server, each node instance (configuration) can have multiple environments. You can specify these under the key environments which allows four types: dev , test,acc and prod .

If you do not want to specify any environment, you should only specify the key application .

In the end, the configuration file should have the following structure:

application:
  api_key: api-key-from-server
  server_url: https://api.distributedlearning.ai
  port: 5000
  api_path: '/api'
  task_dir: tasks
  encryption:
    enabled: true
    private_key: /path/to/private_key.pem
  docker_registries:
    - registry: harbor.vantage6.ai
      username: docker-registry-user
      password: your-password!
  allowed_images:
    - ^harbor.vantage6.ai/[a-zA-Z]+/[a-zA-Z]+
  databases:
    default: /path/to/database.csv
    alternative: /path/to/other/database.csv
  logging:
      level:        DEBUG             # Can be on of 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
      file:         node.log          # Filename of logfile
      use_console:  True              # Log output to the console?
      backup_count: 5                 # Number of logs to keep
      max_size:     1024              # Specified in kB (i.e. 1024 means a maximum file size of 1MB)
      format:       "%(asctime)s - %(name)-14s - %(levelname)-8s - %(message)s"
      datefmt:      "%H:%M:%S"

📰 Parameter description

🔒 Encryption

Both the server and node need to agree if encryption is used or not. Encryption at the server is managed at collaboration level, i.e. each collaboration determines whenever encryption of all communication is used or not. You can enable or disable encryption in the configuration file by setting the enabled key in the encryption section.

application:
  ...
  encryption:
    enabled: true
    private_key: /path/to/private_key.pem
  ...

To generate a new private key and upload the public key to the server you can use the command vnode create-private-key. You can also generate the key yourself and upload it by using the /organization/<id> endpoint.

🏳 Allowed Images

To control which algorithms are allowed at the node you can set the allowed_images key in the configuration file. This is expected to be a valid regular expression.

application:
  ...
  allowed_images:
    - ^harbor.vantage6.ai/[a-zA-Z]+/[a-zA-Z]+
  ...

🐳 Docker Login

If you are using a private docker repository the environment in which the node needs to be logged in to that repository. In case the Dockerized version of vantage6 is used you need to specify this in the configuration file using the docker_registries key.

application:
  ...
  docker_registries:
    - registry: harbor.vantage6.ai
      username: docker-registry-user
      password: your-password!
  ...

📃 Extra Environment Variables [2.0.1+]

It is possible to define additional environment variables for the algorithm containers. For example, this could be useful for sharing credentials for accessing a data-store.

application:
    ...
    algorithm_env:
        var1: value1
        var2: 2
    ...