Docker-only Install
This is not the intended use! However there are use-cases where you do not want to use our CLI and start the Node docker container manually.
Pull the latest Node docker image:
docker pull harbor.distributedlearning.ai/infrastructure/node
This Docker container expects multiple mounts, a volume, and a environment variable.
Mount path | Description |
/mnt/database.csv | Contains the database file that contains the data that the Node is allowed to access |
/mnt/log | Directory where the log files can be stored at the host machine |
/mnt/data | Docker data volume which is used for transferring data between node and algorithms |
/mnt/config | Directory containing the configuration file |
/var/run/docker.sock | Contains the docker.sock from the host system. Usually: //var/run/docker.sock |
/mnt/private_key.pem | Optional. If provided the private key will be used to decrypt any incoming messages. |
The node expects a single Docker volume, in which it can share data with the algorithm containers. You can give this volume any name as you specify its name in the environment variables.
docker volume create datavolume-name
The configuration_name is the filename without the .yaml extension, and the environment is the name of the environment you want to use: test , prod , acc ,dev or application, see this section how the configuration file should be formatted.
docker run \-e DATA_VOLUME_NAME="some-name" \-v some-name:/mnt/data-volume \-v /host/database.csv:/mnt/database.csv \-v /host/log:/mnt/log \-v datavolume-name:/mnt/data \-v /host/config:/mnt/config \-v //var/run/docker.sock:/var/run/docker.sock \-v /host/private_key.pem:/mnt/private_key \harbor.distributedlearning.ai/infrastructure/node \configuration_name environment
docker run `-e DATA_VOLUME_NAME="some-name" `-v some-name:/mnt/data-volume `-v /host/database.csv:/mnt/database.csv `-v /host/log:/mnt/log `-v /host/data:/mnt/data `-v /host/config:/mnt/config `-v //var/run/docker.sock:/var/run/docker.sock `-v /host/private_key.pem:/mnt/private_key `harbor.distributedlearning.ai/infrastructure/node `configuration_name environment