Installation

Netcfg can be installed via pip:

$ pip install netcfg

Usage

After installation, there are two parts to netcfg. First, a daemon process must be launched with root privileges and given access to the Docker IPC socket:

$ netcfg daemon

By default, netcfg stores configuration under /var/lib/netcfg/netcfg.json, but this location may be overriden by using the --config command-line argument. After the daemon is running in the background, one can setup netcfg configuration by calling the netcfg script.

First, one should define one or more networks:

$ netcfg create foo0 bridge

The first argument specifies the network name and the other specifies the network type. Currently only networks with type bridge are supported, but netcfg implements different network types as modules so new ones could be added.

Then, we can attach networks to one or more containers:

$ netcfg attach my_container_a foo0 --address 10.42.0.1/24
$ netcfg attach my_container_b foo0 --address 10.42.0.2/24

Currently only static addressing can be configured (IPv4 and IPv6 are supported) and multiple addresses may be specified. In case one only wants an address-less L2 veth device, no address argument should be given.

Existing configuration can be shown by using:

$ netcfg show
{
  "containers": {
    "my_container_a": {
      "name": "my_container_a",
      "networks": {
        "foo0": {
          "address": [
            "10.42.0.1/24"
          ]
        }
      }
    },
    "my_container_b": {
      "name": "my_container_b",
      "networks": {
        "foo0": {
          "address": [
            "10.42.0.2/24"
          ]
        }
      }
    }
  },
  "networks": {
    "foo0": {
      "destroy_on_stop": false,
      "name": "foo0",
      "type": "bridge"
    }
  }
}

If the containers are running, networks will be configured immediately. Otherwise, networks will be configured when the named containers are started.