External Connectivity

netlab contains several mechanisms that allow you to manage physical labs, add physical devices to virtual labs, connect to the outside world from virtual lab devices, or use network management software packaged as containers or virtual machines with your virtual labs.

Outbound Connectivity

libvirt, VirtualBox and containerlab try to add IPv4 default routes to lab devices. libvirt and Virtualbox use a DHCP option, containerlab installs a default route into the container network namespace[1]. Most network devices running in a virtual lab are thus able to reach external destinations.

Most box-building recipes for libvirt and Virtualbox Vagrant plugins recommend using a management VRF for the management interface. The default route is thus installed into the management VRF, and the client you’re using on the network device has to be VRF-aware to reach external destinations. For example, you’ll have to use a command similar to ping vrf name destination to ping external IP addresses.

Connecting to Lab Devices

libvirt and containerlab providers create configuration files that connect all lab devices to a management network. Together with the default route configured on network devices, it’s always possible to reach the management IP address of every device in your lab, but you have to fix the routing in the external network – the management network IPv4 prefix has to be reachable from the external network.

Alternatively, use graphite for GUI-based SSH access to your lab network or port forwarding to map VM/container management TCP ports to the host ports. Port forwarding is always used with Virtualbox, and configurable with libvirt and containerlab providers. Use netlab report mgmt to display the host-to-lab-device TCP port mapping.

Finding the Management IP Addresses

netlab report mgmt command displays the management IP addresses[2] of the lab devices, protocol used to configure the devices (SSH, NETCONF or docker), and the username/password used by netlab to configure the device.

Alternatively, you could use Ansible inventory to find the same information:

  • Run ansible-inventory --host _device-name_ to display the Ansible variables for the specified lab device.

  • Look for ansible_host variable or ipv4 value in mgmt dictionary.

Finally, you could display node information in YAML format with the netlab inspect –node nodename command, or analyze the nodes dictionary in the netlab.snapshot.yml file with yq or a custom script.

Control-Plane Connectivity

If you need control-plane connectivity to your lab devices (for example, you’d like to run BGP with a device outside of your lab), consider running your additional devices as virtual machines in the lab. Please see Unknown Devices and Unprovisioned Devices for more details.

To connect libvirt virtual machines or containerlab containers to the outside world, set libvirt.uplink or clab.uplink link attribute on any link in your topology.

VirtualBox uses a different connectivity model. It maps device TCP/UDP ports into host TCP/UDP ports. The default ports mapped for each network device are ssh, http and netconf. It’s possible to add additional forwarded ports to the defaults.providers.virtualbox.forwarded parameter; the details are beyond the scope of this tutorial.

VirtualBox can connect VMs to the external world. That capability is not part of netlab functionality; please feel free to submit a Pull Request implementing it.

Unprovisioned Devices

The easiest way to add network management software (or any third-party workload) to your lab is to deploy it as a node in your network:

  • Define an extra linux node in your lab topology

  • Use image node attribute to specify a Vagrant box or container image to use.

The lab provisioning process will configure the static routes on your VM/container to allow it to reach all other devices in your lab.

The VM device provisioning process will fail if your VM does not contain Python (used by Ansible) or the necessary Linux CLI commands (example: ip to add static routes); container interface addresses and routing tables are configured from the Linux server.

If you want to use a VM that cannot be configured as a Linux host, put that node into the unprovisioned group, for example:

defaults.device: iosv
    device: linux
    image: awesome-sw

    members: [ nms ]


Devices in the unprovisioned group will not get IP addresses on interfaces other than the management interface, or static routes to the rest of the network.

As they are still connected to the management network, they can always reach the management interfaces of all network devices.

Unmanaged Devices

In advanced scenarios connecting your virtual lab with the outside world, you might want to include external devices into your lab topology without managing or provisioning them[3].

For example, if you want to have a BGP session with an external router:

  • Define the external router as another device in your lab topology.

  • Use static IP prefixes on the link between the virtual devices and the external router to ensure the virtual devices get IP addresses from the subnet configured on the external router

  • Define BGP AS numbers used by your devices and the external router – netlab will automatically build IBGP/EBGP sessions between lab devices and the external device

  • Use unmanaged node attribute on the external node to tell netlab not to include it in Ansible inventory or Vagrant/containerlab configuration files

Here is the resulting topology file using an Arista vEOS VM running BGP with an external Arista EOS switch. The lab is using libvirt public network to connect the VM to the outside world:

defaults.device: eos
module: [ bgp ]
    bgp.as: 65000
    unmanaged: True
    bgp.as: 65001
- vm:
  libvirt.public: True

Managing Physical Devices

If you want to create configurations for a prewired physical lab, use the external provider.

Before using netlab with a physical lab, you’ll have to create a lab topology that specifies the specify management IP addresses and interface names for all devices in your lab. Once that’s done, save the topology as a blueprint for further lab work.

Starting with the physical lab blueprint topology, add addressing plans (if needed), configuration modules, and configuration module parameters. Use netlab up to start the data transformation process and configuration deployment on physical devices.

Please note that netlab does not contain a cleanup procedure for physical devices – you’ll have to remove the device configurations before starting the next lab.