Config Engine ​

Overview ​

The Config Engine combines a powerful templating language with the knowledge of your network topology.

Templates allow us to render CLI snippets using variables, the real power behind Config Engine comes from the knowledge of your network. At a high level this allows us:

1. Declare logical variables as part of the topology. E.g add an IP address to a node/link.
2. Easily template network links and nodes. Variables are pre-processed to allow templates to simply use local and far-end variables.
3. Send commands to a router (we know the kind of router). This includes knowledge of how to log in and even enable configuration transactions when required. This is covered in the next section

The Config Engine uses Golang's Text Templates as the templating language. We'll cover the basics with many examples, but there are also great external resources covering the Go template syntax.

Variables ​

Variables provide custom input to the templates, and they are specified directly in your topology file. They can be placed under Nodes or Links.

Below you see variables defined for a node. It is located under the node's config.vars branch.

topology:
    nodes:
        pe_1:
            config:
                vars:
                    key: value
                    key_2: value_2
                    key_n: value_n

Similar to the rest of the topology file, defaults can be provided for variables on the kind of the node, or under defaults

topology:
  kinds:
    vr-sros:
      image: ghcr.io/nokia/srlinux
      config:
          key: value
  nodes:
    pe_1:
      kind: vr-sros
    pe_2:
      kind: vr-sros

Variables on links ​

Links can also have variables. These are specified in the topology file next to the link endpoints. Generally link variables are specified in pairs, where each entry is applicable to the endpoint at the same position.

In the example below, 1/1/c1/1 applies to endpoint pe1:eth1 and 1/1/c2/1 applies to endpoint pe2:eth2

topology:
    links:
      - endpoints: ["pe1:eth1", "pe2:eth2"]
        vars:
            port: ['1/1/c1/1', '1/1/c2/1']
            instance_id: 1

In some cases a value might be applicable to both endpoint, and then you only need to specify it as a single value (e.g. instance_id above).

Template files typically target nodes and not links, so link variables are added to each node as a list named clab_links.

Node Role ​

The role of the node points to the template that needs to be deployed on that node. Examples of roles can be leaves, spines or PEs, or CEs. If you have no specific roles in your topology, the bare minimum might be that you have different templates for different types (or kinds) of nodes.

Template files ​

Template files are located in the supplied --template-paths

The naming convention is shown below

The template names are specific on the input using the --template-names / -l flags. The config engine supports deploying a list of templates.

Magic variables ​

We've mentioned that link variables are included with the node variables, but there are also a lot of pre-processing of you variables. Typically when a a variable is used as input (or even generated as output) of this pre-processing step, it starts with clab_

Let's start with a summary of all known magic variables, or jump straight to the examples

The following variables indicate the basic structure and are mostly assembled directly from the topology file

On the nodes you can use the following input.

Each link in the clab_links array will have the following structure:

Magic variable examples ​

In the most basic case, we can specify only a System IP as clab_system_ip. Config Engine generates three variables clab_link_ip and clab_link_name on each link.

topology:
  nodes:
    pe1:
      config:
        vars:
          clab_system_ip: 10.0.0.1/32
    pe2:
      config:
        vars:
          clab_system_ip: 10.0.0.2/32

  links:
    - endpoints: ["pe1:eth1", "pe2:eth1"]