TOSCA Interface

A TOSCA interface is available for configuring and controlling CORD. It is auto-generated from the set of models configured into the POD manifest, and includes both core and service-specific models.

What is TOSCA?

Topology and Orchestration Specification for Cloud Applications (TOSCA) is an OASIS standard language to describe a topology of cloud based web services, their components, relationships, and the processes that manage them. The TOSCA standard includes specifications to describe processes that create or modify web services. You can read more about it on the OASIS website.

CORD extends the TOSCA specification to support custom models for services, allow operators to manage them with a simple and well-known YAML interface.


The xos-tosca container autogenerates the TOSCA interface starting from the xproto definition. When the xos-tosca container starts, it connects to xos-core via the gRPC API to fetch all the xproto definition of the onboarded models. This includes both core and service models. Then using the xos-genx toolchain, it will generates the corresponding TOSCA specifications.

For example, the xproto definition of a compute node in XOS is:

message Node::node_policy (XOSBase) {
     required string name = 1 [max_length = 200, content_type = "stripped", blank = False, help_text = "Name of the Node", null = False, db_index = False];
     required manytoone site_deployment->SiteDeployment:nodes = 2 [db_index = True, null = False, blank = False];

which is then transformed in the following TOSCA spec:

tosca_definitions_version: tosca_simple_yaml_1_0


        derived_from: tosca.nodes.Root
        description: "An XOS Node"
                type: tosca.capabilities.xos.Node
                type: boolean
                default: false
                description: Allow to reference existing models in TOSCA recipes
                type: string
                required: false
                description: "Name of the Node"

        derived_from: tosca.relationships.Root
        valid_target_types: [ tosca.capabilities.xos.SiteDeployment ]

        derived_from: tosca.capabilities.Root
        description: Node

In TOSCA terminology, the above woule be called a TOSCA node type (although confusingly, it's defined for the Node model in CORD, which represents a server).


Once CORD is up and running, a node can be added to a POD using the TOSCA interface by uploading the following recipe:

tosca_definitions_version: tosca_simple_yaml_1_0

description: Load a compute node in XOS

   - custom_types/node.yaml


    # A compute node
      type: tosca.nodes.Node
        name: Grateful Vest

In TOSCA terminology, the above would be called a TOSCA node template.

Where to find the generated specs?

On any running CORD POD, the TOSCA apis are accessible as:

curl http://<head-node-ip>:<head-node-port>/xos-tosca | python -m json.tool

And it will return a list of all the recipes with the related url:

  "image": "/custom_type/image",
  "site": "/custom_type/site",

For examples, to site the TOSCA spec of the Site model, you can use the URL:

curl http://<head-node-ip>:<head-node-port>/xos-tosca/custom_type/site

How to load a TOSCA recipe in the system

The xos-tosca container exposes two endpoint:

POST http://<cluster-ip>:<tosca-port>/run
POST http://<cluster-ip>:<tosca-port>/delete

To load a recipe via curl you can use this command:

curl -H "xos-username:" -H "xos-password: <xos-password>" -X POST --data-binary @<path/to/file> http://<cluster-ip>:<tosca-port>/run

_If you installed the xos-core charts without modifications, the tosca-port is 30007

results matching ""

    No results matching ""