kubecrd

Create Kubernetes CRD using Python dataclasses


Keywords
apischema, custom-resource-definition, k8s, kubernetes, kubernetes-crd, kubernetes-operators, openapi-specification
License
Apache-2.0
Install
pip install kubecrd==0.5.0

Documentation

Kube CRD

The primary purpose of this project is to simplify working with Kubernetes Custom Resources. To achieve that it provides a base class, kubecrd.KubeResourceBase that can create Python dataclassses into Kubernetes Custom Resources and also generate and install Custom Resource Definitions for those resource into the K8s cluster directly.

>>> from dataclasses import dataclass, field
>>> from uuid import UUID
>>> from kubecrd import KubeResourceBase
>>> from apischema import schema
>>> @dataclass
... class Resource(KubeResourceBase):
...     __group__ = 'example.com'
...     __version__ = 'v1alpha1'
...
...     name: str
...     tags: list[str] = field(
...         default_factory=list,
...         metadata=schema(
...            description='regroup multiple resources',
...            unique=False,
...         ),
...     )
>>> print(Resource.crd_schema())
apiVersion: apiextensions.k8s.io/v1
kind: CustomResourceDefinition
metadata:
  name: resources.example.com
spec:
  group: example.com
  names:
    kind: Resource
    plural: resources
    singular: resource
  scope: Namespaced
  versions:
  - name: v1alpha1
    schema:
      openAPIV3Schema:
        properties:
          spec:
            properties:
              name:
                type: string
              tags:
                default: []
                description: regroup multiple resources
                items:
                  type: string
                type: array
                uniqueItems: false
            required:
            - name
            type: object
        type: object
    served: true
    storage: true
<BLANKLINE>

Create CRD in K8s Cluster

It is also possible to install the CRD in a cluster using a Kubernetes Client object:

from from kubernetes import client, config
config.load_kube_config()
k8s_client = client.ApiClient()
Resource.install(k8s_client)

You can then find the resource in the cluster:

» kubectl get crds/resources.example.com
NAME                    CREATED AT
resources.example.com   2022-03-20T03:58:25Z

$ kubectl api-resources | grep example.com
resources     example.com/v1alpha1                  true         Resource

Installation of resource is idempotent, so re-installing an already installed resource doesn't raise any exceptions if exist_ok=True is passed in:

Resource.install(k8s_client, exist_ok=True)

Serialization

You can serialize a Resource such that it is suitable to POST to K8s:

>>> example = Resource(name='myResource', tags=['tag1', 'tag2'])
>>> import json
>>> print(json.dumps(example.serialize(), sort_keys=True, indent=4))
{
    "apiVersion": "example.com/v1alpha1",
    "kind": "Resource",
    "metadata": {
        "name": "..."
    },
    "spec": {
        "name": "myResource",
        "tags": [
            "tag1",
            "tag2"
        ]
    }
}

Objects can also be serialized and saved directly in K8s:

example.save(k8s_client)

Where client in the above is a Kubernetes client object. You can also use asyncio with kubernetes_asyncio client and instead do:

await example.async_save(k8s_async_client)

Deserialization

You can deserialize the JSON from Kubernetes API into Python CR objects.

$ cat -p testdata/cr.json
{
 "apiVersion": "example.com/v1alpha1",
 "kind": "Resource",
 "metadata": {
     "generation": 1,
     "name": "myresource1",
     "namespace": "default",
     "resourceVersion": "105572812",
     "uid": "02102eb3-968b-418a-8023-75df383daa3c"
 },
 "spec": {
     "name": "bestID",
     "tags": [
         "tag1",
         "tag2"
     ]
 }
 }

by using from_json classmethod on the resource:

>>> import json
>>> with open('testdata/cr.json') as fd:
...     json_schema = json.load(fd)
>>> res = Resource.from_json(json_schema)
>>> print(res.name)
bestID
>>> print(res.tags)
['tag1', 'tag2']

This also loads the Kubernetes's V1ObjectMeta and sets it as the .metadata property of CR:

>>> print(res.metadata.namespace)
default
>>> print(res.metadata.name)
myresource1
>>> print(res.metadata.resource_version)
105572812

Watch

It is possible to Watch for changes in Custom Resources using the standard Watch API in Kubernetes. For example, to watch for all changes in Resources:

async for happened, resource in Resource.async_watch(k8s_async_client):
    print(f'Resource {resource.metadata.name} was {happened}')

Or you can use the block sync API for the watch:

for happened, resource in Resource.watch(k8s_client):
    print(f'Resource {resource.metadata.name} was {happened}')

Installing

Kube CRD can be install from PyPI using pip or your favorite tool:

$ pip install kubecrd