Skip to content

gtsystem/lightkube

Repository files navigation

lightkube

Coverage Status pypi supported versions

Modern lightweight kubernetes module for python

Highlights

  • Simple interface shared across all kubernetes APIs.
  • Extensive type hints to avoid common mistakes and to support autocompletion.
  • Models and resources generated from the swagger specifications using standard dataclasses.
  • Load/Dump resource objects from YAML.
  • Support for async/await
  • Support for installing a specific version of the kubernetes models (1.16 to 1.31)
  • Lazy instantiation of inner models.
  • Fast startup and small memory footprint as only needed models and resources can be imported.
  • Automatic handling of pagination when listing resources.

This module is powered by httpx.

Installation

This module requires python >= 3.7

pip install lightkube

Usage

Read a pod

from lightkube import Client
from lightkube.resources.core_v1 import Pod

client = Client()
pod = client.get(Pod, name="my-pod", namespace="default")
print(pod.namespace.uid)

List nodes

from lightkube import Client
from lightkube.resources.core_v1 import Node

client = Client()
for node in client.list(Node):
    print(node.metadata.name)

Watch deployments

from lightkube import Client
from lightkube.resources.apps_v1 import Deployment

client = Client()
for op, dep in client.watch(Deployment, namespace="default"):
    print(f"{dep.namespace.name} {dep.spec.replicas}")

Create a config map

from lightkube.resources.core_v1 import ConfigMap
from lightkube.models.meta_v1 import ObjectMeta

config = ConfigMap(
    metadata=ObjectMeta(name='my-config', namespace='default'),
    data={'key1': 'value1', 'key2': 'value2'}
)

client.create(config)

Replace the previous config with a different content

config.data['key1'] = 'new value'
client.replace(config)

Patch an existing config adding a label

patch = {'metadata': {'labels': {'app': 'xyz'}}}
client.patch(ConfigMap, name='my-config', namespace='default', obj=patch)

Remove the label app

# When using PatchType.STRATEGIC (default), setting a value of a key/value to None, will remove the current item 
patch = {'metadata': {'labels': {'app': None}}}
client.patch(ConfigMap, name='my-config', namespace='default', obj=patch)

Delete a namespaced resource

client.delete(ConfigMap, name='my-config', namespace='default')

Create resources defined in a file

from lightkube import Client, codecs

client = Client()
with open('deployment.yaml') as f:
    for obj in codecs.load_all_yaml(f):
        client.create(obj)

Scale a deployment

from lightkube.resources.apps_v1 import Deployment
from lightkube.models.meta_v1 import ObjectMeta
from lightkube.models.autoscaling_v1 import ScaleSpec

obj = Deployment.Scale(
    metadata=ObjectMeta(name='metrics-server', namespace='kube-system'),
    spec=ScaleSpec(replicas=1)
)
client.replace(obj)

Create and modify resources using server side apply

Note: field_manager is required for server-side apply. You can specify it once in the client constructor or when calling apply(). Also apiVersion and kind need to be provided as part of the object definition.

from lightkube.resources.core_v1 import ConfigMap
from lightkube.models.meta_v1 import ObjectMeta

client = Client(field_manager="my-manager")
config = ConfigMap(
    # note apiVersion and kind need to be specified for server-side apply
    apiVersion='v1', kind='ConfigMap',
    metadata=ObjectMeta(name='my-config', namespace='default'),
    data={'key1': 'value1', 'key2': 'value2'}
)

res = client.apply(config)
print(res.data)
# prints {'key1': 'value1', 'key2': 'value2'}

del config.data['key1']
config.data['key3'] = 'value3'

res = client.apply(config)
print(res.data)
# prints {'key2': 'value2', 'key3': 'value3'}

Stream pod logs

from lightkube import Client

client = Client()
for line in client.log('my-pod', follow=True):
    print(line)

Unsupported features

The following features are not supported at the moment:

  • Special subresources attach, exec, portforward and proxy.
  • auth-provider authentication method is not supported. The supported authentication methods are token, username + password and exec.