[CloudCoil](https://github.com/cloudcoil/cloudcoil) - Production-ready Python client for Kubernetes with async support
I've been working on improving the Python development experience for Kubernetes, and I'm excited to share CloudCoil - a modern K8s client that brings features like async/await, type safety, and integrated testing to the Python ecosystem.
Why another Kubernetes client?
In the Python ecosystem, we've been missing features that Go developers take for granted - things like robust client implementations, proper type safety, and integrated testing tools. CloudCoil aims to fix this by providing:
1) Production-focused features:
* 🔥 Elegant, Pythonic API - Feels natural to Python developers
* ⚡ Async First - Native async/await support for high performance
* 🛡️ Type Safe - Full mypy support and runtime validation
* 🧪 Testing Ready - Built-in pytest fixtures for K8s integration tests
* 📦 Zero Config - Works with your existing kubeconfig
* 🪶 Minimal Dependencies - Only requires httpx, pydantic, and pyyaml
2) First-class operator support:
- [cert-manager](https://github.com/cloudcoil/models-cert-manager)
- [FluxCD](https://github.com/cloudcoil/models-fluxcd)
- [Kyverno](https://github.com/cloudcoil/models-kyverno)
(More coming soon - let me know what you'd like to see!)
3) Rich features for production use:
Resource watching with async support:
async for event_type, pod in await core.v1.Pod.async_watch(
field_selector="metadata.name=mypod"
):
if event_type == "DELETED":
break
Smart wait conditions:
pod = core.v1.Pod.get("test-pod")
status = await pod.async_wait_for({
"succeeded": lambda _, pod: pod.status.phase == "Succeeded",
"failed": lambda _, pod: pod.status.phase == "Failed"
}, timeout=300)
Dynamic CRD support:
DynamicCRD = resources.get_dynamic_resource(
"MyCustomResource",
"example.com/v1"
)
resource = DynamicCRD(
metadata={"name": "example"},
spec={"someField": "value"}
).create()
4) Installation:
Choose your K8s version:
# Latest version
pip install cloudcoil[kubernetes]
# Specific K8s version
pip install cloudcoil[kubernetes-1-32]
The project is Apache 2.0 licensed and ready for production use. We'd especially love feedback from:
- Teams using Python for K8s automation
- Anyone building operators/controllers in Python
- DevOps engineers managing multiple clusters
Links:
* GitHub: https://github.com/cloudcoil/cloudcoil
* Docs: https://cloudcoil.github.io/cloudcoil
* PyPI: https://pypi.org/project/cloudcoil
Looking forward to your feedback, especially on what operators you'd like to see supported next!
* fix: add application/yaml section in api-concepts
* nit fix: breaking the request and response into distinct code blocks
* fix: split the request and response into separate code blocks
* Update deprecation-guide.md
**targetAverageUtilization** is no longer supported by **autoscaling/v2**
* Remove sample and add document reference
* Update notable changes for HPA in 1.25 and 1.26
* cite a document metric usage
* Update deprecation-guide.md
* Document CEL option, library and language feature availability by version
* Update content/en/docs/reference/using-api/cel.md
Co-authored-by: Tim Bannister <tim@scalefactory.com>
* Apply feedback
* Feedback applied
* noop commit to trigger build
---------
Co-authored-by: Tim Bannister <tim@scalefactory.com>
Sambhav Kothari
windsonsea
Kubernetes Prow Robot
Abu Kashem
Christian Schlotter
yongruilin
Ben Luddy
Lukasz Szaszkiewicz
abhilasha
Sagnik Pal
Tim Bannister
kohbis
Daniel Chan
Ivan ROGER
Marek Siarkowicz
Jordan Liggitt
Ritikaa96
Jun Duan
Keigo Kurita
harsh
John Huang
Francis Huang
Steve Kuznetsov
cici37
Alexander Zielenski
Abhay Krishna Arunachalam
Joe Betz
Takashiidobe
Sergey Kanzhelev
Rui Chen
Kubernetes Prow Robot