The Rook CRDs are evolving over time and to verify that clients make use of the correct definitions, this project provides generated API classes to let users write correct clients easily.
Right now, it supports three operators:
- Ceph
- Edgefs
- Cassandra
It is used to type check client code against the Rook API using mypy.
Inspired by kubernetes-client/python
Main uses case is the mgr/rook orchestrator module of the Ceph MGR
This project contains a generate_model_classes.py
which can generate
model classes from Kubernetes CRDs and is independent of Rook.
Usage:
python generate_model_classes.py <crds.yaml> <output-folder>
To install rook-client
, run:
pip install -e 'git+https://github.com/ceph/rook-client-python#egg=rook-client'
def objectstore(api_name, name, namespace, instances):
from rook_client.ceph import cephobjectstore as cos
rook_os = cos.CephObjectStore(
apiVersion=api_name,
metadata=dict(
name=name,
namespace=namespace
),
spec=cos.Spec(
metadataPool=cos.MetadataPool(
failureDomain='host',
replicated=cos.Replicated(
size=1
)
),
dataPool=cos.DataPool(
failureDomain='osd',
replicated=cos.Replicated(
size=1
)
),
gateway=cos.Gateway(
port=80,
instances=instances
)
)
)
return rook_os.to_json()
To re-generate the python files, run
./generate.sh