kubectl-openstack

kubectl-openstack

A tool that helps you manage OpenStack infrastructure CLI access for Cluster API OpenStack (CAPO) clusters.

Usage

Typically you simply specify the name of one of your CAPO clusters:

$ kubectl get openstackcluster -A
NAMESPACE   NAME           CLUSTER        READY   NETWORK                                SUBNET                                 BASTION IP
my-ns1      my-cluster-1   my-cluster-1   true    xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx   xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
my-ns1      my-cluster-2   my-cluster-1   true    xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx   xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx

The --management-cluster flag can be skipped. More on that later. If the cluster name is not unique across namespaces then --namepsace flag is required.

$ ./kubectl-openstack login --management-cluster my-mc my-cluster-1
Writing "my-mc-my-cluster-1" cloud to /home/username/.config/openstack/clouds.yaml

To use the cloud run:

    openstack --os-cloud="my-mc-my-cluster-1" server list

If --management-cluster flag flag is omitted then it is inferred as the second segment of the API URL.

Example:

$ kubectl cluster-info
Kubernetes control plane is running at https://api.my-mc.test.gigantic.io:6443
CoreDNS is running at https://api.my-mc.test.gigantic.io:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

The inferred Management Cluster name would be my-mc.

The full --help output:

kubectl-openstack login --help
Usage of ./kubectl-openstack:
      --clouds-file string          absolute path to the clouds.yaml file (default "$HOME/.config/openstack/clouds.yaml")
  -f, --force                       force overwriting existing cloud (if it exists) in the clouds file
      --kubeconfig string           absolute path to the kubeconfig file (default "$HOME/.kube/config")
      --management-cluster string   (optional) name of the management cluster, if not set will be inferred from the API URL
  -n, --namespace string            (optional) namespace of the OpenstackCluster resource, required only if the cluster name is ambiguous

Things to do

• Makefiles with install
• Write installation instructions
• Add to Krew
• Readme with placeholders for --help
• Improve command package structure (the login command is currently a fixed string in main.go)
• Extract a package with generic collection functions (contains, keys, etc.)
• Add --version flag
• Add selfupdate command