fgimenez/terraform-kubana repository overview

kubana lets you deploy easily a kubernetes cluster over an openstack installation without neutron enabled (tested on the havana release). It is based on https://github.com/openstack/osops-tools-contrib⁠. terraform⁠ takes care of the infrastructure setup and kubeadm⁠ of the installation of the k8s services.

First of all, a base image must be available on the openstack side to be used by the k8s nodes. Given the usual openstack access environment variables are present you can create it using packer⁠ from the images directory, executing:

$ packer build -var 'source_image_name=<my_source_image_name>' k8s_base.json

being my_source_image_name an image available on your openstack installation and suitable to be used with k8s (tested with ubuntu 16.04).

For the actual deplyment kubana uses a container with a prebuilt terraform version patched to play well with old openstack api responses. You should create a terraform.tfvars file with variables customized to your environment, for example:

floatingip_pool = "my_floating_ip_pool"
compute_count = 6
whitelist_network = "aa.bb.cc.dd/ee"

With that in place you can spin up the cluster with:

$ sudo docker run --rm -v ${PWD}:/app \
-e OS_USERNAME=${OS_USERNAME} \
-e OS_PASSWORD=${OS_PASSWORD} \
-e OS_TENANT_NAME=${OS_TENANT_NAME} \
-e OS_AUTH_URL=${OS_AUTH_URL} \
-e OS_REGION_NAME=${OS_REGION_NAME} \
fgimenez/terraform-kubana apply

The output from the execution let's you know how to access the controller node, something like:

kubernetes-controller = $ ssh -i ./dummy_keypair/cloud.key [email protected]

From there you can copy the /etc/kubernetes/admin.conf file to your host on ~/.kube/config to access the cluster with the kubectl cli tool.

You can teardown the cluster with:

$ sudo docker run --rm -v ${PWD}:/app \
-e OS_USERNAME=${OS_USERNAME} \
-e OS_PASSWORD=${OS_PASSWORD} \
-e OS_TENANT_NAME=${OS_TENANT_NAME} \
-e OS_AUTH_URL=${OS_AUTH_URL} \
-e OS_REGION_NAME=${OS_REGION_NAME} \
fgimenez/terraform-kubana destroy