/ansible-terraform-microk8s-openstack

Primary LanguageHCL

Terraform+Ansible microk8s OpenStack deployment

This is an example of how to deploy a microk8s cluster in an OpenStack instance using Terraform and Ansible. It is based on the following repositories:

Quick start

pip install ansible
  • Log in Openstack sourcing the openrc file that your Openstack instance provides.
. project_YYYXXXX-openrc.sh

In order to configure this deployment the file group_vars/all must be used.

Name Description Default
network this is the name of the network the Virtual Machines will be attached to. You can get the list with openstack network list. -
keypair this is the name of the public ssh key stored in OpenStack that will be added to the Virtual Machines. You can get the list of keys installed in OpenStack with openstack keypair list. You must choose one key from that list. -
private_key_path this is the path on you computer where Terraform will find the private key. This private key has to be the pair of the public key selected in keypair. -
cidr_list list of CIDRs (an IP range) that will be able to access the cluster. 193.166.1.0/24,193.166.2.0/24, 193.166.80.0/23, 193.166.85.0/24
cidr_ssh list of CIDRs (an IP range) that will be able to SSH to the cluster nodes 0.0.0.0/0 # Any IP
instance_count number of worker nodes. 0 # the master node will run the applications
flavor Openstack flavor for the Virtual Machines. standard.medium
nfs_volume_size size in Gigabyes for the NFS volume. Resize is not supported, when changing the size, a cluster recreation is recommended. 50
instance_master_name name of the master Virtual machine. microk8s-master
instance_prefix prefix of the name of the worker nodes. A hiphen and a number will be added to form the worker node name. By default the first node will be called microk8s-node-0, the second microk8s-node-1 and so on. microk8s-node
nfs_node_name name of the NFS virtual machine. microk8s-nfs
ssh_user name of the username to login in the Virtual machines. It must correspond to the one configured on the OS image used. ubuntu
microk8s_version, version to install. 1.24
microk8s_plugins, allows to enable or disable individual microk8s plugins.

After all variables are properly set, simply run:

terraform -chdir=terraform init
# This must be run only once to initiliaze Terraform's plugins
ansible-playbook site.yaml

This will deploy the Virtual Machines and configure them. The cluster will be ready for use.

Kubectl configuration

kubectl is the command line tool to interact with a Kubernetes cluster. The master node has kubectl installed by default. It is possible to install and configure kubectl in any remote computer to interact with this cluster. There are 3 steps:

  1. Install kubectl
  2. Make sure that the computer's IP belongs to one of the CIDRs configured in cidr_list.
  3. Generate and copy a config file:
export MASTER_IP=<microk8s_master>
ssh ubuntu@$MASTER_IP microk8s.config >config
sed -i "s#\(server: https://\)[0-9\.]*:#\1$MASTER_IP:#" config

Note: Replace <microk8s_master> with the master's floating ip

Note 2: In order to tell kubectl where is the config file simply do kubectl --kubeconfig config. Or copy it to the default path: $HOME/.kube/config.

Example application deployment

We will deploy a minimal example, with an Ingress, Service and Deployment

  1. Edit examples/ingress.yaml and change XXXX.kaj.pouta.csc.fi for the DNS of the master's floating ip.

  2. Run kubectl create

    kubectl --kubeconfig config create -f examples/deployment.yaml -f examples/service.yaml -f examples/ingress.yaml

  3. Visit http://<microk8s_master>/ you should be greeted by:

    "Directory listing for /"

Un-deploy

terraform -chdir=terraform destroy

This will destroy the cluster and all its data