How to integrate Canonical Kubernetes with etcd

Integrating etcd with your Canonical Kubernetes deployment provides a robust, distributed key-value store that is essential for storing critical data needed for Kubernetes’ clustering operations. This guide will walk you through the process of deploying Canonical Kubernetes with an external etcd cluster.

What you will need

  • A Juju controller with access to a cloud environment (see the Juju setup guide for more information).

Warning

Once you deploy your Canonical Kubernetes cluster with a particular datastore, you cannot switch to a different datastore post-deployment. Planning for your datastore needs ahead of time is crucial, particularly if you opt for an external datastore like etcd.

Preparing the Deployment

  1. Creating the Deployment Model: Begin by creating a Juju model specifically for your Canonical Kubernetes cluster deployment.

juju add-model my-cluster

  1. Deploying Certificate Authority: etcd requires a secure means of communication between its components. Therefore, we require a certificate authority such as EasyRSA or Vault. Check the respective charm documentation for detailed instructions on how to deploy a certificate authority. In this guide, we will be using EasyRSA.

juju deploy easyrsa

Deploying etcd

  • Single Node Deployment:

    • To deploy a basic etcd instance on a single node, use the command:

      juju deploy etcd

      This setup is straightforward but not recommended for production environments due to a lack of high availability.

  • High Availability Setup:

    • For environments where high availability is crucial, deploy etcd across at least three nodes:

      juju deploy etcd -n 3

      This ensures that your etcd cluster remains available even if one node fails.

Integrating etcd with EasyRSA

Now you have to integrate etcd with your certificate authority; this will issue the required certificates for secure communication between etcd and your Canonical Kubernetes cluster:

juju integrate etcd easyrsa

Deploying Canonical Kubernetes

Deploy the control plane units of Canonical Kubernetes with the command:

juju deploy k8s --config datastore=etcd -n 3

This command deploys 3 units of the Canonical Kubernetes control plane (k8s) and configures them to use etcd as the backing datastore, ensuring high availability.

Important

Remember to run juju expose k8s. This will open the required ports to reach your cluster from outside.

Integrating Canonical Kubernetes with etcd

Now that we have both the etcd datastore deployed alongside our Canonical Kubernetes cluster, it is time to integrate our cluster with our etcd datastore.

juju integrate k8s etcd

This step integrates the k8s charm (Control Plane units) with the etcd hosts, allowing the Kubernetes cluster to use the etcd units as an external datastore.

Final Steps

Verify the Deployment: After completing the deployment, it’s essential to verify that all components are functioning correctly. Use the juju status command to inspect the current status of your cluster.

Model       Controller  Cloud/Region   Version  SLA          Timestamp
my-cluster  canonicaws  aws/us-east-1  3.4.2    unsupported  16:02:18-05:00

App      Version  Status  Scale  Charm    Channel        Rev  Exposed  Message
easyrsa  3.0.1    active      1  easyrsa  latest/stable   58  no       Certificate Authority connected.
etcd     3.4.22   active      3  etcd     latest/stable  760  no       Healthy with 3 known peers
k8s      1.29.4   active      3  k8s      latest/edge     33  yes      Ready

Unit        Workload  Agent  Machine  Public address  Ports     Message
easyrsa/0*  active    idle   0        35.172.230.66             Certificate Authority connected.
etcd/0*     active    idle   1        34.204.173.161  2379/tcp  Healthy with 3 known peers
etcd/1      active    idle   2        54.225.4.183    2379/tcp  Healthy with 3 known peers
etcd/2      active    idle   3        3.208.15.61     2379/tcp  Healthy with 3 known peers
k8s/0       active    idle   4        54.89.153.117   6443/tcp  Ready
k8s/1*      active    idle   5        3.238.230.3     6443/tcp  Ready
k8s/2       active    idle   6        34.229.202.243  6443/tcp  Ready

Machine  State    Address         Inst id              Base          AZ          Message
0        started  35.172.230.66   i-0b6fc845c28864913  [email protected]  us-east-1f  running
1        started  34.204.173.161  i-05439714c88bea35f  [email protected]  us-east-1f  running
2        started  54.225.4.183    i-07ecf97ed29860334  [email protected]  us-east-1c  running
3        started  3.208.15.61     i-0be91170809d7dccc  [email protected]  us-east-1b  running
4        started  54.89.153.117   i-07906e76071b69721  [email protected]  us-east-1c  running
5        started  3.238.230.3     i-0773583e7a5fbf07e  [email protected]  us-east-1f  running
6        started  34.229.202.243  i-0f03b9833a338380c  [email protected]  us-east-1b  running