How to deploy on AKS

Azure Kubernetes Service (AKS) allows you to quickly deploy a production ready Kubernetes cluster in Azure. To access the AKS Web interface, go to https://portal.azure.com/.

Summary

• Install AKS and Juju tooling
• Create a new AKS cluster
• Bootstrap Juju on AKS
• Deploy charms
• Display deployment information
• Clean up

Install AKS and Juju tooling

Install Juju and Azure CLI tool:

sudo snap install juju
sudo apt install --yes azure-cli

Follow the installation guides for:

• az - the Azure CLI

To check they are all correctly installed, you can run the commands demonstrated below with sample outputs:

~$ juju version
3.4.2-genericlinux-amd64

~$ az --version
azure-cli                         2.61.0

core                              2.61.0
telemetry                          1.1.0

Dependencies:
msal                              1.28.0
azure-mgmt-resource               23.1.1
...
Your CLI is up-to-date.

Authenticate

Login to your Azure account:

az login

Create a new AKS cluster

Export the deployment name for further use:

export JUJU_NAME=aks-$USER-$RANDOM

This following examples in this guide will use the single server AKS in location eastus - feel free to change this for your own deployment.

Create a new Azure Resource Group:

az group create --name aks --location eastus

Bootstrap AKS with the following command (increase nodes count/size if necessary):

az aks create -g aks -n ${JUJU_NAME} --enable-managed-identity --node-count 1 --node-vm-size=Standard_D4s_v4 --generate-ssh-keys

Sample output:

{
  "aadProfile": null,
  "addonProfiles": null,
  "agentPoolProfiles": [
    {
      "availabilityZones": null,
      "capacityReservationGroupId": null,
      "count": 1,
      "creationData": null,
      "currentOrchestratorVersion": "1.28.9",
      "enableAutoScaling": false,
      "enableEncryptionAtHost": false,
      "enableFips": false,
      "enableNodePublicIp": false,
...

Dump newly bootstraped AKS credentials:

az aks get-credentials --resource-group aks --name ${JUJU_NAME} --context aks

Sample output:

...
Merged "aks" as current context in ~/.kube/config

Bootstrap Juju on AKS

Bootstrap Juju controller:

juju bootstrap aks aks

Sample output:

Creating Juju controller "aks" on aks/eastus
Bootstrap to Kubernetes cluster identified as azure/eastus
Creating k8s resources for controller "controller-aks"
Downloading images
Starting controller pod
Bootstrap agent now started
Contacting Juju controller at 20.231.233.33 to verify accessibility...

Bootstrap complete, controller "aks" is now available in namespace "controller-aks"

Now you can run
	juju add-model <model-name>
to create a new model to deploy k8s workloads.

Create a new Juju model (k8s namespace)

juju add-model welcome aks

[Optional] Increase DEBUG level if you are troubleshooting charms:

juju model-config logging-config='<root>=INFO;unit=DEBUG'

Deploy charms

The following command deploys PostgreSQL K8s:

juju deploy postgresql-k8s --trust -n 3 --channel 14/stable

Sample output:

Deployed "postgresql-k8s" from charm-hub charm "postgresql-k8s", revision 247 in channel 14/stable on ubuntu@22.04/stable

Check the status:

juju status --watch 1s

Sample output:

Model    Controller  Cloud/Region  Version  SLA          Timestamp
welcome  aks         aks/eastus    3.4.2    unsupported  17:53:35+02:00

App             Version  Status  Scale  Charm           Channel       Rev  Address       Exposed  Message
postgresql-k8s  14.11    active      3  postgresql-k8s  14/stable     247  10.0.237.223  no       Primary

Unit               Workload  Agent  Address      Ports  Message
postgresql-k8s/0*  active    idle   10.244.0.19         Primary
postgresql-k8s/1   active    idle   10.244.0.18         
postgresql-k8s/2   active    idle   10.244.0.17  

Display deployment information

Display information about the current deployments with the following commands:

~$ kubectl cluster-info 
Kubernetes control plane is running at https://aks-user-aks-aaaaa-bbbbb.hcp.eastus.azmk8s.io:443
CoreDNS is running at https://aks-user-aks-aaaaa-bbbbb.hcp.eastus.azmk8s.io:443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy
Metrics-server is running at https://aks-user-aks-aaaaa-bbbbb.hcp.eastus.azmk8s.io:443/api/v1/namespaces/kube-system/services/https:metrics-server:/proxy

~$ az aks list
...
        "count": 1,
        "currentOrchestratorVersion": "1.28.9",
        "enableAutoScaling": false,
...

~$ kubectl get node
NAME                                STATUS   ROLES   AGE   VERSION
aks-nodepool1-31246187-vmss000000   Ready    agent   11m   v1.28.9

Clean up

To clean the AKS cluster, resources and juju cloud, run the following commands:

juju destroy-controller aks --destroy-all-models --destroy-storage --force

List all services and then delete those that have an associated EXTERNAL-IP value (load balancers, …):

kubectl get svc --all-namespaces
kubectl delete svc <service-name> 

Next, delete the AKS resources (source: Deleting an all Azure VMs)

az aks delete -g aks -n ${JUJU_NAME}

Finally, logout from AKS to clean the local credentials (to avoid forgetting and leaking):

az logout