This is in connection to the topic of Using Kubernetes with Juju. See that page for background information.
This tutorial will demonstrate the use of the ‘aws-integrator’ charm with the AWS cloud to make Kubernetes dynamic persistent volumes (PVs) available for use with Kubernetes-specific charms.
The following prerequisites are assumed as a starting point for this tutorial:
- You’re using Ubuntu 18.04 LTS.
v.2.5.0is installed. See the Installing Juju page.
- A credential for the ‘aws’ cloud has been added to Juju. See the Using Amazon AWS with Juju page.
- Sufficient permissions are assigned to the above credential in order for ‘aws-integrator’ to perform operations (see Permissions Requirements; this tutorial assigns the IAM security policy of ‘AdministratorAccess’).
Let’s begin by creating a controller. We’ll call it ‘aws-k8s’:
juju bootstrap aws aws-k8s
We’ll deploy Kubernetes using the ‘kubernetes-core’ bundle, which will give us a minimalist cluster. We’ll add the integrator charm to the mix by means of an overlay bundle that we’ll store in file
- ['aws-integrator', 'kubernetes-master']
- ['aws-integrator', 'kubernetes-worker']
See Overlay bundles for details on overlays.
We can now deploy the cluster like so:
juju deploy kubernetes-core --overlay k8s-aws-overlay.yaml
juju trust aws-integrator
trust command grants ‘aws-integrator’ access to the credential used in the
bootstrap command. This charm acts as a proxy for the Juju machines, acting as Kubernetes nodes, to create and attach dynamic storage volumes in the AWS backing cloud.
When using a managed Kubernetes service like Azure’s AKS or Google’s GKE, using an integrator charm is not required. These Kubernetes clusters have storage built in.
It will take about ten minutes to arrive at a stable
status command output:
Model Controller Cloud/Region Version SLA Timestamp
default aws-k8s aws/us-east-1 2.5.0 unsupported 03:28:12Z
App Version Status Scale Charm Store Rev OS Notes
aws-integrator 1.15.71 active 1 aws-integrator jujucharms 8 ubuntu
easyrsa 3.0.1 active 1 easyrsa jujucharms 195 ubuntu
etcd 3.2.10 active 1 etcd jujucharms 338 ubuntu
flannel 0.10.0 active 2 flannel jujucharms 351 ubuntu
kubernetes-master 1.13.2 active 1 kubernetes-master jujucharms 542 ubuntu exposed
kubernetes-worker 1.13.2 active 1 kubernetes-worker jujucharms 398 ubuntu exposed
Unit Workload Agent Machine Public address Ports Message
aws-integrator/0* active idle 0 22.214.171.124 ready
easyrsa/0* active idle 0/lxd/0 10.57.10.22 Certificate Authority connected.
etcd/0* active idle 0 126.96.36.199 2379/tcp Healthy with 1 known peer
kubernetes-master/0* active idle 0 188.8.131.52 6443/tcp Kubernetes master running.
flannel/1 active idle 184.108.40.206 Flannel subnet 10.1.101.1/24
kubernetes-worker/0* active idle 1 220.127.116.11 80/tcp,443/tcp Kubernetes worker running.
flannel/0* active idle 18.104.22.168 Flannel subnet 10.1.11.1/24
Machine State DNS Inst id Series AZ Message
0 started 22.214.171.124 i-06b046ea0ade98e9c bionic us-east-1a running
0/lxd/0 started 10.57.10.22 juju-06e5d4-0-lxd-0 bionic us-east-1a Container started
1 started 126.96.36.199 i-04c67dc1d633c2794 bionic us-east-1b running
Adding the cluster to Juju
We’ll now copy over the cluster’s main configuration file and then use the
add-k8s command to add the cluster to Juju’s list of known clouds. Here, we arbitrarily call the new cloud ‘k8s-cloud’:
juju scp kubernetes-master/0:config ~/.kube/config
juju add-k8s k8s-cloud
The success of this operation can be confirmed by running
Adding a model
When we add a Kubernetes cluster to Juju we effectively have two clouds being managed by one controller. For us, they are named ‘aws’ and ‘k8s-cloud’. So when we want to create a model we’ll need explicitly state which cloud to place the new model in. We’ll do this now by adding a model called ‘k8s-model’ to cloud ‘k8s-cloud’:
juju add-model k8s-model k8s-cloud
Dynamic persistent volumes
As opposed to static Kubernetes persistent volumes, dynamic PVs do not need to be created in advance. They will be created on an as-needed basis by the cluster. This is generally the preferred method.
Tutorial Setting up static Kubernetes storage shows how to set up static PVs and includes information on how to inspect a cluster and its various objects using the
Creating Juju storage pools
The storage pool name for operator storage must be called ‘operator-storage’ while the pool name for workload storage is arbitrary. Here, our charm has storage requirements so we’ll need a pool for it. We’ll call it ‘k8s-pool’. It is this workload storage pool that will be used at charm deployment time.
For dynamic AWS volumes, the Kubernetes provisioner is
kubernetes.io/aws-ebs. We will also request a general purpose SSD drive by passing the
Our two storage pools are therefore created like this:
juju create-storage-pool operator-storage kubernetes \
juju create-storage-pool k8s-pool kubernetes \
Deploying a Kubernetes charm
We can now deploy a Kubernetes charm. For example, here we deploy a charm by requesting the use of the ‘k8s-pool’ workload storage pool we just set up:
juju deploy cs:~juju/gitlab-k8s
juju deploy cs:~juju/mariadb-k8s --storage database=k8s-pool,10M
juju add-relation gitlab-k8s mariadb-k8s
The output to
juju status should soon look similar to this:
Model Controller Cloud/Region Version SLA Timestamp
k8s-model aws k8s-cloud 2.5.0 unsupported 18:57:16Z
App Version Status Scale Charm Store Rev OS Address Notes
gitlab-k8s active 1 gitlab-k8s jujucharms 0 kubernetes 10.152.183.184
mariadb-k8s active 1 mariadb-k8s jujucharms 0 kubernetes 10.152.183.221
Unit Workload Agent Address Ports Message
gitlab-k8s/0* active idle 10.1.11.14 80/TCP
mariadb-k8s/0* active idle 10.1.11.13 3306/TCP
Congratulations, you deployed a Kubernetes workload using dynamically provisioned volumes through the use of the AWS integrator charm!
Removing configuration and software
To remove all traces of Kubernetes and its configuration follow these steps:
juju destroy-model -y --destroy-storage k8s-model
juju remove-k8s k8s-cloud
rm -rf ~/.kube
This leaves us with Juju installed as well as an AWS controller. To remove even those things proceed as follows:
juju destroy-controller -y --destroy-all-models aws-k8s
sudo snap remove juju
Consider the following tutorials:
- Setting up static Kubernetes storage
- Using Juju with MicroK8s
- Multi-cloud controller with GKE and auto-configured storage
- Installing Kubernetes with CDK and using auto-configured storage
To gain experience with a standalone (non-Juju) MicroK8s installation check out Ubuntu tutorial Install a local Kubernetes with MicroK8s.