kubernetes-hands-on/17-volumes

Running a stateful application: volumes

Introduction

In this section you will learn how to deploy a stateful application, mysql in this example.

As you know a pod is mortal, meaning it can be destroyed by Kubernetes anytime, and with it it's local data, memory, etc. So it's perfect for stateless applications. Of course, in the real world we need a way to store our data, and we need this data to be persistent in time.

So how can we deploy a stateful application with a persistent storage in Kubernetes? Let's deploy a mysql.

Volumes

We need to review what a volume is before continuing with the deployment of our mysql. As stated above, the disk of a pod is destroyed with it, so it's lost. For a database it would be nice if we could keep the data between restarts of the pods. Here comes the volume.

We can see a pod as something that requests CPU & RAM. We can see a volume as something that requests a storage on disk. Kubernetes handles a lot of different kind of volumes - 26 as this file hands on is written - from local disk storage to s3.

Here we will use PersistentVolumeClaim, it's an abstraction over the hard drives of the Kubernetes nodes - a fancy name for local hard drive.

Let's create the volume where our mysql data will be stored.

First we create the PersistentVolume. It is a piece of storage in the cluster that has been provisioned by a cluster administrator. It is a resource in the cluster just like a node is a resource of the cluster.

apiVersion: v1
kind: PersistentVolume
metadata:
  name: mysql-pv-volume
spec:
  capacity:
    storage: 1Gi
  accessModes:
    - ReadWriteOnce
  hostPath:
    path: "/mnt/data"

Let's review some parameters:

• capacity: the capacity of the volume
  • storage: the volume size - here 1Gb
• accessModes: how this volume will be accessed, here ReadWriteOnce
  • ReadWriteOnce: the volume can be mounted as read-write by a single node
  • ReadOnlyMany: the volume can be mounted read-only by many nodes
  • ReadWriteMany: the volume can be mounted as read-write by many nodes
• hostPath: where the storage will be stored on the host, here /mnt/data

Apply it:

kubectl apply -f 01-simple-mysql-pv.yml

Now that we have a storage, we need to claim it, make it available for our pods. So we need a PersistentVolumeClaim. It is a request for storage by a user. It is similar to a pod. Pods consume node resources and PersistentVolumeClaim consume PersistentVolume resources.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: mysql-pv-claim
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 1Gi

The manifest is pretty similar to the PersistentVolume:

kubectl apply -f 02-simple-mysql-pvc.yml

Stateful application

Now let's create the deployment of mysql:

kubectl apply -f 03-simple-mysql-deployment.yml

There is a bunch of parameters we haven't seen yet:

• strategy: the strategy of updates of the pods
  • type: Recreate. This instructs Kubernetes to not use rolling updates. Rolling updates will not work, as you cannot have more than one Pod running at a time.
• env: the list of environment variables to pass to the container
  • name: the name of the env variable
  • value: the value of the env variable
• volumeMounts: the volumes, think directories, to give access to the container
  • name: the name of the volume to mount
  • mountPath: where in the container to mount the volume
• volumes: the volumes to request access to
  • name: the name of the volume, same as volumeMounts.name
  • persistentVolumeClaim: the PersistentVolumeClaim we want
    • claimName: the name of the claim

Let's finish by creating a service to have stable DNS entry inside our cluster.

kubectl apply -f 04-simple-mysql-service.yml

Finally let's access the mysql:

kubectl run -it --rm --image=mysql:5.6 --restart=Never mysql-client -- mysql -h mysql -ppassword

If you don't see a command prompt, try pressing enter.

mysql> show databases;
+--------------------+
| Database           |
+--------------------+
| information_schema |
| mysql              |
| performance_schema |
+--------------------+
3 rows in set (0.02 sec)

Create a new database in mysql:

mysql> CREATE DATABASE testing;
Query OK, 1 row affected (0.01 sec)

Now delete the service and the deployment:

kubectl delete service,deployment --all

Recreate them, reconnect to mysql and see if you still have the database testing you created.

Clean up

kubectl delete service,deployment,pvc,pv,pod --all