Persistent Volumes

Objective: Make data persistent in OKD.

Now the game servers are running but every time a new deployment of the applications is started all of the data is reset. This means that the server configuration, plugins, and worlds are all erased when there is a new deployment.

This may seem strange but that is because containers, which make up the cluster, are ephemeral by nature; meaning that any data they hold is temporary. This is perfect for services that are stateless, or don't need to store data, like web servers, but not so good for stateful services like databases or well, Minecraft servers.

Persistent Volumes

So our Minecraft server is stateful. How do we save the data? In the containerized world there are a few ways to do this but the way we are going to use is persistent volumes.

A volume is just a place to store data. When you tell a container to use a volume you are mounting that volume to the container.

In the OpenShift web console navigate to: Application -> Deployments -> hub -> Configuration

Here you will find an overview of the deployment configuration for the hub pod. A deployment configuration (dc) contains instructions that tell the cluster how to rollout a pod. Under Volumes you can see the volumes available to the pod. Notice that the type field it says empty dir. This means that OpenShift has no where to save data during a new deployment so all data in the container will be lost.

To change this, you can create a volume that is persistent and modify the deployment config to use that volume.

Take note of the volume name that mounts to the directory /data in the pod.

---

Creating Persistent Volumes

There are a ton of ways you can setup the storage for a volume. You could use NFS, or GlusterFS, even something like Amazon's Elastic Block Storage (AWS EBS). In this exercise we will be simply be creating a volume backend directory on the local Minishift virtual machine.

Creating the directory:

The first thing you will need to do is create a folder on the Minishift virtual machine. First connect to the vm by typing minishift ssh into the terminal. Next type sudo -i to gain super user privileges

You will now create the folder for the volume to mount to

Create a new directory called hub inside a directory called minecraft-server.

mkdir -p /mnt/sda1/var/lib/minishift/openshift.local.volumes/pv/minecraft-server/hub

Then set it's permission to rwx for user group and other.

chmod 777 -R /mnt/sda1/var/lib/minishift/openshift.local.volumes/pv/minecraft-server/hub

Now that the directory is accessible you there is one more thing you have to do. SELinux, by default doesn't allow container volumes to access storage on their host machine. So you have to change the context of the folder to allow it.

chcon -Rt svirt_sandbox_file_t /mnt/sda1/var/lib/minishift/openshift.local.volumes/pv

---

Persistent Volume:

Next you have to mount a PersistentVolume to the directory you just created making is accessible by OKD.

• Exit minishift ssh back to your machine.

• Login to the cluster with oc login -u system:admin

  • If the oc command isn't recognized type minishift oc-env and see the command to add it to add it to your $PATH

• Once logged in switch to the minecraft-server project.

• Create and edit a new file called hubpv.yml.

  • Here you will define the persistent volume.
  • This definition will mount the folder you just created as persistent volume called hub.
  • Add the following to mount the folder you created as a persistent volume:

apiVersion: v1
kind: PersistentVolume
metadata:
 name: hub
spec:
 capacity:
  storage: 5Gi
 accessModes:
  - ReadWriteOnce
 hostPath:
  path: /mnt/sda1/var/lib/minishift/openshift.local.volumes/pv/minecraft-server/hub

• Now add this definition for the persistent volume to the minecraft-server project

  oc create -f hubpv.yml

---

Persistent Volume Claim:

Now there is a volume resource sitting out there in the minecraft-project ether.

To use this volume you first have to claim it. Using a PersistentVolumeClaim(pvc) you can request the PersistentVolume resource. Then use it to claim as a volume in a pod making OpenShift take the volume backing the claim, mounting it to the pod.

• First you'll need to create another new file called hubpvc.yml
• Edit this file and enter the following:

kind: PersistentVolumeClaim
apiVersion: v1
metadata:
 name: hubclaim
spec:
 accessModes:
   - ReadWriteOnce
 resources:
  requests:
   storage: 5Gi
selector:
 name: hub

• Save and quit that file then create a new persistent claim using this definition:

oc create -f hubpvc.yml

---

Patching the Deployment Config

Now that you've made the pvc you have to add it to the hub pod's deployment config.

oc set volume dc/hub --add --name=hub-storage -t pvc --claim-name=hubclaim --overwrite

Be sure to replace hub-storage with the volume you noted earlier from the deployment config that mounts to /data in the pod.

The above command modifies the current deployment config of hub (--overwrite).

It makes the volume called hub-storage(--name=hub-storage) use the persistent volume claim(-t pvc) called hubclaim(--claim-name=hubclaim).

Container Data is Persistent

It took a few steps but now you have made the /data directory in this pod persistent!

To go over the steps in brief again you:

1. Added a local volume backend directory
2. Created a PersistentVolume
3. Created a PersistentVolumeClaim
4. Added the PersistentVolumeClaim to the deployment config of the hub pod

SMP and Bungee

Now repeat these steps for the other pods smp, bungee changing the name hub appropriately

For bungee the container directory is /server not /data