A persistent volume in an AKS Cluster represents the storage that has been provisioned for use with the AKS pods. Even if the pods are destroyed/recreated, the data stored in the persistent volumes will stay independent of the pod lifecycle.
In this example we will explore how to use already existing Azure disks as persistent volumes in an AKS Cluster, and use it to store the monogdb files.
We will create 2 Managed Disks and place the mongodb data and config files on the mounted Azure disks. Finally we will delete the pod and recreate it and attach the same Azure disks which hold mongodb data files and see that the data is persistent.
NOTE: All the commands given below can be executed either on the Azure Shell or the CentOS jumpbox.
We will start with deleting all the existing deployments of our demo application and making sure that the old instances of DB, WEB and API pods are not running.
kubectl get deploymentsNAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
heroes-api-deploy 1 1 1 1 22d
heroes-db-deploy 1 1 1 1 22d
heroes-web-deploy 1 1 1 0 22dDelete all deployments that start with heroes-*
kubectl delete deployments heroes-api-deploy
kubectl delete deployments heroes-db-deploy
kubectl delete deployments heroes-web-deployFor each of these commands should get a confirmation that the deployment was sucessfully deleted:
deployment.extensions "heroes-api-deploy" deleted
deployment.extensions "heroes-db-deploy" deleted
deployment.extensions "heroes-web-deploy" deletedAfter that, there shouldn't be any pods starting with heroes-* left.
kubectl get podsNo resources found.In case there are a few pods left you can delete them by using the following command:
kubectl delete pod heroes-db-deploy-... --grace-period=0 --forceBefore mounting an Azure-managed disk as a Kubernetes volume, the disk to be mounted must be in the same resource group where the AKS Node resides.
Get the Resource Group where the AKS Nodes resides by executing below command. Replace the resourcegroup name and AKS cluster name with the values from your lab.
NODEGROUP=`az resource show --resource-group <RG name of AKS cluster> --name <AKS Clustername> --resource-type Microsoft.ContainerService/managedClusters --query properties.nodeResourceGroup -o tsv`
echo $NODEGROUP
az disk create \
--resource-group $NODEGROUP \
--name mongodb-datadisk \
--size-gb 2 \
--sku Standard_LRS \
--query id --output tsv
Once the disk has been created, you should see the last portion of the output like the following. This value is the disk ID, which is used when mounting the datadisk. Keep this value copied to a notepad file.
/subscriptions/<SUBSCRIPTION_ID>/resourceGroups/MC_HackFest05_Kubecluster05_eastus/providers/Microsoft.Compute/disks/mongodb-datadisk
az disk create \
--resource-group $NODEGROUP \
--name mongodb-configdisk \
--size-gb 2 \
--sku Standard_LRS \
--query id --output tsv
Once the disk has been created, copy the disk ID to a notepad, which is used when mounting the configdisk.
/subscriptions/<SUBSCRIPTION_ID>/resourceGroups/MC_HackFest05_Kubecluster05_eastus/providers/Microsoft.Compute/disks/mongodb-configdisk
Mount the Azure disks into your pod by configuring the volume in the deployment specification.
Create a new file named heroes-db-azdisk.yaml under ~/blackbelt-aks-hackfest/labs/helper-files/ with the following contents in your CentOS jumpbox/Azure Shell.
NOTE: Update the ACR server name and the diskURI for both datadisk and configdisk with the disk IDs obtained while creating the respective disks. Also, take note of the mountPath, which is the path where the Azure disk is mounted inside the heroes-db pod.
apiVersion: v1
kind: Service
metadata:
name: mongodb
labels:
name: mongodb
spec:
type: ClusterIP
ports:
- name: http
port: 27017
targetPort: 27017
selector:
name: heroes-db
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
name: heroes-db-deploy
labels:
name: heroes-db
spec:
strategy:
rollingUpdate:
maxSurge: 1
maxUnavailable: 1
type: RollingUpdate
template:
metadata:
labels:
name: heroes-db
spec:
imagePullSecrets:
- name: acr-secret
containers:
- image: <acrname>.azurecr.io/azureworkshop/rating-db:v1
name: heroes-db-cntnr
resources:
requests:
cpu: "20m"
memory: "55M"
ports:
- containerPort: 27017
name: heroes-db
volumeMounts:
- mountPath: /data/db
name: azuredisk-db
- mountPath: /data/configdb
name: azuredisk-configdb
imagePullPolicy: Always
volumes:
- name: azuredisk-db
azureDisk:
kind: Managed
diskName: mongodb-datadisk
diskURI: <datadisk ID>
- name: azuredisk-configdb
azureDisk:
kind: Managed
diskName: mongodb-configdisk
diskURI: <configdisk ID>
restartPolicy: AlwaysSave the heroes-db-azdisk.yaml file.
Apply the yaml file to create the heroes-db pod with mounted Azure Disks.
kubectl apply -f heroes-db-azdisk.yamlEdit the heroes-web-api.yaml file and make sure that the MONGO_URI is pointing to the Mongodb running in the heroes-db pod.
env:
- name: MONGODB_URI
value: mongodb://mongodb:27017/webratings
ports:Apply the yaml file to create the heroes-web and heroes-api pods
kubectl apply -f heroes-web-api.yamlWait for all the pods to be in the Running status.
kubectl get podNAME READY STATUS RESTARTS AGE
heroes-api-deploy-77f5fdcbb-xxq46 1/1 Running 0 8m
heroes-db-deploy-678745655b-f82vj 1/1 Running 0 4m
heroes-web-deploy-5dffc9c976-cdll5 1/1 Running 0 8mVerify the mount points of azure disks inside the DB pod with the below commands: NOTE: You should see 2 disks of 2GB each mounted to the /data/db and /data/configdb paths.
kubectl exec -it <DB pod name> bashroot@heroes-db-deploy-678745655b-f82vj:/# df -Th
Filesystem Type Size Used Avail Use% Mounted on
overlay overlay 30G 4.2G 25G 15% /
tmpfs tmpfs 1.7G 0 1.7G 0% /dev
tmpfs tmpfs 1.7G 0 1.7G 0% /sys/fs/cgroup
/dev/sdc ext4 2.0G 304M 1.5G 17% /data/db
/dev/sdd ext4 2.0G 3.0M 1.8G 1% /data/configdb
/dev/sda1 ext4 30G 4.2G 25G 15% /etc/hosts
shm tmpfs 64M 0 64M 0% /dev/shm
tmpfs tmpfs 1.7G 12K 1.7G 1% /run/secrets/kubernetes.io/serviceaccount
tmpfs tmpfs 1.7G 0 1.7G 0% /sys/firmware
root@heroes-db-deploy-678745655b-f82vj:/#Run the mongo command and list the databases by running the command 'show dbs' in the mongo shell.
root@heroes-db-deploy-678745655b-vq7l5:/# mongo
MongoDB shell version v3.6.1
connecting to: mongodb://127.0.0.1:27017
MongoDB server version: 3.6.1
>
> show dbs
admin 0.000GB
config 0.000GB
local 0.000GB
>
At this point there will be only 3 default databases namely admin, local and config. Type exit to exit the mongo shell
Run the import.sh file to create the webratings database and load it with initial data.
root@heroes-db-deploy-678745655b-f82vj:/#cd /
root@heroes-db-deploy-678745655b-f82vj:/# ./import.sh
2018-07-02T11:48:16.546+0000 connected to: localhost
2018-07-02T11:48:16.608+0000 imported 4 documents
2018-07-02T11:48:16.617+0000 connected to: localhost
2018-07-02T11:48:16.710+0000 imported 72 documents
2018-07-02T11:48:16.719+0000 connected to: localhost
2018-07-02T11:48:16.787+0000 imported 2 documentsRun the mongo command again and list the databases by running the command 'show dbs' in the mongo shell.
root@heroes-db-deploy-678745655b-vq7l5:/# mongo
MongoDB shell version v3.6.1
connecting to: mongodb://127.0.0.1:27017
MongoDB server version: 3.6.1
>
> show dbs
admin 0.000GB
config 0.000GB
local 0.000GB
webratings 0.000GB
>This time, you will see the webratings database also listed in the output.
NOTE: The imported webratings database information will be stored in the mounted Azure disks.
Type 'exit' 2 times to exit from the mongo shell and the db pod.
Browse the heroes web application and add some ratings. Make a note of the current ratings.
Now delete the database pod deployment
kubectl delete deployment heroes-db-deployBrowse the heroes web application and refresh the ratings. You will not be able to see the ratings now.
Now, again apply the yaml file for the db pod, heroes-db-azdisk.yaml to recreate the DB pod.
This will create a fresh heroes-db pod with the same Azure disks mounted.
kubectl apply -f heroes-db-azdisk.yaml Wait for the pod to be started and verify the mount points of azure disks inside the newly created DB pod.
[root@CentoS01 helper-files]# kubectl exec -it <db pod> bash
root@heroes-db-deploy-678745655b-f82vj:/# df -Th
Filesystem Type Size Used Avail Use% Mounted on
overlay overlay 30G 4.2G 25G 15% /
tmpfs tmpfs 1.7G 0 1.7G 0% /dev
tmpfs tmpfs 1.7G 0 1.7G 0% /sys/fs/cgroup
/dev/sdc ext4 2.0G 304M 1.5G 17% /data/db
/dev/sdd ext4 2.0G 3.0M 1.8G 1% /data/configdb
/dev/sda1 ext4 30G 4.2G 25G 15% /etc/hosts
shm tmpfs 64M 0 64M 0% /dev/shm
tmpfs tmpfs 1.7G 12K 1.7G 1% /run/secrets/kubernetes.io/serviceaccount
tmpfs tmpfs 1.7G 0 1.7G 0% /sys/firmware
root@heroes-db-deploy-678745655b-f82vj:/#Run the mongo command and list the databases. The DB pod should now automatically use the database files stored in the mounted Azure disks and will populate the webratings database.
root@heroes-db-deploy-678745655b-vq7l5:/# mongo
MongoDB shell version v3.6.1
connecting to: mongodb://127.0.0.1:27017
MongoDB server version: 3.6.1
>
> show dbs
admin 0.000GB
config 0.000GB
local 0.000GB
webratings 0.000GB
>Browse the heroes web application and check the ratings. You will see the same ratings which you noted earlier.