Challenge 08 - Storage

< Previous Challenge - Home - Next Challenge >

Introduction

Not all containers can be stateless. What happens when your application needs to have some persistent storage?

Description

In this challenge we will be creating Azure data disks and using the Kubernetes storage mechanism to make them available to be attached to running containers. This will give MongoDB a place to storage its data without losing it when the container is stopped and restarted.

• Verify what happens without persistent storage
  • Make sure that you are using the latest version of the Fabmedical container images, either the ones you built or the pre-built ones available here:
    • whatthehackmsft/content-api:v2
    • whatthehackmsft/content-web:v2
  • Delete the MongoDB pod created in Challenge 6.
    • Kubernetes should automatically create a new pod to replace it.
    • Connect to the new pod and you should observe that the “contentdb” database is no longer there since the pod was not configured with persistent storage.
  • Delete your existing MongoDB deployment
• Redeploy MongoDB with dynamic persistent storage
  • NOTE: Some types of persistent volumes (specifically, Azure disks) are associated with a single zone, see this document. Since we enabled availability zones in challenge 3, we need to guarantee that the two volumes and the node that the pod runs on are in the same zone.
    - Fortunately, the default azure-disk storage class uses the “volumeBindingMode: WaitForFirstConsumer” setting which forces kubernetes to wait for a workload to be scheduled before provisioning the disks. In other words, AKS is smart enough to create the disk in the same availability zone as the node. - You could optionally create your own storage class to specifically define the volumeBindingMode, but that’s not necessary for this exercise.
  • Create two Persistent Volume Claims (PVC) using the new Storage Class, one for data and one for config. - Look in the Resources/Challenge 8 folder for starter templates
  • Modify your MongoDB deployment to use the PVCs. - Look in the /Challenge-08 folder of the Resources.zip package for starter templates
  • Deploy MongoDB
    • Examine the automatically provisioned Persistent Volumes (PVs) and verify that both are in the same zone.
    • Check that the disk and the node that the pod runs on are in the same zone
• Verify that persistent storage works

  • Verify that MongoDB is working fine by connecting to the corresponding MongoDB Pod in interactive mode. Make sure that the disks are associated correctly (bold and italic below)

    • kubectl exec -it <mongo-db pod name> bash
    • root@mongo-db678745655b-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@mongo-db678745655b-f82vj:/# mongosh --version

        MongoDB shell version v3.6.1
        connecting to: mongodb://127.0.0.1:27017
        MongoDB server version: 3.6.1
      

  • Re-load the sample content (Speakers & Sessions data) in to MongoDB by running the content-init job as you did earlier during Challenge 7.
  • Make sure that the contentdb database is populated by connecting to the MongoDB pod with an interactive terminal and verify the database “contentdb” exists.
    • root@mongo-db678745655b-f82vj:/# 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
        contentdb   0.000GB
        local       0.000GB
      

  • Destroy the MongoDB pod to prove that the data persisting to the disk
    • kubectl delete pod <mongo-db-pod>
  • Wait for kubernetes to recreate the pod
  • Once the pod is created, verify that data is persisted by following the previous MongoDB verification step.
  • Verify the API can retrieve the data by viewing the speaker and sessions pages in the browser:
    • http://<yourWebServicePIP>:3000/speakers.html
    • http://<yourWebServicePIP>:3000/sessions.html

Success Criteria

1. Verify that speaker and session data is imported into MongoDB
2. Verify that the data isn’t lost after you stop and restart the MongoDB pod.
3. Verify that this new MongoDB instance and its data are used in the application.

Advanced Challenges

Discuss with your coach what would happen if the Node running the MongoDB pod were to crash or be shut down. Would Kubernetes be able to re-deploy the pod on a different node? How do availability zones play into this?