CloudDiscordGitHub

Migrate Memphis storage between storageClass's

This tutorial guides you on migrating Persistent Volumes (PVs) from the gp2 storageClass to the new default gp3 storageClass. Before implementing this process in production, it is advisable to perform a thorough test in a testing environment.

Important:

  • The migration process includes PVC removal and heavily relies on replication between brokers. Before beginning, validate that there are no errors or warnings.

  • To avoid data loss, ensure all critical stations with data are configured with at least 3 replicas.

  • Every step should be executed with continuous log monitoring, paying extra attention to the replication process during the broker startup.

Step 1: Create a new storageClass.yaml file with the required configuration. 

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: gp3-new
  annotations:
    storageclass.kubernetes.io/is-default-class: 'true'
provisioner: ebs.csi.aws.com
volumeBindingMode: WaitForFirstConsumer
parameters:
  csi.storage.k8s.io/fstype: xfs
  type: gp3

Step 2: Apply the New StorageClass Configuration

Begin by applying the configuration for the new storageClass:

kubectl apply -f storageClass.yaml

Step 3:Edit the Old StorageClass Configuration

Edit the old storageClass to change it from the default setting. In this example, the old storageClass is named "gp2."

kubectl edit storageclasses.storage.k8s.io gp2

Make the following change in the configuration:

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"storage.k8s.io/v1","kind":"StorageClass","metadata":{"annotations":{"storageclass.kubernetes.io/is-default-class":"true"},"name":"gp2"},"parameters":{"fsType":"ext4","type":"gp2"},"provisioner":"kubernetes.io/aws-ebs","volumeBindingMode":"WaitForFirstConsumer"}
    storageclass.kubernetes.io/is-default-class: "false" ## Change from "true" to "false"
  creationTimestamp: "2022-02-01T12:07:20Z"
  labels:
    k8slens-edit-resource-version: v1
  name: gp2
  resourceVersion: "214718038"
  uid: 5c09a9e5-06a4-4436-bda0-250b279e0475
parameters:
  fsType: ext4
  type: gp2
provisioner: kubernetes.io/aws-ebs
reclaimPolicy: Delete
volumeBindingMode: WaitForFirstConsumer

Step 4: Verify the Configuration

Check the updated storageClass configuration:

kubectl get storageclasses.storage.k8s.io
NAME               PROVISIONER             RECLAIMPOLICY   VOLUMEBINDINGMODE      ALLOWVOLUMEEXPANSION   AGE
gp2                kubernetes.io/aws-ebs   Delete          WaitForFirstConsumer   false                  614d
gp3-ol             kubernetes.io/aws-ebs   Delete          WaitForFirstConsumer   true                   328d
gp3-new (default)  ebs.csi.aws.com         Delete          WaitForFirstConsumer   true                   248d

Step 5: Update PVCs

Delete the PVC (data-memphis-metadata-0) associated with the old storageClass. The pod itself will also be deleted, resulting in a new PVC assigned to the new storageClass

Ensure the metadata finishes the sync process before going to the next pod.

$ kubectl get pvc -n memphis
NAME                       STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
data-memphis-metadata-0    Bound    pvc-b946195d-b1c9-42f1-8541-a650ed37b82d   8Gi        RWO            gp2            36m
data-memphis-metadata-1    Bound    pvc-d30e2baf-a74b-48b4-b1f9-d49c7366501c   8Gi        RWO            gp2            35m
memphis-js-pvc-memphis-0   Bound    pvc-d34787a7-9af2-49c4-b2d2-44418ff9beb2   30Gi       RWO            gp2            31m
memphis-js-pvc-memphis-1   Bound    pvc-888a916e-57a4-4973-bf79-86b924c5e927   30Gi       RWO            gp2            29m
memphis-js-pvc-memphis-2   Bound    pvc-f7bf4f73-69d3-4a5e-89ec-754fd21616ba   30Gi       RWO            gp2            27m

Delete resources:

$ kubectl delete pvc data-memphis-metadata-0 -n memphis
persistentvolumeclaim "data-memphis-metadata-0" deleted
^C%
$ kubectl delete pod memphis-metadata-0 -n memphis
pod "memphis-metadata-0" deleted

Check the status of the PVCs:

$ kubectl get pvc -n memphis
NAME                       STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
data-memphis-metadata-0    Bound    pvc-b946195d-b1c9-42f1-8541-a650ed37b82d   8Gi        RWO            gp3-new        36m
data-memphis-metadata-1    Bound    pvc-d30e2baf-a74b-48b4-b1f9-d49c7366501c   8Gi        RWO            gp2            35m
memphis-js-pvc-memphis-0   Bound    pvc-d34787a7-9af2-49c4-b2d2-44418ff9beb2   30Gi       RWO            gp2            31m
memphis-js-pvc-memphis-1   Bound    pvc-888a916e-57a4-4973-bf79-86b924c5e927   30Gi       RWO            gp2            29m
memphis-js-pvc-memphis-2   Bound    pvc-f7bf4f73-69d3-4a5e-89ec-754fd21616ba   30Gi       RWO            gp2            27m

Step 7: Continue With Other Stateful Resources

Proceed with the migration for the remaining stateful resources. Remember to delete them individually and validate the sync status before each step.

Important!!! Deleting and validating stateful resources one by one is crucial for a smooth migration process.

Last updated

Was this helpful?