Restore a running mysql cluster

[cpanato]

Hello I did the following thing: 1- create a mysql cluster without a database using this sample manifest

apiVersion: mysql.presslabs.org/v1alpha1
kind: MysqlCluster
metadata:
  name: mm1
spec:
  replicas: 1
  secretName: test
  volumeSpec:
    persistentVolumeClaim:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 10Gi

2- wait for the the pod be healthy 3- updated the manifest to use a backup and apply the restore

apiVersion: mysql.presslabs.org/v1alpha1
kind: MysqlCluster
metadata:
  name: mm1
spec:
  replicas: 1
  secretName: test

  initBucketURL: s3://xxxxoooooxxxxooooo/backup11.gz
  initBucketSecretName: test-restore
  volumeSpec:
    persistentVolumeClaim:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 10Gi

I saw the pos was in terminating state and then it start a new one, however the backup was not applied.

Does this is possible to achieve?

cc @AMecea

[cpanato]

saw this: https://github.com/presslabs/mysql-operator/blob/103e6a9db5ce9de6c7017f24c5f0f6cb43cf4349/pkg/sidecar/appclone.go#L30

i tried to remove but then the pod never came up

[AMecea]

Hi @cpanato,

To init a cluster from a backup you have to set initBucketURL from the first time (when the cluster is created). The initBucketURL is used only when the disk empty.

In your case, you can delete the PVC and wait for pod to be recreated (maybe scale to 0 then remove the PVC then scale back to 1)

[cpanato]

@AMecea I will try that. will be a good addition if we have a controller that apply a restore

[cpanato]

thanks that worked :)