Backup and Restore Elasticsearch database using Stash

Stash 0.9.0+ supports backup and restoration of Elasticsearch clusters. This guide will show you how you can backup and restore your Elasticsearch database with Stash.

Before You Begin

• At first, you need to have a Kubernetes cluster, and the kubectl command-line tool must be configured to communicate with your cluster. If you do not already have a cluster, you can create one by using Minikube.
• Install Stash in your cluster following the steps here.
• Install Elasticsearch addon for Stash following the steps here.
• Install KubeDB in your cluster following the steps here. This step is optional. You can deploy your database using any method you want. We are using KubeDB because KubeDB simplifies many of the difficult or tedious management tasks of running a production grade databases on private and public clouds.
• If you are not familiar with how Stash backup and restore Elasticsearch databases, please check the following guide here.

You have to be familiar with following custom resources:

• AppBinding
• Function
• Task
• BackupConfiguration
• RestoreSession

To keep things isolated, we are going to use a separate namespace called demo throughout this tutorial. Create demo namespace if you haven’t created yet.

$ kubectl create ns demo
namespace/demo created

Note: YAML files used in this tutorial are stored here.

Backup Elasticsearch

This section will demonstrate how to backup an Elasticsearch database. Here, we are going to deploy an Elasticsearch database using KubeDB. Then, we are going to backup this database into a GCS bucket. Finally, we are going to restore the backed up data into another Elasticsearch database.

Deploy Sample Elasticsearch Database

Let’s deploy a sample Elasticsearch database and insert some data into it.

Create Elasticsearch CRD:

Below is the YAML of a sample Elasticsearch crd that we are going to create for this tutorial:

apiVersion: kubedb.com/v1alpha1
kind: Elasticsearch
metadata:
  name: sample-elasticsearch
  namespace: demo
spec:
  version: "6.4.0"
  storageType: Durable
  storage:
    storageClassName: "standard"
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 1Gi
  terminationPolicy: Delete

Create the above Elasticsearch crd,

$ kubectl apply -f https://github.com/stashed/elasticsearch/raw/6.4.0/docs/examples/backup/elasticsearch.yaml
elasticsearch.kubedb.com/sample-elasticsearch created

KubeDB will deploy an Elasticsearch database according to the above specification. It will also create the necessary secrets and services to access the database.

Let’s check if the database is ready to use,

$ kubectl get es -n demo sample-elasticsearch
NAME                   VERSION       STATUS    AGE
sample-elasticsearch   6.4.0         Running   3m35s

The database is Running. Verify that KubeDB has created a Secret and a Service for this database using the following commands,

$ kubectl get secret -n demo -l=kubedb.com/name=sample-elasticsearch
NAME                        TYPE     DATA   AGE
sample-elasticsearch-auth   Opaque   9      5m14s
sample-elasticsearch-cert   Opaque   4      5m14s

$ kubectl get service -n demo -l=kubedb.com/name=sample-elasticsearch
NAME                          TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)    AGE
sample-elasticsearch          ClusterIP   10.102.109.209   <none>        9200/TCP   5m36s
sample-elasticsearch-master   ClusterIP   10.107.148.74    <none>        9300/TCP   5m36s

Here, we have to use service sample-elasticsearch and secret sample-elasticsearch-auth to connect with the database. KubeDB creates an AppBinding crd that holds the necessary information to connect with the database.

Verify AppBinding:

Verify that the AppBinding has been created successfully using the following command,

$ kubectl get appbindings -n demo
NAME                   AGE
sample-elasticsearch   4m16s

Let’s check the YAML of the above AppBinding,

$ kubectl get appbindings -n demo sample-elasticsearch -o yaml

apiVersion: appcatalog.appscode.com/v1alpha1
kind: AppBinding
metadata:
  labels:
    app.kubernetes.io/component: database
    app.kubernetes.io/instance: sample-elasticsearch
    app.kubernetes.io/managed-by: kubedb.com
    app.kubernetes.io/name: elasticsearch
    app.kubernetes.io/version: 6.4.0
    kubedb.com/kind: Elasticsearch
    kubedb.com/name: sample-elasticsearch
  name: sample-elasticsearch
  namespace: demo
spec:
  clientConfig:
    service:
      name: sample-elasticsearch
      port: 9200
      scheme: http
  secret:
    name: sample-elasticsearch-auth
  secretTransforms:
  - renameKey:
      from: ADMIN_USERNAME
      to: username
  - renameKey:
      from: ADMIN_PASSWORD
      to: password
  type: kubedb.com/elasticsearch
  version: "6.4.0"

Stash uses the AppBinding crd to connect with the target database. It requires the following two fields to set in AppBinding’s Spec section.

• spec.clientConfig.service.name specifies the name of the service that connects to the database.
• spec.secret specifies the name of the secret that holds necessary credentials to access the database.
• spec.type specifies the types of the app that this AppBinding is pointing to. KubeDB generated AppBinding follows the following format: <app group>/<app resource type>.

Creating AppBinding Manually:

If you deploy Elasticsearch database without KubeDB, you have to create the AppBinding crd manually in the same namespace as the service and secret of the database.

The following YAML shows a minimal AppBinding specification that you have to create if you deploy Elasticsearch database without KubeDB.

apiVersion: appcatalog.appscode.com/v1alpha1
kind: AppBinding
metadata:
  name: my-custom-appbinding
  namespace: my-database-namespace
spec:
  clientConfig:
    service:
      name: my-database-service
      port: 9200
      scheme: http
  secret:
    name: my-database-credentials-secret
  # type field is optional. you can keep it empty.
  # if you keep it emtpty then the value of TARGET_APP_RESOURCE variable
  # will be set to "appbinding" during auto-backup.
  type: elasticsearch

Connection information:

• Address: localhost:9200
• Username: Run following command to get username

$ kubectl get secrets -n demo sample-elasticsearch-auth -o jsonpath='{.data.\ADMIN_USERNAME}' | base64 -d
admin

• Password: Run the following command to get the password

$ kubectl get secrets -n demo sample-elasticsearch-auth -o jsonpath='{.data.\ADMIN_PASSWORD}' | base64 -d
mti4sbyy

Insert Sample Data:

Now, we are going to exec into the database pod and create some sample data. At first, find out the database pod using the following command,

$ kubectl get pods -n demo --selector="kubedb.com/name=sample-elasticsearch"
NAME                     READY   STATUS    RESTARTS   AGE
sample-elasticsearch-0   1/1     Running   0          7m33s

Now, let’s exec into the pod and create a table,

$ kubectl exec -it -n demo sample-elasticsearch-0 bash
~ curl -XPUT --user "admin:mti4sbyy" "localhost:9200/test/snapshot/1?pretty" -H 'Content-Type: application/json' -d'
{
    "title": "Snapshot",
    "text":  "Testing instant backup",
    "date":  "2018/02/13"
}
'

~ curl -XGET --user "admin:mti4sbyy" "localhost:9200/test/snapshot/1?pretty"
{
  "_index" : "test",
  "_type" : "snapshot",
  "_id" : "1",
  "_version" : 1,
  "found" : true,
  "_source" : {
    "title" : "Snapshot",
    "text" : "Testing instant backup",
    "date" : "2018/02/13"
  }
}

Now, we are ready to backup this sample database.

Prepare Backend

We are going to store our backed up data into a GCS bucket. At first, we need to create a secret with GCS credentials then we need to create a Repository crd. If you want to use a different backend, please read the respective backend configuration doc from here.

Create Storage Secret:

Let’s create a secret called gcs-secret with access credentials to our desired GCS bucket,

$ echo -n 'changeit' > RESTIC_PASSWORD
$ echo -n '<your-project-id>' > GOOGLE_PROJECT_ID
$ cat downloaded-sa-json.key > GOOGLE_SERVICE_ACCOUNT_JSON_KEY
$ kubectl create secret generic -n demo gcs-secret \
    --from-file=./RESTIC_PASSWORD \
    --from-file=./GOOGLE_PROJECT_ID \
    --from-file=./GOOGLE_SERVICE_ACCOUNT_JSON_KEY
secret/gcs-secret created

Create Repository:

Now, crete a Respository using this secret. Below is the YAML of Repository crd we are going to create,

apiVersion: stash.appscode.com/v1alpha1
kind: Repository
metadata:
  name: gcs-repo
  namespace: demo
spec:
  backend:
    gcs:
      bucket: appscode-qa
      prefix: /demo/elasticsearch/sample-elasticsearch
    storageSecretName: gcs-secret

Let’s create the Repository we have shown above,

$ kubectl apply -f https://github.com/stashed/elasticsearch/raw/6.4.0/docs/examples/backup/repository.yaml
repository.stash.appscode.com/gcs-repo created

Now, we are ready to backup our database to our desired backend.

Backup

We have to create a BackupConfiguration targeting respective AppBinding crd of our desired database. Then Stash will create a CronJob to periodically backup the database.

Create BackupConfiguration:

Below is the YAML for BackupConfiguration crd to backup the sample-elasticsearch database we have deployed earlier.,

apiVersion: stash.appscode.com/v1beta1
kind: BackupConfiguration
metadata:
  name: sample-elasticsearch-backup
  namespace: demo
spec:
  schedule: "*/5 * * * *"
  task:
    name: elasticsearch-backup-6.4.0
  repository:
    name: gcs-repo
  target:
    ref:
      apiVersion: appcatalog.appscode.com/v1alpha1
      kind: AppBinding
      name: sample-elasticsearch
  interimVolumeTemplate:
    metadata:
      name: stash-tmp-storage
    spec:
      accessModes: [ "ReadWriteOnce" ]
      storageClassName: "standard"
      resources:
        requests:
          storage: 1Gi
  retentionPolicy:
    name: keep-last-5
    keepLast: 5
    prune: true

Here,

• spec.schedule specifies that we want to backup the database at 5 minutes interval.
• spec.task.name specifies the name of the task crd that specifies the necessary Function and their execution order to backup an Elasticsearch database.
• spec.target.ref refers to the AppBinding crd that was created for sample-elasticsearch database.
• spec.interimVolumeTemplate specifies a PVC template where the dumped data will be stored temporarily before uploading to the backend.

Let’s create the BackupConfiguration crd we have shown above,

$ kubectl apply -f https://github.com/stashed/elasticsearch/raw/6.4.0/docs/examples/backup/backupconfiguration.yaml
backupconfiguration.stash.appscode.com/sample-elasticsearch-backup created

Verify CronJob:

If everything goes well, Stash will create a CronJob with the schedule specified in spec.schedule field of BackupConfiguration crd.

Verify that the CronJob has been created using the following command,

$ kubectl get cronjob -n demo
NAME                          SCHEDULE      SUSPEND   ACTIVE   LAST SCHEDULE   AGE
sample-elasticsearch-backup   */5 * * * *   False     0        <none>          10s

Wait for BackupSession:

The sample-elasticsearch-backup CronJob will trigger a backup on each scheduled slot by creating a BackupSession crd.

Wait for the next schedule. Run the following command to watch BackupSession crd,

$ kubectl get backupsession -n demo -w
NAME                                     INVOKER-TYPE          INVOKER-NAME                  PHASE       AGE
sample-elasticsearch-backup-1570098367   BackupConfiguration   sample-elasticsearch-backup   Running     9s
sample-elasticsearch-backup-1570098367   BackupConfiguration   sample-elasticsearch-backup   Running     65s
sample-elasticsearch-backup-1570098367   BackupConfiguration   sample-elasticsearch-backup   Succeeded   79s

We can see above that the backup session has succeeded. Now, we are going to verify that the backed up data has been stored in the backend.

Verify Backup:

Once a backup is complete, Stash will update the respective Repository crd to reflect the backup. Check that the repository gcs-repo has been updated by the following command,

$ kubectl get repository -n demo gcs-repo
NAME       INTEGRITY   SIZE        SNAPSHOT-COUNT   LAST-SUCCESSFUL-BACKUP   AGE
gcs-repo   true        1.030 KiB   1                39s                      3m20s

Now, if we navigate to the GCS bucket, we are going to see backed up data has been stored in demo/elasticsearch/sample-elasticsearch directory as specified by spec.backend.gcs.prefix field of Repository crd.

Note: Stash keeps all the backed up data encrypted. So, data in the backend will not make any sense until they are decrypted.

Restore Elasticsearch

In this section, we are going to restore the database from the backup we have taken in the previous section. We are going to deploy a new database and initialize it from the backup.

Stop Taking Backup of the Old Database:

At first, let’s stop taking any further backup of the old database so that no backup is taken during restore process. We are going to pause the BackupConfiguration crd that we had created to backup the sample-elasticsearch database. Then, Stash will stop taking any further backup for this database.

Let’s pause the sample-elasticsearch-backup BackupConfiguration,

$ kubectl patch backupconfiguration -n demo sample-elasticsearch-backup --type="merge" --patch='{"spec": {"paused": true}}'
backupconfiguration.stash.appscode.com/sample-elasticsearch-backup patched

Now, wait for a moment. Stash will pause the BackupConfiguration. Verify that the BackupConfiguration has been paused,

$  kubectl get backupconfiguration -n demo sample-elasticsearch-backup
NAME                          TASK                         SCHEDULE      PAUSED   AGE
sample-elasticsearch-backup   elasticsearch-backup-6.4.0   */5 * * * *   true     3m8s

Notice the PAUSED column. Value true for this field means that the BackupConfiguration has been paused.

Deploy Restored Database:

Now, we have to deploy the restored database similarly as we have deployed the original sample-psotgres database. However, this time there will be the following differences:

• We have to use the same secret that was used in the original database. We are going to specify it using spec.databaseSecret field.
• We have to specify spec.init section to tell KubeDB that we are going to use Stash to initialize this database from backup. KubeDB will keep the database phase to Initializing until Stash finishes its initialization.

Below is the YAML for Elasticsearch crd we are going deploy to initialize from backup,

apiVersion: kubedb.com/v1alpha1
kind: Elasticsearch
metadata:
  name: restored-elasticsearch
  namespace: demo
spec:
  version: "6.4.0"
  storageType: Durable
  databaseSecret:
    secretName: sample-elasticsearch-auth # use same secret as original the database
  storage:
    storageClassName: "standard"
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 1Gi
  init:
    stashRestoreSession:
      name: sample-elasticsearch-restore
  terminationPolicy: Delete

Here,

• spec.init.stashRestoreSession.name specifies the RestoreSession crd name that we are going to use to restore this database.

Let’s create the above database,

$ kubectl apply -f https://github.com/stashed/elasticsearch/raw/6.4.0/docs/examples/restore/restored-elasticsearch.yaml
elasticsearch.kubedb.com/restored-elasticsearch created

If you check the database status, you will see it is stuck in Initializing state.

$ kubectl get es -n demo restored-elasticsearch
NAME                     VERSION       STATUS         AGE
restored-elasticsearch   6.4.0         Initializing   38s

Create RestoreSession:

Now, we need to create a RestoreSession crd pointing to the AppBinding for this restored database.

Check AppBinding has been created for the restored-elasticsearch database using the following command,

$ kubectl get appbindings -n demo restored-elasticsearch
NAME                     AGE
restored-elasticsearch   29s

If you are not using KubeDB to deploy database, create the AppBinding manually.

Below is the YAML for the RestoreSession crd that we are going to create to restore backed up data into restored-elasticsearch database.

apiVersion: stash.appscode.com/v1beta1
kind: RestoreSession
metadata:
  name: sample-elasticsearch-restore
  namespace: demo
  labels:
    kubedb.com/kind: Elasticsearch # this label is mandatory if you are using KubeDB to deploy the database. Otherwise, Elasticsearch crd will be stuck in `Initializing` phase.
spec:
  task:
    name: elasticsearch-restore-6.4.0
  repository:
    name: gcs-repo
  target:
    ref:
      apiVersion: appcatalog.appscode.com/v1alpha1
      kind: AppBinding
      name: restored-elasticsearch
  interimVolumeTemplate:
    metadata:
      name: stash-tmp-storage
    spec:
      accessModes: [ "ReadWriteOnce" ]
      storageClassName: "standard"
      resources:
        requests:
          storage: 1Gi
  rules:
  - snapshots: [latest]

Here,

• metadata.labels specifies a kubedb.com/kind: Elasticsearch label that is used by KubeDB to watch this RestoreSession.
• spec.task.name specifies the name of the Task crd that specifies the Functions and their execution order to restore an Elasticsearch database.
• spec.repository.name specifies the Repository crd that holds the backend information where our backed up data has been stored.
• spec.target.ref refers to the AppBinding crd for the restored-elasticsearch database.
• spec.interimVolumeTemplate specifies a PVC template to store the restored data temporarily before inserting into the targeted Elasticsearch database.
• spec.rules specifies that we are restoring from the latest backup snapshot of the database.

Warning: Label kubedb.com/kind: Elasticsearch is mandatory if you are using KubeDB to deploy the database. Otherwise, the database will be stuck in Initializing state.

Let’s create the RestoreSession crd we have shown above,

$ kubectl apply -f https://github.com/stashed/elasticsearch/raw/6.4.0/docs/examples/restore/restoresession.yaml
restoresession.stash.appscode.com/sample-elasticsearch-restore created

Once, you have created the RestoreSession crd, Stash will create a job to restore. We can watch the RestoreSession phase to check if the restore process is succeeded or not.

Run the following command to watch RestoreSession phase,

$ kubectl get restoresession -n demo sample-elasticsearch-restore -w
NAME                           REPOSITORY-NAME   PHASE       AGE
sample-elasticsearch-restore   gcs-repo          Running     5s
sample-elasticsearch-restore   gcs-repo          Succeeded   43s

So, we can see from the output of the above command that the restore process succeeded.

Verify Restored Data:

In this section, we are going to verify that the desired data has been restored successfully. We are going to connect to the database and check whether the table we had created in the original database is restored or not.

At first, check if the database has gone into Running state by the following command,

$ kubectl get es -n demo restored-elasticsearch
NAME                     VERSION       STATUS    AGE
restored-elasticsearch   6.4.0         Running   2m16s

Now, find out the database pod by the following command,

$ kubectl get pods -n demo --selector="kubedb.com/name=restored-elasticsearch"
NAME                       READY   STATUS    RESTARTS   AGE
restored-elasticsearch-0   1/1     Running   0          48m

Now, exec into the database pod and list available tables,

$ kubectl exec -it -n demo restored-elasticsearch-0 bash

~ curl -XGET --user "admin:mti4sbyy" "localhost:9200/test/snapshot/1?pretty"
{
  "_index" : "test",
  "_type" : "snapshot",
  "_id" : "1",
  "_version" : 1,
  "found" : true,
  "_source" : {
    "title" : "Snapshot",
    "text" : "Testing instand backup",
    "date" : "2018/02/13"
  }
}

So, from the above output, we can see the document test that we had created in the original database sample-elasticsearch is restored in the restored database restored-elasticsearch.

Cleanup

To cleanup the Kubernetes resources created by this tutorial, run:

kubectl delete -n demo restoresession sample-elasticsearch-restore
kubectl delete -n demo backupconfiguration sample-elasticsearch-backup
kubectl delete -n demo es sample-elasticsearch restored-elasticsearch
kubectl delete -n demo repository gcs-repo