You are looking at the documentation of a prior release. To read the documentation of the latest release, please visit here.

New to KubeDB? Please start here.

Snapshot and Restore Using S3 Repository Plugin

The S3 repository plugin adds support for using AWS S3 as a repository for Snapshot/Restore. It also works with S3 compatible other mediums such as Linode Object Storage.

For the demo, we are going to show you how to snapshot a KubeDB managed Elasticsearch and restore data from previously taken snapshot.

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 kind.

Now, install the KubeDB operator in your cluster following the steps here.

To keep things isolated, this tutorial uses a separate namespace called demo throughout this tutorial.

$ kubectl create namespace demo
namespace/demo created

$ kubectl get namespace
NAME                 STATUS   AGE
demo                 Active   9s

Note: YAML files used in this tutorial are stored in guides/elasticsearch/quickstart/overview/yamls folder in GitHub repository kubedb/docs

Create S3 Compatible Storage

We are going to use the Linode Object Storage which is S3 compatible. But you can any S3 compatible storage which suits you best. Let’s create a sample-s3-bucket to store snapshot and later restore from it.

create sample s3 bucket

You also need to create access_key and secret_key so that your Elasticsearch Cluster can connect to the bucket.

Deploy Elasticsearch Cluster and Populate Data

For the demo, we are going to use Elasticsearch docker images from KubeDB distribution with the pre-installed S3 repository plugin.

Secure Client Settings

To make the plugin works we need to create a k8s secret with the Elastisearch secure settings:

apiVersion: v1
kind: Secret
metadata:
  name: es-secure-settings
  namespace: demo
stringData:
  password: strong-password
  s3.client.default.access_key: 6BU5GFIIUC2********
  s3.client.default.secret_key: DD1FS5NAiPf********

N.B.: Here, the password is the Elasticsearch KEYSTROE_PASSWORD, if you do not provide it, default to empty string ("").

Let’s create the k8s secret with secure settings:

$ kubectl apply -f secure-settings-secret.yaml
secret/es-secure-settings created

In S3 Client Settings, If you do not configure the endpoint, it default to s3.amazonaws.com. Since we are using Linode Bucket instead of AWS S3, we need to configure the endpoint too. Let’s create another secret with custom client configurations:

apiVersion: v1
kind: Secret
metadata:
  name: es-custom-config
  namespace: demo
stringData:
  elasticsearch.yml: |-
    s3.client.default.endpoint: us-east-1.linodeobjects.com    

N.B.: In Elasticsearch, only secure setting goes to elasticsearch.keystore, others are put into elasticsearch.yml config file. That’s why two different k8s secrets are used.

Let’s create the k8s secret with custom configurations:

$ kubectl apply -f custom-configuration.yaml
secret/es-custom-config created

Deploy Elasticsearch Cluster

Now that we have deployed our configuration secrets, it’s time to deploy our Elasticsearch instance.

apiVersion: kubedb.com/v1alpha2
kind: Elasticsearch
metadata:
  name: sample-es
  namespace: demo
spec:
  # Custom configuration, which will update elasticsearch.yml
  configSecret:
    name: es-custom-config
  # Secure settings which will be stored in elasticsearch.keystore
  secureConfigSecret:
    name: es-secure-settings
  enableSSL: true
  # we are using ElasticsearchVersion with pre-installed s3 repository plugin
  version: kubedb-xpack-7.14.0
  storageType: Durable
  replicas: 3
  storage:
    storageClassName: "standard"
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 1Gi

Let’s deploy the Elasticsearch and wait for it to become ready to use:

$ kubectl apply -f https://github.com/kubedb/docs/raw/v2023.01.17/docs/guides/elasticsearch/plugins-backup/s3-repository/yamls/elasticsearch.yaml
elasticsearch.kubedb.com/sample-es created
$ kubectl get es -n demo -w
NAME        VERSION               STATUS   AGE
sample-es   kubedb-xpack-7.14.0            0s
sample-es   kubedb-xpack-7.14.0   Provisioning   19s
sample-es   kubedb-xpack-7.14.0   Ready          41s

Populate Data

To connect to our Elasticsearch cluster, let’s port-forward the Elasticsearch service to local machine:

$ kubectl port-forward -n demo svc/sample-es 9200
Forwarding from 127.0.0.1:9200 -> 9200
Forwarding from [::1]:9200 -> 9200

Keep it like that and switch to another terminal window:

$ export ELASTIC_USER=$(kubectl get secret -n demo sample-es-elastic-cred -o jsonpath='{.data.username}' | base64 -d)

$ export ELASTIC_PASSWORD=$(kubectl get secret -n demo sample-es-elastic-cred -o jsonpath='{.data.password}' | base64 -d)

$ curl -XGET -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/_cluster/health?pretty"
{
  "cluster_name" : "sample-es",
  "status" : "green",
  "timed_out" : false,
  "number_of_nodes" : 3,
  "number_of_data_nodes" : 3,
  "active_primary_shards" : 1,
  "active_shards" : 2,
  "relocating_shards" : 0,
  "initializing_shards" : 0,
  "unassigned_shards" : 0,
  "delayed_unassigned_shards" : 0,
  "number_of_pending_tasks" : 0,
  "number_of_in_flight_fetch" : 0,
  "task_max_waiting_in_queue_millis" : 0,
  "active_shards_percent_as_number" : 100.0
}

So, our cluster status is green. Let’s create some indices with dummy data:

$ curl -XPOST -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/products/_doc?pretty" -H 'Content-Type: application/json' -d '
{
    "name": "KubeDB",
    "vendor": "AppsCode Inc.",
    "description": "Database Operator for Kubernetes"
}
'

$ curl -XPOST -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/companies/_doc?pretty" -H 'Content-Type: application/json' -d '
{
    "name": "AppsCode Inc.",
    "mission": "Accelerate the transition to Containers by building a Kubernetes-native Data Platform",
    "products": ["KubeDB", "Stash", "KubeVault", "Kubeform", "ByteBuilders"]
}
'

Now, let’s verify that the indexes have been created successfully.

$ curl -XGET -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/_cat/indices?v&s=index&pretty"
health status index            uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   .geoip_databases oiaZfJA8Q5CihQon0oR8hA   1   1         42            0     81.6mb         40.8mb
green  open   companies        GuGisWJ8Tkqnq8vhREQ2-A   1   1          1            0     11.5kb          5.7kb
green  open   products         wyu-fImDRr-Hk_GXVF7cDw   1   1          1            0     10.6kb          5.3kb

Repository Settings

The s3 repository type supports a number of settings to customize how data is stored in S3. These can be specified when creating the repository.

Let’s create the _snapshot repository sample_s3_repo with our bucket name sample-s3-bucket:

$ curl -k -X PUT -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/_snapshot/sample_s3_repo?pretty" -H 'Content-Type: application/json' -d'
{
  "type": "s3",
  "settings": {
    "bucket": "sample-s3-bucket"
  }
}
'
{
  "acknowledged" : true
}

We’ve successfully created our repository. Ready to take our first snapshot.

Create a Snapshot

A repository can contain multiple snapshots of the same cluster. Snapshots are identified by unique names within the cluster. For more details, visit Create a snapshot.

$ curl -k -X PUT -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/_snapshot/sample_s3_repo/snapshot_1?wait_for_completion=true&pretty"

{
  "snapshot" : {
    "snapshot" : "snapshot_1",
    "uuid" : "JKoF5sgtS3WPBQ8A_OvWbw",
    "repository" : "sample_s3_repo",
    "version_id" : 7140099,
    "version" : "7.14.0",
    "indices" : [
      ".geoip_databases",
      "companies",
      "products"
    ],
    "data_streams" : [ ],
    "include_global_state" : true,
    "state" : "SUCCESS",
    "start_time" : "2021-08-24T14:45:38.930Z",
    "start_time_in_millis" : 1629816338930,
    "end_time" : "2021-08-24T14:46:16.946Z",
    "end_time_in_millis" : 1629816376946,
    "duration_in_millis" : 38016,
    "failures" : [ ],
    "shards" : {
      "total" : 3,
      "failed" : 0,
      "successful" : 3
    },
    "feature_states" : [
      {
        "feature_name" : "geoip",
        "indices" : [
          ".geoip_databases"
        ]
      }
    ]
  }
}

We’ve successfully taken our first snapshot.

Delete Data and Restore a Snapshot

Let’s delete all the indices:

$ curl -k -u "$ELASTIC_USER:$ELASTIC_PASSWORD" -X DELETE "https://localhost:9200/_all?pretty"
{
  "acknowledged" : true
}

List and varify the deletion:

$ curl -XGET -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/_cat/indices?v&s=index&pretty"
health status index            uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   .geoip_databases oiaZfJA8Q5CihQon0oR8hA   1   1         42            0     81.6mb         40.8mb

For more details about restore, visit Restore a snapshot.

Let’s restore the data from our snapshot_1:

$ curl -k -u "$ELASTIC_USER:$ELASTIC_PASSWORD" -X POST "https://localhost:9200/_snapshot/sample_s3_repo/snapshot_1/_restore?pretty" -H 'Content-Type: application/json' -d'
{
  "indices": "companies,products"
}
'

{
  "accepted" : true
}

We’ve successfully restored our indices.

N.B.: We only wanted to restore the indices we created, but if you want to overwrite everything with the snapshot data, you can do it by setting include_global_state to true while restoring.

Varify Data

To varify our data, let’s list the indices:

$ curl -XGET -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/_cat/indices?v&s=index&pretty"
health status index            uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   .geoip_databases oiaZfJA8Q5CihQon0oR8hA   1   1         42            0     81.6mb         40.8mb
green  open   companies        drsv-5tvQwCcte7bkUT0uQ   1   1          1            0     11.7kb          5.8kb
green  open   products         7TXoXy5kRFiVgZDuyqffQA   1   1          1            0     10.6kb          5.3kb

Check the content inside:

$ curl -XGET -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/products/_search?pretty"
{
  "took" : 3,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 1,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "products",
        "_type" : "_doc",
        "_id" : "36SEeHsBS6UMHADkEvJw",
        "_score" : 1.0,
        "_source" : {
          "name" : "KubeDB",
          "vendor" : "AppsCode Inc.",
          "description" : "Database Operator for Kubernetes"
        }
      }
    ]
  }
}
$ curl -XGET -k -u  "$ELASTIC_USER:$ELASTIC_PASSWORD" "https://localhost:9200/companies/_search?pretty"
{
  "took" : 3,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 1,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "companies",
        "_type" : "_doc",
        "_id" : "4KSFeHsBS6UMHADkGvL5",
        "_score" : 1.0,
        "_source" : {
          "name" : "AppsCode Inc.",
          "mission" : "Accelerate the transition to Containers by building a Kubernetes-native Data Platform",
          "products" : [
            "KubeDB",
            "Stash",
            "KubeVault",
            "Kubeform",
            "ByteBuilders"
          ]
        }
      }
    ]
  }
}

So, we have successfully retored our data from the snapshot.