The backup & restoration procedure described below is based on the Opensearch snapshot mechanism.

Output directory

Set the directory where snapshots will be saved, add to the ${OPENSEARCH_HOME}/config/opensearch.yml file in the “Paths” section: path.repo: ["/mnt/snapshots"]

Then restart all OpenSearch nodes.

Then run the command:

$ curl -X PUT 'http://localhost:9200/_snapshot/my-snapshots-repository' -d '{
 "type": "fs",
 "settings": {
     "location": "/mnt/snapshots",
     "compress: true
 }
}'

PUT /_snapshot/my-snapshots-repository
{
    "type": "fs",
    "settings": {
        "location": "/mnt/snapshots",
    }
}

To view the information in the created directory, run the command:

$ curl -X GET 'http://localhost:9200/_snapshot/'

GET /_snapshot

Finally, to check that the created directory is functional, run the command:

$ curl -X POST 'http://localhost:9200/_snapshot/my-snapshots-repository/_verify'

POST /_snapshot/my-snapshots-repository/_verify

This command lists all nodes where verification has been successful.

Snapshot generation

A directory can contain several snapshots for a single cluster. A snapshot is identified by a unique name within a cluster. The following command creates the my-snapshot snapshot in the directory created previously:

$ curl -X PUT 'http://localhost:9200/_snapshot/my-snapshots-repository/my-snapshot?wait_for_completion=true'

PUT /_snapshot/my-snapshots-repository/my-snapshot?wait_for_completion=true 

The wait_for_completion parameter specifies whether or not the request should return a response immediately after snapshot initialization (default) or wait for snapshot completion. During snapshot initialization, information on all previous snapshots is loaded into memory, which means that for large directories this can take several seconds (or even minutes).

By default, a snapshot of all open and started indexes is created. However, it is possible to specify only the indexes to be included in the snapshot with the following command:

$ curl -X PUT 'http://localhost:9200/_snapshot/my-snapshots-repository' -d '{
 "indices": "index_1,index_2",
 "ignore_unavailable": "true",
 "include_global_state": false
}'

PUT /_snapshot/my-snapshots-repository/my-snapshot
{
    "indices": "index_1,index_2",
    "ignore_unavailable": "true",
    "include_global_state": false
}

To view snapshot information, enter the following command:

$ curl -X GET 'http://localhost:9200/_snapshot/my-snapshots-repository/my-snapshot

GET /_snapshot/my-snapshots-repository/my-snapshot

Finally, you can delete a snapshot with the command:

$ curl -X DELETE 'http://localhost:9200/_snapshot/my-snapshots-repository/my-snapshot

DELETE /_snapshot/my-snapshots-repository/my-snapshot

Restoration

A snapshot can be restored with the following command:

$ curl -X POST 'http://localhost:9200/_snapshot/my-snapshots-repository/my-snapshot/_restore'

POST /_snapshot/my-snapshots-repository/my-snapshot/_restore

This command restores all snapshot indexes. However, it is possible to restore only certain indexes with the following command:

$ curl -X POST 'http://localhost:9200/_snapshot/my-snapshots-repository/_restore' -d '{
 "indices": "index_1,index_2",
 "ignore_unavailable": "true",
 "include_global_state": false,
 "rename_pattern": "index_(.+)",
 "rename_replacement": "restored_index_$1"
}'

POST /_snapshot/my-snapshots-repository/my-snapshot/_restore
{
    "indices": "index_1,index_2",
    "ignore_unavailable": "true",
    "include_global_state": false,
    "rename_pattern": "index_(.+)",
    "rename_replacement": "restored_index_$1"
}

An index can only be restored if it is closed. If the index does not exist in the cluster, it is created when the snapshot is restored.

Changing index parameters during restoration

During restoration, it is possible to modify certain index parameters. In the example below, the index_1 index is restored with 3 replicas and a default refresh interval of 1s:

$ curl -X POST 'http://localhost:9200/_snapshot/my-snapshots-repository/_restore' -d '{
 "indices": "index_1",
 "index_settings": {
     "index.number_of_replicas": 3
 },
 "ignore_index_settings": [
     "index.refresh_interval
 ]
}'

POST /_snapshot/my-snapshots-repository/my-snapshot/_restore
{
    "indices": "index_1",
    "index_settings": {
        "index.number_of_replicas": 3
    },
    "ignore_index_settings": [
        "index.refresh_interval
    ]
}

Restoration to a different cluster

A snapshot is not cluster-specific. A snapshot from cluster A can be restored to another cluster B.

Save the directory containing the snapshot in cluster B and start the restore process.

Snapshot management

Snapshot status

• The list of running snapshots can be viewed with the command:

$ curl -X GET 'http://localhost:9200/_snapshot/_status'

GET /_snapshot/_status

• The search can be narrowed down to a directory:

$ curl -X GET 'http://localhost:9200/_snapshot/my-snapshots-repository/_status'

GET /_snapshot/my-snapshots-repository/_status

• The status of a particular snapshot can also be viewed:

$ curl -X GET 'http://localhost:9200/_snapshot/my-snapshots-repository/my-snapshot/_status'

GET /_snapshot/my-snapshots-repository/my-snapshot/_status

Progression

The GET /_snapshot/my-snapshots-repository/my-snapshot/_status command is used to view snapshot information before the snapshot is stopped. This makes it possible to view snapshot information while the snapshot is running, unlike the GET /_snapshot/my-snapshots-repository/my-snapshot command, which waits until the snapshot has finished before displaying the information.

Stop

If a snapshot has been executed by mistake, or if the execution is abnormally long, it can be stopped using a delete snapshot operation. The operation will then stop the snapshot before deleting it.

To cancel a restore, the indices being restored must be deleted. All data from deleted indices will also be deleted from the cluster.