Backup Role

Configure Gravity’s inbuilt backup system to upload snapshots to S3 or create them locally.

  2 minute read  

Snapshots

Gravity’s inbuilt backup regularly saves etcd snapshots to a user-configured S3 endpoint and bucket. These snapshots can then be used to restore a cluster from scratch.

Starting with 0.4.5, Gravity will also keep 1 snapshot locally without any additional configuration, in the /data/backup directory. This snapshot is created/updated with the same schedule as configured in the Role configuration’s Cron Schedule This snapshot can be restored just like other snapshots.

Restore

  1. Move the snapshot file into a volume accessible by the Gravity container.

  2. In the Gravity container, run gravity cli snapshot status /path/to/snapshot to ensure the snapshot is a valid file.

  3. Run gravity cli snapshot restore /path/to/snapshot --data-dir /data/restore. This will restore the snapshot to /data/restore, but the following steps are required to make Gravity use the restored data.

  4. Stop the container.

  5. Remove the etcd folder in the data volume.

  6. Rename restore to etcd.

  7. Start the container.

  8. Run gravity cli etcdctl member list -w table in the Gravity container to list all nodes.

  9. If the value under PEER ADDRS is http://localhost:2380, run the following command:

    # Replace <ID> with the value `ID` from the command above.
gravity cli etcdctl member update <ID> --peer-urls http://$INSTANCE_IP:2380

Snapshots don’t include any cluster information, so if you were using a cluster, you’ll have to re-join nodes.

CLI

Additionally, there are CLI tools to export/import all data in a more human-readable format:

cli export

Running gravity cli export in the container will export all keys in the database into a JSON file.

Example:

gravity cli export /data/dump.json

cli import

Running gravity cli import in the container will import all keys from a JSON file into the database.

Example:

gravity cli import /data/dump.json
Role configuration