The snapshot script will shutdown the node for as long as the archive and upload process takes, so use a dedicated node for creating snapshots.
The node data directory can be restored from a .tar, .tar.gz or .lz4 file stored on a public URL.
The file can be obtained from the following sources:
- Direct URL to the archive file
- Base URL file listing, where the archive matches a given pattern.
- snapshot.json generated by O-Line Snapshot backup feature.
Note that snapshots will be restored in-process, without downloading the snapshot to disk first. This saves disk space but is slower to extract, and could be made configurable in the future.
| Variable | Description | Default | Examples |
|---|---|---|---|
DOWNLOAD_SNAPSHOT |
Force bootstrapping from snapshot. If unset the node will only restore a snapshot if the data contents are missing |
1 |
|
SNAPSHOT_URL |
A URL to a .tar, .tar.gz or .lz4 file |
http://135.181.60.250/akash/akashnet-2_2021-06-16.tar |
|
SNAPSHOT_BASE_URL |
A base URL to a directory containing backup files | http://135.181.60.250/akash |
|
SNAPSHOT_JSON |
A URL to a snapshot.json as detailed in Snapshot backup |
https://cosmos-snapshots.s3.filebase.com/akash/pruned/snapshot.json |
|
SNAPSHOT_FORMAT |
The format of the snapshot file | tar.gz |
tar/tar.zst |
SNAPSHOT_PATTERN |
The pattern of the file in the SNAPSHOT_BASE_URL |
$CHAIN_ID.*$SNAPSHOT_FORMAT |
foobar.*tar.gz |
SNAPSHOT_DATA_PATH |
The path to the data directory within the archive | snapshot/data |
|
SNAPSHOT_WASM_PATH |
The path to the wasm directory within the archive, if exists outside of data | snapshot/wasm |
|
SNAPSHOT_PRUNING |
Type of snapshot to download, e.g. archive, pruned, default. |
pruned |
archive |
O-Line includes a script to automatically snapshot a node and upload the resulting archive to any S3 compatible service like Filebase.
At a specified time (or day), the script will shut down the tendermint server, create an archive of the data directory and upload it.
Snapshots older than a specified time can also be deleted. Finally a JSON metadata file is created listing the current snapshots. The server is then restarted and monitored.
See an example of a snapshot node deployment.