EndKeep takes scheduled snapshots of your Bedrock world while the server is online.
It stores snapshots in the directory configured by storage.path (default: backups/).
Follow the installation instructions: Windows uses
command-line installation; Linux also supports placing the wheel in plugins/.
Check /backup status, take a snapshot with /backup create, and run
/backup maintenance to process it. View recovery points with /backup list.
Commands are available to the console and operators by default (endkeep.admin).
| Command | Description |
|---|---|
/backup status |
Show backup status and current work |
/backup create |
Take a snapshot now |
/backup list |
List available recovery points |
/backup maintenance |
Process snapshots waiting for maintenance |
/backup maintenance full |
Also apply retention and run configured verification |
/backup verify |
Check repository structure |
/backup verify deep |
Check stored data and restore history more thoroughly |
/backup export <id> |
Export a recovery point to a separate world directory |
/backup delete <id> |
Preview deleting a recovery point |
/backup rollover <id> |
Preview discarding recovery points older than the selected one |
/backup confirm |
Confirm the most recent delete or rollover preview |
/backup cancel |
Request cancellation of supported background work |
/backup reload |
Reload EndKeep configuration |
Deleting or rolling over a snapshot is a two-step operation: run the command
to view what will change, then run /backup confirm. Preview information is
cleared on reload or restart. Delete and rollover jobs can be cancelled before the atomic commit; exports
can also be cancelled. Make an independent copy before changing recovery history.
Default backup times are 12:00, 16:30, 20:30 and 23:45 in the server's local timezone. Maintenance runs at 06:00 and 18:30. The first maintenance run also handles retention and verification.
Snapshots are kept for 7 days or at least the latest 28 recovery points, whichever retains more. Missed backup times are not replayed.
Edit plugins/endkeep/config.toml and run /backup reload to change these
settings. See Configuration.
EndKeep normally reserves at least 5 GiB of free disk space, plus enough
room for a new snapshot. On smaller disks, adjust storage.min_free_space_gib
carefully. If storage is full or backup processing fails, new captures may be
paused to preserve earlier recovery points. Check /backup status and the
server log before retrying.
If /backup status shows health=FAILED, maintenance and backup deletion are
blocked until a successful /backup verify deep. Preserve the repository and
investigate the error rather than deleting the health marker manually.
Keep a separate off-server copy of your backups. Backups stored on the same disk cannot protect against disk or host failure.
Use /backup export <id> to create a separate restored world while Endstone
is running. To recover when the server cannot start, use the standalone restore
tool. Neither method overwrites your active world.
Follow the Recovery guide before replacing a live world.
On server shutdown, the worker cancels background work at a safe checkpoint and exits. An atomic commit already in progress finishes first; uncommitted RAW snapshots remain available for the next startup. This also applies after an unexpected server process exit. Unacknowledged results are recovered on the next startup.
Plugin disable allows 15 seconds for a newly enabled instance to reconnect. Running repository jobs continue through the handoff. If the plugin stays unloaded, the worker safely cancels its job and exits.
Startup briefly retries an unavailable worker. If it remains alive but cannot
respond, EndKeep refuses a second writer. Check backups/worker.log and keep
the backup directory intact.