Backups

A backup is a snapshot: a tar archive plus a manifest in backups/<name>/. Take one from the Backups tool, from a schedule, or from the command line.

spawnloft backup survival --scope standard --label before-update --keep 10

Choose a scope

Scope	Contents
plugins	plugins/ and mods/
worlds	The active world set
config	Root configuration files (server.properties, bukkit.yml, spigot.yml, paper*.yml, permissions.yml and similar) and config/
standard (default)	Plugins, active worlds and config
full	Everything except cache/, libraries/, versions/ and logs/

Only the active world is ever included. An inactive world is not backed up until you switch to it.

Back up a running server

A running server is flushed first: save-off and save-all flush over RCON, and save-on afterwards, so the snapshot is coherent rather than a torn copy of a world mid-write. This happens for every snapshot from any path: the panel, a schedule, the command line, an AI assistant, plugin installs, and the snapshot taken before a Minecraft version upgrade. If the flush cannot be done, the snapshot is still taken and its manifest says so.

tar may report a warning for a file the running server holds locked. That is expected on hot snapshots and is not treated as failure.

Schedule backups

In Schedule, or with spawnloft task add:

spawnloft task add survival --do backup --daily 03:00
spawnloft task add survival --do verify --weekly SUN --at 04:00

Trigger	Flag
Daily	--daily 03:00
Weekly	--weekly SUN --at 03:00
Every n hours	--hourly <n>
Every n minutes	--minutes <n>
At sign-in	--on-logon

Retention (keep, 1 to 365) removes only snapshots the same schedule produced, never one taken by hand or before a reset. Tasks run while you are signed in whether or not SpawnLoft is open; on Linux, turn on lingering for them to survive logout (spawnloft task linger on). See How it works.

Keep a second copy

spawnloft config set-backup-mirror /mnt/other-drive/spawnloft-backups

Every new snapshot is copied to the mirror as it is taken, and retention deletions follow. Servers and backups on one drive fail together; a mirror on another drive is the cheapest protection against that. spawnloft config set-backup-mirror off stops mirroring.

Verify

spawnloft verify survival --all

verify reads each archive end to end, so every gzip block is decompressed and checked, and compares the entries with the manifest. A snapshot missing a world is caught the week it was taken rather than the day it is needed. It exits non-zero on any failure, and a scheduled verify action reports failures to the server's Discord webhook.

Restore

1. Stop the server. Restore is refused while it runs, because extracting over files a live server holds open corrupts a world.
2. In Backups, choose the snapshot and Restore, or run spawnloft restore survival [ref] --yes. The default reference is latest.

Restore extracts over the server in place and deletes nothing, so a file added after the snapshot was taken survives. To return to exactly what the snapshot holds, remove the members it lists first.

Databases

standard and full snapshots of a server with an attached database include a databases/ dump. Restore imports it back into the database it came from, which has to be running. MySQL databases also have their own Backups tool with downloadable SQL dumps. See Databases.

Single-file backups

When an AI assistant changes a configuration file, that file is snapshotted alone as before-edit_config_…. Restoring it puts back that one file and touches nothing else.

Where snapshots live

backups/ under the data folder. Settings can point it at another drive; that changes where new snapshots go and never moves existing ones.