Match this symptom
A crash, update or configuration change damaged progress and the server must return to a known-good point.
The answer first
Stop the process, preserve the current ShooterGame/Saved directory, and restore a complete timestamp-matched backup rather than replacing only the map file. Verify tribes, characters and map state privately before reopening.
What belongs in the backup
- The complete ShooterGame/Saved directory, including map, tribe, player and configuration state needed by the deployment.
- The launch arguments, map name, mod list and server version used by that save.
- A checksum or size record so an unexpectedly empty backup is detected before an emergency.
- At least one copy outside the game installation and preferably on another storage device.
Use ARK's save interval carefully
The server setup reference documents AutoSavePeriodMinutes and the SavedArks backup controls. More frequent saves reduce ordinary progress loss, but they do not replace independent snapshots: corruption and unwanted changes can be preserved just as efficiently as good state.
Rollback procedure
- 1Announce maintenance, stop the server cleanly and confirm the process has exited.
- 2Copy the current ShooterGame/Saved directory to a quarantine location. Never delete it before recovery is accepted.
- 3Choose a complete backup from before the incident and restore it as one set. Keep map, tribe and player files from the same recovery point.
- 4Start privately. Verify the map, a representative base, tribe ownership and at least one returning character.
- 5If the result is wrong, stop and select another known-good point. Keep a record of every attempted timestamp.
After a successful rollback
- Create a fresh backup of the verified recovered state.
- Identify the original cause before re-enabling automatic updates or the last configuration change.
- Tell players exactly how much time was rolled back.
- Retain the pre-rollback quarantine copy until the incident is closed.
Prove that map, tribe, and survivor files belong together
Inventory the active map name and the map-specific save files before choosing a point. ARK progress is distributed across the map state and related tribe/player profiles; restoring only the largest map file can leave characters detached from tribes or ownership. Use timestamps, backup naming, checksums, and the server log to select a set captured under the same build, map, and mod order.
During private verification, test a returning survivor rather than only a new character. Inspect tribe membership and ownership at a known base, a tame, inventory/engrams, and any cluster upload/download state the deployment relies on. Save and shut down cleanly, then repeat after a second start. If any relationship is missing, stop and test another complete recovery point rather than combining individual profile files from different timestamps.
For clustered servers, keep the cluster directory and every participating map's maintenance timestamp in the incident record. A healthy restored map can still reference transfers produced by a later or incompatible cluster state, so test upload/download only on a quarantined copy before restoring inter-map travel.