# Move one Rust world while its neighbour stays open

[Rust server guides](https://rustsc.com/learn/index.html.md) / [Backups and recovery](https://rustsc.com/learn/recovery/index.html.md)

A move is easier to verify when the two worlds have independent addresses and staff can name which one is closed. Rehearse the move with a spare copy first. On the real day, only one version of the moved world should accept players.

## Record both paths before the handoff

In this example, Willow Weekly moves to a new host while Willow Monthly stays on the old one. Write the old and new Weekly addresses, the unchanged Monthly address, the two save identities and which work belongs to each host. Keep Weekly's scheduling and announcement changes separate from any jobs still needed for Monthly.

Before the scheduled move, prove a supported test restore of Weekly on the destination. Keep that rehearsal off live links. Check the expected game, framework, map and a known world marker. Tell staff what passing looks like and who has authority to reopen.

## Give every step a check

Keep the old Weekly instance and its final save available until the destination passes.

| When | Weekly | Monthly |
| --- | --- | --- |
| Close the moving world | Send the notice, stop accepting joins, save and stop Weekly cleanly, then take its final checkpoint. | A helper verifies a player can still join Monthly. |
| Bring up the destination privately | Use the proven recovery steps and check the selected checkpoint, current version, world and permissions before public joins. | Keep the usual address and verify no new Weekly job targets Monthly. |
| Open Weekly on the new host | Have an outside player verify the new address or DNS record and the expected world. Publish the updated Weekly link and leave the old Weekly closed. | Repeat the Monthly join check and leave its links and jobs active. |

## Set a point to fall back

If Weekly fails before players are admitted on the destination, leave the destination closed and reopen the old Weekly using the known final state after checking which jobs and addresses would have to return. If players have already made progress on the destination, close public joins there and decide how to handle that progress before moving back.

In a disposable Linux rebuild, a prepared replacement started from the selected world backup after the required framework had been installed. That run exercised one game at a time and did not test public DNS or a neighbouring live world. Verify those two join paths yourself before announcing the move is complete.

## Sources

- [Facepunch server identity and settings](https://wiki.facepunch.com/rust/Creating-a-server)
- [Facepunch Rust DNS records](https://wiki.facepunch.com/rust/dns-records)
- [Google SRE on assigning incident work](https://sre.google/sre-book/managing-incidents/)

## Related resources

- [How to move a Rust server hostname to a new host](https://rustsc.com/learn/hosting/move-your-rust-server-hostname/index.html.md)
- [Prove a Rust backup belongs to the right world](https://rustsc.com/learn/recovery/prove-a-backup-belongs-to-the-right-rust-world/index.html.md)
- [Move a server to a new host](https://rustsc.com/support/guides/move-to-new-host/index.html.md)

---

HTML version: https://rustsc.com/learn/recovery/move-one-rust-world-while-the-other-stays-open/
