Tools / Conan Exiles
Conan Exiles Save Recovery & Transfer
World won't load, or moving a single-player world to a dedicated server? A Conan world is one
SQLite database, game_0.db, and the server keeps rolling backups
game_0_backup_1.db, game_0_backup_2.db… Point this page at your
Saved folder: it runs a structural SQLite check on the live
world and every backup - not just a size guess - shows which pass and when each was
saved, and rebuilds a clean game_0.db you can drop back in.
Your world never leaves your browser. It reads only file names, sizes,
modified-times and each database's 100-byte SQLite header, copies the chosen database into a zip
locally, and never modifies your originals - the zip keeps the file it would replace as a backup,
and deliberately drops the stale -wal/-shm sidecars.
Pick your Saved folder
On a dedicated server the world lives in ConanSandbox/Saved/ (on a Supercraft server,
con/data/Saved/); single-player keeps game.db in the same
Saved tree. Pick that folder. Stop the server / close the game first so nothing is
mid-write.
Reads names, sizes, times & each .db header locally. Nothing is uploaded.
Want this on a real server? Supercraft hosts Conan Exiles →
How Conan Exiles saves - and why restores fail
Conan Exiles (the Enhanced / UE5 build) stores the whole world - buildings, thralls, characters -
in a single SQLite database, game_0.db, inside ConanSandbox/Saved/. The
engine rotates numbered backups game_0_backup_1.db, game_0_backup_2.db…
alongside it. When the live world breaks - a crash during a save, an interrupted upload, a bad
copy - the last intact copy is usually one of those backups.
The trap is the two sidecar files SQLite writes while the server runs:
game_0.db-wal (the write-ahead log) and game_0.db-shm. They belong to
the current database. If you copy a different game_0.db into place and leave
the old -wal/-shm behind, SQLite replays them onto the new file and
corrupts it - the most common reason a "restored" world still won't load. This tool never carries
those sidecars into the zip and tells you to delete them first.
Because the world is SQLite, its structure is checkable: the tool verifies the
SQLite format 3 header magic, that the file is a whole number of pages, and that it
is not shorter than the size the database's own header declares - so a truncated or
header-damaged game_0.db is flagged instead of handed back to you. A structural
check like this catches the common corruption, but it can't detect every kind of internal
damage - if a world that passes still won't load, restore a backup.
Manual recovery (no JavaScript needed)
- Stop the Conan Exiles server (or close single-player) so nothing writes to the database.
- Open
ConanSandbox/Saved/and copygame_0.dband the newestgame_0_backup_*.dbsomewhere safe. - Delete
game_0.db-wal,game_0.db-shmandgame_0.db-journalif present - they describe the old database. - Rename the current
game_0.dbtogame_0.db.old, then copy your chosen backup in and rename it togame_0.db. - Start the server and confirm the world loads before deleting the
.oldfile.
Moving a world instead of fixing one? The same steps package a transfer - rename a
single-player game.db to game_0.db for the server. Prefer it hosted with
daily automated backups? Run a Conan Exiles server with Supercraft.
FAQ
Is my save really not uploaded?
Really - there's no upload endpoint on this page. Reading the folder, checking each database's header, and building the recovery zip all happen in your browser; the network tab shows zero requests carrying your world.
Can it recover a world with no intact copy?
No tool can rebuild data that isn't on disk. If the live game_0.db and every backup
fail the SQLite structural check, it says so honestly instead of handing you a broken database.
Does it work for Isle of Siptah and modded servers?
Yes - the map and mods are chosen at launch, not by the database name, so every Exiled Lands,
Siptah, or modded world is the same game_0.db this recovers.