Choose the right workflow before touching files
A planned update and a suspected data-loss incident need different first moves. The steps below are a maintenance rehearsal. If a character or base is already missing, do not treat saving again as a recovery method. Keep evidence of the state you found and work only on copies.
| Your situation | First action | Do not do this yet |
|---|---|---|
| The world is healthy; you want to update | Arrange a maintenance window and record an identifiable checkpoint before stopping. | Update the only installation before preserving its data and build details. |
| A base, character or inventory is unexpectedly missing | Restrict access, stop further writes through your host's controls and preserve existing data plus available backups. | Force a new save, create a replacement character or overwrite files while diagnosing. |
| A rented host hides the save folders | Ask how to export a complete dated copy and test it without replacing the live slot. | Assume a backup button means you can download or restore every part of the world. |
| You want to move a co-op host or change player identities | Find documentation for that exact migration and keep an untouched source copy. | Use this same-server backup drill as an identity-conversion procedure. |
Enable local backup history, but keep a separate copy
The 1.0.4 configuration reference exposes bIsUseBackupSaveData. Enabling it creates a backup directory within the save-data location and adds disk load. On Windows, edit the active Pal/Saved/Config/WindowsServer/PalWorldSettings.ini, not DefaultPalWorldSettings.ini. Preserve the existing file and modify the parameter inside its existing settings structure; the fragment below is not a complete configuration file.
The reference lists five 30-second snapshots, six 10-minute snapshots, twelve hourly snapshots and seven daily snapshots. These are retention tiers, not proof that a particular point is available: inspect the files and timestamps your server actually produced. A setting enabled in an editor is not evidence of a completed backup.
Our recommendation is to keep a dated copy on another device or protected storage as well. History on the same disk cannot protect against losing that disk. Never delete older known-good copies merely because a fresh archive exists. Backups can contain passwords, player identifiers and private server details; restrict access and do not post them in public support threads.
bIsUseBackupSaveData=TrueReferences: Pocketpair: configuration parameters and automatic backup history
Capture a healthy maintenance checkpoint
For this planned-maintenance path, first confirm that the running world is the one you intend to preserve. Note a base landmark, a recognizable Pal and an inventory detail. Keep private build and mod records separately from anything you share publicly.
Pocketpair documents in-game administration through an AdminPassword configured beforehand. Authenticate privately using /AdminPassword followed by your own password. The commands below are entered in the game's chat as an administrator, not in Windows PowerShell. /Info reports server information, /Save writes world data and /Shutdown schedules a stop. The reference distinguishes /DoExit as a forced stop; it is not the routine path here.
- Tell the group when maintenance starts. Record /Info output, client version, operating system, enabled mods and the active installation directory. Ask players to finish their session.
- Only while the world is healthy, run /Save. Then issue /Shutdown 60 Maintenance and allow the countdown to complete. Confirm the process has stopped; prevent a service manager or hosting panel from immediately restarting it during the copy.
- Locate the actual installation's Pal/Saved directory and confirm it contains the world data you were just using. If multiple installations or worlds exist and you cannot identify the active one, stop here; do not guess from a familiar folder name.
- Copy the entire confirmed Saved directory into a new, uniquely dated folder outside the server installation, preserving its structure. Do not copy over yesterday's backup. Separately retain the mod list, relevant custom configuration outside Saved and the build record. A Saved copy alone does not archive server executables or every mod.
- Check for copy errors and unexpected empty files or folders. If you create an archive, extract a copy into an empty temporary location and check that the expected folders are present. This checks packaging, not whether the game can load it.
- Copy the dated package to your separate storage, verify that copy can be read, and keep the original backup untouched during the restore drill.
/Info
/Save
/Shutdown 60 MaintenanceReferences: Pocketpair: in-game administrator commands · Pocketpair: dedicated-server deployment
The restore drill: prove the copy away from the live world
Use a separate test installation or a host-provided isolated test slot. Keep it off public listings and unavailable to ordinary players. It must not point at the live data directory. Prefer the same known server build and mod set first; a newer executable accepting an older save is a separate compatibility question, not something this guide guarantees.
If you cannot obtain a compatible test environment or isolate its data, retain the backup but label it untested. Do not call a failed cross-version experiment proof that the backup itself is bad. The official deployment page explains installation; the checks here are GamerWin's proposed validation procedure, not an official recovery guarantee.
- Initialize the test installation if needed, then stop it completely before copying any data. Confirm both its installation path and its intended test address; never restore into the original production path.
- Move the test installation's disposable Saved folder aside, keeping it recoverable. Place a COPY of the captured Saved folder in the corresponding test location without adding an extra nested Saved directory. Use the host's documented restore mechanism if it manages the paths.
- Keep the captured source untouched. Review the test copy's settings for public visibility, address conflicts and access restrictions before starting it. Do not expose administration interfaces to make the test easier.
- Join the test address with the same intended account and platform. Inspect the earlier world checkpoint and character state. A server-started message is insufficient. If you are asked to make a new character, stop and investigate the account, world selection and copied data instead of saving a replacement.
- On the disposable test world only, make a harmless identifiable change, save, stop cleanly, then restart and reconnect. Record whether that change persists. Keep the modified test files separate from the untouched backup.
- Record the actual outcome and stop the test instance. If a real production rollback later becomes necessary, preserve the then-current production state first and agree the lost-progress window with the group; do not automatically promote this rehearsal's modified files.
References: Pocketpair: dedicated-server deployment · Pocketpair: in-game administrator commands
Use evidence-based pass and stop conditions
These are acceptance criteria to apply to your own drill, not measured results. Record failures as well as successes. They help distinguish an unreadable archive from a wrong-world selection or a save that loads but cannot retain progress.
| Check | Evidence to record | If it fails |
|---|---|---|
| Package is readable | The separate copy opens and contains the expected data/configuration tree. | Investigate copy errors, space and access before updating the live installation. |
| Correct world loads | The recorded base landmark and recognizable Pal are present on the test instance. | Stop; confirm the active installation and selected world. Do not overwrite the live world. |
| Correct character loads | The intended account/platform has its expected character and recorded inventory detail. | Do not create a replacement. Check identity and world selection; seek version-specific help. |
| Progress survives restart | A harmless change made only in the test copy remains after a clean restart. | Check write permissions, active data location and startup configuration. |
| Restore can be repeated | The untouched backup, build/mod record and test notes remain available. | Keep the backup labelled partially checked instead of claiming recovery is proven. |
Worked planning example: choose the rollback point, not just the newest file
Illustrative scenario, not an observed server incident: your last verified copy is stamped 18:00. Players build until 19:20, and a later update leaves the world unusable. Restoring that 18:00 copy would discard up to 80 minutes of subsequent progress. A newer automatic snapshot is only a candidate until you check that it predates the problem and restores correctly.
For a planned update, take a fresh healthy checkpoint immediately before maintenance to reduce that window. For a discovered loss, preserve all available candidates first. Do not choose solely by file size or overwrite every older copy with the newest one.
Private restore record — fill in your own observations
Capture time and time zone:
Server/client build; platform; mod set:
Confirmed source installation and world:
Separate backup location:
Isolated test address and data directory:
Base / Pal / character checks:
Restart-persistence result:
Failure details or remaining uncertainty:
Keep passwords out of this record.Before you play
- Healthy maintenance and suspected-loss workflows kept separate.
- Confirmed active data copied while the server was stopped.
- Untouched copy retained outside the live installation and on separate storage.
- Restore checked on isolated data, with character and restart tests recorded.
- Build/mod limits and any untested conditions written down.
Frequently asked questions
Does enabling automatic backups replace an external copy?
No. The documented backup directory is within the save-data location. Keep a separate copy if you need protection against loss of the host or disk.
Can I restore only one world file and keep newer player data?
This guide does not establish that mixed snapshots are consistent. Preserve complete dated copies and seek instructions for the exact version and recovery case before attempting selective repair.
Does a successful test guarantee the next update will load this save?
No. It establishes only the outcome you observed with that build, mod set and test conditions. Keep the untouched backup and repeat relevant checks when those conditions change.
Sources & revision notes
Editorial revision: September 16, 2026. Source notes give the scope of each reference; the revision date is not a gameplay test date. Official documentation describes the publisher's system; community sources can contain errors. Use your installed game's own descriptions when they disagree.
- Pocketpair: configuration parameters and automatic backup history (Official)
Read September 16, 2026; page labelled 1.0.4. Supports active INI location, bIsUseBackupSaveData, disk-load warning and retention tiers; does not certify a restore.
- Pocketpair: in-game administrator commands (Official)
Read September 16, 2026; page labelled 1.0.4. Supports authentication, Info, Save, timed Shutdown and the distinction from force-exit. Commands were not executed against a game server by GamerWin.
- Pocketpair: dedicated-server deployment (Official)
Read September 16, 2026; page labelled 1.0.4. Installation context only. Isolation, copy retention and pass/fail criteria are our operational recommendations; no historical-build download or cross-version compatibility is promised.
Published a backup-focused maintenance and isolated-restore plan with official command/configuration references, explicit data-loss stop conditions, proposed acceptance checks and a labelled 80-minute rollback example. No restoration was performed by GamerWin.
How we research and correct guides