Skip to main content

Upgrade, backup and recovery

Change procedure​

  1. Announce a maintenance window and stop new Agent/Work activity.
  2. Record version, image/package identity and deployment configuration.
  3. Create and verify a complete backup. Back up linked folders separately.
  4. Download and verify the candidate before stopping the current deployment.
  5. Upgrade through the supported platform manager.
  6. Verify sign-in, data, models, files, Agent runs, Scheduler, Office and Browser runtimes.
  7. Retain the previous package and matching backup until acceptance is complete.

Linux Docker​

The host manager performs the upgrade. For the default or only instance, run:

sudo clawuno upgrade --confirm-stop

With more than one instance, select the intended instance explicitly:

sudo clawuno upgrade --instance NAME --confirm-stop

To install an exact signed release rather than the latest available release, add --version TARGET_VERSION. Supplying an image digest directly is an advanced deployment path: independently verify that the digest belongs to the intended Clawuno release before using it. The manager downloads the candidate before interruption, stops writers, snapshots persistent state, and recreates the instance using an immutable image reference.

If the new version starts and then fails, the deployment remains stopped with recovery information. Do not start the old image against data that may have been migrated. Restore the matching old image and snapshot together.

Create and verify a host backup​

sudo clawuno backup --confirm-stop
sudo clawuno start
sudo clawuno status

For a named instance, add --instance NAME to each command. The backup command stops the selected instance and deliberately leaves it stopped. start is a required separate step. The Web backup list does not represent these host snapshots.

Recover a failed upgrade​

sudo clawuno recover --confirm-restore
sudo clawuno status

For a named instance, add --instance NAME to both commands. Use the recovery record from that same instance; do not copy another instance's snapshot or run an old image against migrated state.

macOS and Windows application backups​

For macOS, create a manual archive in System Settings → Health → Backup & Storage or with ~/clawuno/bin/backup.sh --type manual. Quit the App before restoring with ~/clawuno/bin/restore.sh BACKUP_FILE, then reopen it and verify the data.

For Windows Desktop, create a manual archive in System Settings → Health → Backup & Storage. To restore, quit Clawuno completely, set $engine = "$env:LOCALAPPDATA\ClawunoDesktop\current\engine" and $env:CLAWUNO_DIR = "$env:USERPROFILE\clawuno" in PowerShell, then run & "$engine\bin\restore.ps1" BACKUP_FILE. Remove the temporary environment variable, reopen Clawuno, verify the data, and retain the safety snapshot created before replacement.

Backup responsibilities​

Docker backups are host-managed and can briefly stop the selected instance. They do not include linked external folders or undo actions already sent to model providers, SaaS systems or collaboration platforms.

Test restoration on an isolated host. A backup that has never been restored is not sufficient evidence of recovery readiness.

0.8.1 platform boundaries​

  • macOS can update through Clawuno → Check for Updates or by installing the signed 0.8.1 DMG over the app.
  • Linux supports the public 0.8.0 to 0.8.1 Docker upgrade through the host manager.
  • Windows 0.8.1 supports a clean Desktop installation only. Do not treat preserved 0.8.0 Engine-only or internal-test data as a validated migration path.