Preserve the installed identity
Record the exact release, kit checksums, installation journal, hostname intent and private configuration before changing software. Keep backups and recovery keys separate from the server.
An interrupted installation uses its ownership journal. With the same verified bundle and hostname intent, inspect the new plan and follow the documented rerun/repair procedure. Changed or unknown files can block repair; preserve them for investigation instead of deleting the journal or resetting panel metadata.
Understand update availability
The repository includes signed-update verification and isolated process-update tests. Installed production upgrades remain blocked because the required full-panel reconstruction proof and installed update-coordinator acceptance are incomplete. The presence of an update interface or valid-looking archive does not remove those prerequisites.
Use only the release-specific approved procedure. Do not copy new binaries over active services, invent an unattended update command or treat a code rollback as restoration of database/host effects.
Separate repair and recovery
A repair attempts to reconcile files/services owned by the same reviewed installation; it is not a migration, full-server restore or safe takeover of another panel. Retain the original state and inspect actual services afterward. Keep the isolated VM and provider snapshot available so an unsuccessful evaluation does not become an experiment on customer workloads.