Skip to content

Full-server recovery

When: the server is lost (hardware, compromise, provider failure) and a new one takes its place. Proven end to end in the 6C T5 drill (docs/superpowers/phase-6-exit-result.md, drill 5).

You need: a fresh AlmaLinux 9 server (KVM, not a container); the recovery sheet of the backup destination (printed with flypanel backup recovery-sheet NAME on the old server and kept offline) — it holds the repository key, so it never goes through chat or email.

  1. Point the new server’s hostname at its IP in DNS first (mail and every zone need it), or the recovery refuses to start.
  2. Install FlyPanel (Install and upgrade) with --hostname set to that name; check flypanel doctor is clean. Do not create any account.
  3. Copy the sheet from your own computer straight to the server, never through chat: cat sheet.txt | ssh root@NEW 'umask 077; cat > /root/recovery.sheet'
  4. flypanel dr restore --from /root/recovery.sheet — restores the newest server snapshot (or --snapshot ID), the PHP versions, mail, then every account with its original UID, resellers first. It keeps the new server’s hostname and IP (--keep-identity takes the old ones).
  5. It can be run again: it resumes from where it stopped. One account failed: fix the cause, then flypanel dr restore --from /root/recovery.sheet --account NAME.
  6. When every account is restored: shred -u /root/recovery.sheet; flypanel doctor; move DNS to the new IP.

It worked when: the report says every account ok, doctor shows 0 errors, sites answer and old passwords work. Escalate when: the version check refuses (“the snapshot was taken by a newer FlyPanel”): install that version or newer first.