Bare-metal restore of a physical server
Physical server hardware-failed — restore the OS to dissimilar hardware. Process is recoverable but easy to get wrong on first attempt.
Indicators
- Physical server unrecoverable — disk, mainboard or PSU failure beyond field repair
- Replacement hardware available but different chassis / chipset
- Bare-metal recovery image last verified historically
- Time pressure — business waiting on restore completion
Likely causes
- Not applicable — this is a recovery procedure entry.
Diagnostic steps
-
Verify the backup is a bare-metal-capable image (not just files) — Windows Server Backup, Veeam Agent, Datto, etc.
-
Confirm replacement hardware HCL — RAID controller, NIC, storage all have drivers in the recovery environment
-
Boot recovery media; perform restore to the new hardware
-
Inject drivers post-restore if dissimilar hardware (Windows Server: dism /image:C: /add-driver / inf-only)
-
First boot: address activation (KMS / VLSC), AD secure channel reset (Test-ComputerSecureChannel -Repair if domain-joined), service account state
-
Validate app integrity: SQL DB consistency check, IIS bindings, file shares, third-party services
Resolution path
- Boot from recovery media on replacement hardware
- Restore image, inject drivers as needed
- Address activation, secure channel, service accounts
- Validate apps and integration with rest of estate
Prevention
- Annual bare-metal restore test on lab hardware
- Spare matched-hardware kept on retainer for tier-1 servers (or replacement contract)
- Driver pack maintained alongside backup
- Documented post-restore checklist per server
Tools
- Veeam Agent recovery media
- Windows Server Backup recovery media
- Hardware vendor driver pack ISO
- Sysinternals to validate post-restore service state
- PowerShell to script post-restore checks
References
- Microsoft Learn — Bare-metal recovery for Windows Server
- Veeam — Veeam Agent for Microsoft Windows recovery