Skip to main content
Mini PC Lab logo
Mini PC LabMini PCs for Homelabs
tutorials

Home Assistant Backup and Restore: Prove Your Recovery Plan

By Max · October 13, 2026 · Updated November 18, 2026

Small storage enclosures, USB accessories and a circuit board on a desk

Illustration of hardware involved in a home automation recovery.

If you have never restored a Home Assistant backup, you do not yet know whether your recovery plan works. The file may exist, the schedule may run, and the backup page may show a green status while the encryption key, storage target, USB radio, or replacement host remains unprepared.

The reliable workflow has two parts:

  1. Create encrypted backups on a schedule and keep copies away from the Home Assistant host.
  2. Restore one to a spare mini PC or isolated virtual machine, then verify the things that make your home automation useful.

The second step is the one most guides skip. Home Assistant is not just a dashboard. It is a collection of automations, device identities, radio coordinators, apps, databases, network shares, and credentials. A restore that reaches the login screen is a good start, not the finish line.

If you are choosing the host for a new installation, our best mini PC for Home Assistant guide covers the hardware side. This article focuses on backup scope, recovery, and migration.

What a Home Assistant backup contains

Home Assistant Container backs up its settings and can include local history. Home Assistant OS also offers apps, media and shared folders in its backup selection. Check the saved selections before a move. The recommended automatic schedule leaves out some large folders. An external recorder database or network share still needs its own backup.

That distinction matters because a full backup is not the same as a backup of the entire computer. It does not turn your mini PC, virtual machine definition, Docker host, router, network share, or USB coordinator into a portable appliance.

Data or dependencyIn the Home Assistant backupSeparate recovery work
Dashboards, automations, helpers, and configurationYes, when included in the selected backup scopeVerify after restore
Built-in recorder databaseLocal history can be included. An external recorder database needs its own backupDecide whether long history belongs in daily backups
Apps and their data on Home Assistant OSCovered according to the selected backup scope and app caveatsCheck each app after restore
Media and shared filesIncluded only when selectedKeep another copy if excluded
SSL and Home Assistant configuration dataIncluded when the relevant directory is selectedRecheck certificates and remote access
Physical Zigbee, Z-Wave, or Bluetooth radioNo physical hardware is copiedReconnect or migrate the coordinator
Mini PC host, VM definition, Docker Compose, and host OSNoBack up the host layer separately
NAS, router, DNS, and external servicesNoKeep their own configuration and data backups

The official Home Assistant backup documentation recommends reducing backup size when appropriate. Media and shared folders can make a backup slower to create and restore, but excluding them is only sensible when those files have their own recovery path.

Do not decide what to exclude by looking only at the archive size. Home Assistant says the backup file size is not a reliable measure of the total installation size. On Home Assistant OS, open System Information and record Supervisor’s Disk used. Home Assistant requires a healthy target with internet access and at least 2 GB free after uploading the backup. For example, if the source reports 28 GB used, choose more than 30 GB of target capacity and allow room for future history. Home Assistant restore prerequisites.

The encryption key is part of the backup

Home Assistant backups can be encrypted by location, and Home Assistant Cloud backups are encrypted. The encryption key is not useful if it stays only inside the system that may fail.

If you gave a backup action its own password, keep that password with the recovery record for that archive. The default kit key cannot replace it. During backup setup, download the backup emergency kit and store it outside Home Assistant. It contains the information required to decrypt the matching backup. If you change the encryption key later, older backups still need the old key, so keep a clear relationship between each kit and the backups it unlocks.

The official emergency-kit documentation is blunt about the failure mode: if the key is lost and the original system is no longer available to decrypt the backup, the encrypted archive cannot be restored. That makes the emergency kit a recovery credential, not a piece of optional setup paperwork.

Store the kit in a location that does not depend on Home Assistant:

  • An encrypted password manager attachment.
  • A removable drive stored away from the mini PC.
  • A second encrypted storage location with access limited to the household administrator.
  • A printed recovery record for the location and key version, without exposing the key itself in an insecure place.

The best location depends on your threat model. The important boundary is that losing the Home Assistant disk must not also destroy the only copy of the key.

A backup plan that survives a dead mini PC

The default local /backup directory is useful for quick rollback, but it is not a disaster-recovery plan if the boot or data drive dies. Keep at least one copy on another system and ideally one off-site copy.

For a mini PC homelab, a practical layout looks like this:

CopyLocationPurpose
Working copyHome Assistant local backup locationFast rollback after a bad update
Second copyNAS or another storage systemSurvives failure of the Home Assistant disk
Off-site copyHome Assistant Cloud or another remote locationSurvives theft, fire, or a wider household outage

Home Assistant’s automatic-backup settings can schedule backups, remove older copies, create a backup before updates, select what data is included, and enable backup locations. Configure the schedule only after every destination is reachable at the selected time. A network share that is asleep or disconnected turns an otherwise good schedule into a failed copy.

Daily configuration backups are a sensible starting point for an active smart home. Retention should match how far back you may need to roll after a bad integration update. Keep more than one generation so a newly created backup cannot be the only surviving copy of a broken configuration.

The power cost calculator is useful when you add an always-on NAS, spare recovery host, or external backup drive to the setup. A backup strategy should include the equipment that keeps the copies available.

Set up automatic backups

The current Home Assistant workflow is in Settings, System, and Backups.

  1. Open the backup settings and choose the automatic-backup setup.
  2. Download the backup emergency kit before creating the first encrypted copy.
  3. Choose a daily or custom schedule that runs while all backup locations are available.
  4. Enable a pre-update backup if you want each update to create a rollback point.
  5. Choose how many generations to retain.
  6. Select the data that belongs in the backup.
  7. Enable the local, network, or cloud locations you actually maintain.
  8. Create a manual backup. Choose the encrypted download option for a recovery-key drill. The ordinary browser download can decrypt the archive on the way out. Protect any decrypted copy as you would the live configuration.
  9. Record the backup name and the emergency-kit version with your recovery notes.

If the built-in schedule does not match your needs, Home Assistant also documents the backup.create_automatic action. That lets an automation turn on a storage target, wait for it to become available, and then create the backup. Keep this automation simple. A long chain of dependencies can fail at the same time as the system you are trying to protect.

For Home Assistant OS, the command line also exposes a small recovery-oriented workflow:

ha backups list
ha backups new --name before-update
ha backups restore backup-slug

Use the UI for ordinary operation unless the web interface is unavailable. The commands are useful when you are working from a local terminal or planning a controlled recovery procedure.

The restore drill

The safest restore drill uses a spare mini PC or a disposable Home Assistant OS virtual machine. The target must have enough storage, the same broad architecture support, and a network arrangement that cannot let a restored clone issue commands to the real house.

If you use a virtual machine, the VM capacity estimator helps reserve memory and storage for the temporary target. The drill does not need to run forever. It needs to stay alive long enough to prove that the backup, key, integrations, and recovery notes are usable.

Before you start

Write down the current state while the original system is healthy:

  • Home Assistant installation type.
  • Home Assistant version and host architecture.
  • Used disk space.
  • USB radio model and current device path.
  • Network storage locations.
  • Apps or supporting containers that the household depends on.
  • Critical automations that must work after recovery.
  • The location of the backup archive and matching emergency kit.

Take a fresh full backup, download it to a separate computer, and confirm that the emergency kit belongs to that encryption key. Do not use the only copy stored on the source system for the drill.

Prepare an isolated target

Install Home Assistant OS on the spare machine or in a virtual machine. During a drill, keep the target away from the live device network when possible. A restored clone may contain automations that turn lights on, unlock doors, trigger notifications, or call external services.

Do not attach the live Zigbee or Z-Wave coordinator to the clone at the beginning. Do not give the clone the same public hostname or remote-access path. An isolated test network can use a separate radio with its own test devices. Move the household radio only during a controlled handover after stopping the old controller.

For a target VM, use an isolated network or a firewall policy that blocks device-control networks while allowing the administrator to reach the onboarding screen. For a spare mini PC, use a lab network and a temporary address rather than copying the production address.

Restore during onboarding

This is the cleanest path for moving from an old host to a replacement host.

  1. Install a supported Home Assistant target and reach the onboarding welcome screen.
  2. Choose Upload backup and select the downloaded archive from the computer you are using to open the UI.
  3. Select the full backup scope when you need the complete configuration, or choose only the parts required for a focused recovery.
  4. Enter the key from the matching emergency kit.
  5. Start the restore and wait.
  6. Do not refresh the browser while the target is being wiped and reinstalled from the backup. Home Assistant warns that a refresh can show a not-found page while the system is unavailable.
  7. Sign in with the username and password that existed when the backup was created.
  8. Wait for the system and selected apps to finish starting before judging the restore.

The official restore documentation says the onboarding path can migrate a setup to a different device type. It also says that the target needs more free space than the source installation uses. A small archive is not proof that a smaller replacement disk is sufficient.

Restore on the current host

Use an in-place restore when the host still boots but a configuration or update needs to be rolled back.

  1. Take a current backup first if the system is stable enough to do so.
  2. Open Settings, System, and Backups.
  3. Select the backup to restore.
  4. Choose the directories and apps to restore.
  5. Confirm the restore.
  6. Wait for Home Assistant to restart and reconnect.
  7. Use the post-restore checklist before declaring the rollback complete.

Restoring selected data overwrites the corresponding current data. A full restore is more destructive than a partial restore, and the full-backup action documentation warns that changes made after the backup are lost.

Verify the restored system

Begin with login and configuration on the isolated clone. Reserve live device, cloud and network-share checks for the handover after the original system is stopped.

Login and core configuration

  • Sign in with the old credentials.
  • Confirm the expected dashboards, areas, helpers, scenes, scripts, and automations exist.
  • Check Settings, System, and Repairs for new issues.
  • Read the logs for configuration errors and failed apps.
  • Confirm the time zone, units, location, and network address are correct.

Devices and integrations after the old controller is stopped

  • Confirm representative lights, switches, sensors, and climate devices are available.
  • Check MQTT or other supporting services if your installation depends on them.
  • Reconnect the Zigbee or Z-Wave coordinator only during the cutover, after the original controller is stopped.
  • Verify the coordinator path and integration settings after the radio is attached.
  • Check Bluetooth adapters and any USB serial devices.
  • Review unavailable entities instead of assuming they will heal by themselves.

Home Assistant’s ZHA documentation explains that a changed coordinator can require migration or advanced migration. The backup preserves the Home Assistant side of the setup, but it cannot carry the physical radio to the replacement host.

Automations and safety

  • Trigger a harmless test automation and confirm its trace.
  • Confirm that scheduled automations have the expected time zone.
  • Check conditions that depend on presence, sun position, or helper state.
  • Confirm notification targets without sending sensitive information to a real household channel.
  • Review door, lock, alarm, heating, and water-leak automations manually.
  • Keep the clone disconnected from real actuators until you understand every action it can perform.

Do not test a restored clone by letting it control the production house. A recovery drill should prove the configuration without creating a second automation controller.

Storage, remote access, and apps

  • Reconnect network storage and confirm the share is mounted.
  • Check that media and shared folders are present when they were included.
  • Confirm certificates and reverse-proxy settings if the clone uses them.
  • Verify that apps start and retain their configuration.
  • Check that backup locations are still available from the restored system.
  • Create a small new backup from the clone and confirm the backup process still works.

When the checklist passes, record what had to be reconnected or changed. That note is more valuable than a vague statement that the restore succeeded.

Migration paths by installation type

Home Assistant’s supported installation types do not have the same recovery boundary. Choose the path that matches how the system is actually installed.

InstallationWhat the Home Assistant backup helps restoreWhat you must preserve separately
Home Assistant OS on a mini PCConfiguration, selected apps, and supported backup dataBoot media process, physical radios, host accessories, network and router settings
Home Assistant OS in a VMHome Assistant configuration and selected appsVM definition, virtual disk, USB passthrough, hypervisor startup, host backup
Home Assistant ContainerHome Assistant configuration and supported backup dataDocker Compose, image tag, host OS, mounted paths, USB mapping, MQTT and companion containers
Existing host rollbackSelected configuration and app dataA current safety backup before restoring older data

The official installation overview recommends Home Assistant OS for most users and describes Container as a host-managed installation without apps. That difference matters during recovery. A Home Assistant backup does not recreate the Linux host, Docker network, Compose file, or supporting containers that surround a Container installation.

If you run Home Assistant in Docker, preserve the Compose file, environment settings, configuration mount, device mappings, network mode, and supporting services in a separate host backup. Our Docker homelab guide covers the wider host model. Do not assume that restoring the Home Assistant configuration also restores MQTT, Zigbee2MQTT, a database, or a reverse proxy.

If you run Home Assistant OS in Proxmox, a VM snapshot can make a fast rollback, but it should complement the application backup. The Home Assistant backup is easier to move to a new host or different architecture. The VM backup is useful when the entire guest or its disk needs to be recovered.

Rehearse the handover in two stages

First, prove what you can recover with the house disconnected. Give the spare VM a separate address on a private test network. Block its routes to the device LAN, production MQTT broker and cloud-control services, and leave USB radios detached. Cloud integrations and webhooks can reach real devices even when the clone has no route to a local light switch. Home Assistant OS checks internet connectivity before restoring. Allow the update and download endpoints it needs while keeping actuator endpoints blocked. If you cannot separate those paths, use a backup from a disposable installation for the rehearsal and restore the household archive during the controlled cutover.

Restore the selected backup. Sign in with an account that existed when you took it, open a named dashboard and inspect one automation you can recognize. A helper used only inside Home Assistant makes a good test target. An unavailable physical sensor at this point is expected. Do not attach its coordinator just to make the dashboard green.

Then choose a cutover window for the checks that need real hardware. Stop the original controller and prevent its VM or container from automatically restarting. Move one radio, confirm its stable device mapping and reconnect only the network paths the replacement needs. Check a battery sensor reporting a new reading and a harmless light responding to a local command before enabling automations that affect heating, locks or alarms.

DependencySave before the moveFirst check on the replacement
ZHA with the existing Zigbee adapterHome Assistant backup, ZHA network backup, adapter model and stable device pathThe integration sees the original adapter and receives a new report without re-pairing
ZHA with a different adapterZHA network backup and the documented migration path for the old and new radio typesFollow the radio migration procedure and verify device communication before resetting anything
Z-Wave JSDownload the controller’s NVM backup and preserve the driver configuration and network security keys securelyConfirm the expected network and one existing secured device before attempting a new inclusion
MQTT or an external recorder databaseBroker configuration, credentials, persistence and a consistent database backup from the system that owns themA fresh message or history query succeeds under the restored account

ZHA and Zigbee2MQTT store different application state. If you use Zigbee2MQTT, preserve its data directory and follow its adapter-specific recovery guidance. A Z-Wave NVM file describes the adapter’s network. Keep it alongside the driver’s recovery material. ZHA radio migration, Zigbee2MQTT migration guidance and Z-Wave network backup.

Set a rollback point before the first change. Keep the original host and its disk intact. If the replacement cannot see the original coordinator or recover an essential integration, stop the replacement, disconnect it, return the unchanged adapter to the original host and start that system alone. If you already migrated or reset a radio, check its own recovery procedure first. Rolling back the VM disk cannot undo a change written to the adapter.

For a replacement disk, start with the Home Assistant hardware requirements and OS installation guide. Keep an independent copy of the archive and a separately retrievable emergency kit as part of your household backup plan.

Before every update

An update is the most convenient time to create a recovery point because the system is still healthy and the change is known.

Before updating:

  1. Read the release notes for breaking changes.
  2. Confirm the last backup completed and exists away from the host.
  3. Confirm the emergency kit is accessible.
  4. Create a manual backup if the update affects a critical integration.
  5. Check that the target disk and host have enough free space.

After updating:

  1. Check Settings, System, and Repairs.
  2. Read the logs for warnings and failed apps.
  3. Confirm critical device states.
  4. Trigger a safe automation.
  5. Keep the pre-update backup until the new version has survived normal operation.

Home Assistant OS also documents the ha core update --backup command for creating a partial backup before a Core update. The flag is a useful guardrail, but it does not replace an off-host backup or a recovery drill.

Who should skip a full restore drill on the live network

Do not restore a clone onto the same network with the same radios, public hostname, and access path. That creates a second automation controller and can produce actions you did not intend.

If you do not have a spare mini PC, use an isolated VM if your host has enough memory and storage. If neither is available, at least download a backup, verify the emergency kit, inspect the archive, and document the onboarding steps. That is weaker evidence than a real restore, but it is better than leaving the only copy attached to a dead host.

Do not make Home Assistant Cloud, a NAS, or a backup integration your only recovery dependency if the rest of the recovery path depends on the same network, credentials, or power system. Keep the archive and key in separate places and write down how to reach them.

The practical recovery standard

We recommend this minimum standard for a Home Assistant installation on a mini PC:

  • Automatic backups run on a schedule that matches how often the home changes.
  • At least one backup copy is outside the Home Assistant host.
  • The emergency kit is stored outside the host and mapped to the backup key.
  • The host, VM, Docker, radio, router, and network-share dependencies have their own notes or backups.
  • A fresh backup can be downloaded without opening the Home Assistant system.
  • A restore drill has reached the login screen and passed a device, automation, app, storage, and backup checklist.
  • The last successful restore date and any manual reconnection steps are written down.

The goal is not to make recovery look effortless. The goal is to know which parts restore automatically and which parts require a person. That knowledge is what turns a mini PC failure from a rebuild project into a controlled migration.

Sources

Frequently Asked Questions

Where does Home Assistant store backups?

Home Assistant stores backups locally in the backup directory by default. You can configure additional locations such as network storage or Home Assistant Cloud, but a backup that remains only on the Home Assistant host will not help much after that host loses its storage.

What is the Home Assistant backup emergency kit?

The backup emergency kit contains the encryption key and metadata needed to restore encrypted backups. Store it outside the Home Assistant system and keep the kit that matches each encryption key if you rotate keys later.

Can I restore Home Assistant to a mini PC?

Yes. Install a supported Home Assistant target, then restore the backup during onboarding or from the backup page. The target needs more free storage than the source installation uses, and USB radios, network storage, and host-level settings may need separate attention.

Does a Home Assistant backup include Zigbee devices?

The backup includes Home Assistant configuration and integration data, but it does not replace the physical coordinator. Preserve the radio, connect it to the replacement host, and follow the integration migration steps if the coordinator changes.

How often should I test a Home Assistant restore?

Run a restore drill after creating the first backup, then repeat it after major hardware, installation, storage, or encryption-key changes. A scheduled backup is only an assumption until a spare host or isolated virtual machine can restore it and pass a device and automation checklist.