Short answer: an unavailable device usually indicates a failure in one of six layers: power, the device itself, its radio or IP network, the controlling integration, the entity record, or the automation using it. Follow the branches below in order and stop when the evidence identifies the failing layer.
Before starting: choose one affected device. Record its model, connection type, exact message, when it last worked, recent changes, relevant software versions, and what still works.
1. Is there a safety risk?
Yes—heat, smoke, arcing, swelling battery, water, gas, insecure lock, or failed alarm
Stop troubleshooting. Disconnect power only when safe and follow the manufacturer’s emergency instructions or contact an appropriate qualified professional or emergency service.
No obvious safety risk
Continue to scope the failure.
2. How much is affected?
Only one device
Check power or battery, then test the device in its native app or local interface. If native control fails, investigate the device and its connection before Home Assistant.
Every device using one protocol
Check the Zigbee/Z-Wave coordinator, Matter controller or Thread border router, MQTT broker, Bluetooth proxy, bridge, or Wi-Fi access point shared by those devices.
One integration
Open Settings → Devices & services. Look for reauthentication, setup failures, repairs, rate limits, service outages, or a changed device address. Save diagnostics before reloading once.
The entire Home Assistant system
Check host power, storage, network, DNS, time synchronization, Core, Supervisor, and router status. Do not reset endpoint devices.
3. Does native control work?
No
The likely fault is below Home Assistant: power, battery, Wi-Fi/radio membership, cloud service, account, firmware, or the device itself. Restore native operation first.
Yes
The likely fault is discovery, reachability from the Home Assistant network, authentication, integration configuration, entity mapping, or an automation.
4. Identify the connection path
Wi-Fi or Ethernet
Confirm the current IP by MAC address, DHCP reservation, VLAN/firewall route, DNS or mDNS, and the actual port/protocol Home Assistant uses. Router presence alone does not prove reachability.
Zigbee
Check coordinator status, channel interference, mesh routers, signal quality, battery wake behavior, and whether the device remains joined. Avoid re-pairing until coordinator and mesh health are known.
Z-Wave
Check controller connection, node statistics, power, routes, sleeping-node behavior, USB interference, and failed-node status before exclusion or factory reset.
Matter over Thread
Check fabric membership, preferred Thread network, credentials on the commissioning phone, border router, IPv6, multicast, and network reachability separately.
MQTT
Check broker reachability, authentication, subscriptions, discovery replay, birth messages, retained availability, and topic spelling before recreating entities.
Bluetooth
Check adapter or proxy health, signal coverage, passive versus active connections, encryption requirements, and whether another app holds the connection.
5. Integration, entity, or automation?
- If the integration is unhealthy, save its diagnostics and relevant redacted logs. Correct authentication or reachability, then reload it once.
- If the integration is healthy, check whether only one entity is disabled, stale, renamed, duplicated, or governed by an availability timeout.
- If the entity updates correctly, inspect the automation trace: trigger, conditions, selected branch, target, action response, timing, and concurrency mode.
Evidence that the fix worked
- The device remains available longer than its previous failure interval.
- Entity timestamps and values continue updating.
- No recurring authentication, coordinator, or transport error appears.
- A safe manual command succeeds and the expected state is reported back.
- Related automations complete with a correct trace.
When to escalate
Provide the model, connection type, installation type, relevant versions, exact redacted error, first failure time, what still works, changes immediately before failure, and results of the checks above. Never share credentials or private diagnostic archives publicly.
Related: complete smart-home diagnostic decision tree.
Method status: this tool provides a diagnostic sequence; it does not claim hands-on testing of every device. Last reviewed: September 9, 2026.