fix: register ikea_bulb_device_set with registry_entry and handle connection errors with ConfigEntryNotReady

[pablotoledo]

Summary

Two related bugs that together caused the Dirigera integration to require multiple manual reloads before successfully loading after a Home Assistant restart. Both issues were reproduced and verified on a production HA 2025.10.3 instance with the integration running inside a Docker bridge network.

---

Bug 1: ikea_bulb_device_set crashes the hub event listener immediately after setup

Root cause

In light.py, ikea_bulb_device_set registers itself directly into the hub_event_listener device registry:

# Before (line 397)
hub_event_listener.register(self.unique_id, self)

All other entities (via base_classes.py) register a registry_entry wrapper:

hub_event_listener.register(id, registry_entry(self))

When sync_all_device_areas() iterates the device registry and calls registry_entry.entity, it works for all normal entities (because their registry value is a registry_entry object with an .entity property). But for ikea_bulb_device_set, the registry value is the entity itself — a LightEntity subclass with no .entity attribute — causing an immediate AttributeError.

Observed log errors

ERROR [custom_components.dirigera_platform.hub_event_listener] Failed to sync area for device 992c7af5-...: 'ikea_bulb_device_set' object has no attribute 'entity'
WARNING [custom_components.dirigera_platform.hub_event_listener] Failed to create listener or listener exited, will sleep 10 seconds before retrying

This crash happens on every startup that has bulbs grouped in a device set. Once the listener dies, the integration enters a broken state and all subsequent device polling fails.

Fix

# After (light.py line 397)
hub_event_listener.register(self.unique_id, registry_entry(self))

registry_entry is already imported in light.py from .hub_event_listener, so no import changes are needed.

---

Bug 2: async_setup_entry does not handle connection failures gracefully

Root cause

async_setup_entry calls make_devices() with no exception handling. If the hub is temporarily unreachable during HA startup (e.g., brief network unavailability in Docker bridge environments, hub booting up, any transient OS-level network error), the ConnectionError or OSError propagates uncaught.

Home Assistant interprets this as a permanent failure and marks the entry as setup_error — which has no automatic retry. The correct HA pattern for transient connection failures is to raise ConfigEntryNotReady, which triggers HA's built-in exponential backoff retry mechanism.

Observed behaviour

• Entry state: setup_error (permanent, no retry)
• Error in logs: [Errno 101] Network unreachable during get_scenes / get_blinds / get_lights calls within make_devices
• Manual reload required every time

Fix

from homeassistant.exceptions import ConfigEntryNotReady

# In async_setup_entry:
try:
    await platform.make_devices(hass, hass_data[CONF_IP_ADDRESS], hass_data[CONF_TOKEN])
except (ConnectionError, OSError) as err:
    raise ConfigEntryNotReady(
        f"Cannot connect to IKEA Dirigera hub at {hass_data[CONF_IP_ADDRESS]}: {err}"
    ) from err

---

Results after both fixes

Tested on HA 2025.10.3, dirigera_platform installed via HACS, Dirigera hub at 192.168.0.80, HA running in Docker bridge network:

Scenario	Before	After
Integration state after reload	setup_error (no retry)	loaded in ~10s
ikea_bulb_device_set AttributeError	Present on every startup	Gone
Failed to create listener	Triggered every startup	Gone
Manual reloads needed	~20	0

---

Files changed

• custom_components/dirigera_platform/light.py — 1 line changed
• custom_components/dirigera_platform/__init__.py — 6 lines added

This fix was previously submitted and accepted in the nrbrt fork: nrbrt#13