Growatt integration stoped working

[hammarw]

bess-debug-2026-05-19-125518.md

What happened

Growatt integration stoped working, I have deleted and added it once more. But it looks different it has not all componets it had before. And are BESS not working anymore.

Debug bundle

Please drag-and-drop bess-debug-2026-05-19-125518.md from your Downloads folder into the editor below before submitting.
The file contains your version, settings, recent logs, and the raw HA discovery dump that the maintainers need to reproduce the issue.

[johanzander]

@claude-bot analyze

[johanzander]

If your growatt integration has stopped working, maybe you should file an issue on the HA core repo for instead?

[bess-manager-claude-bot]

Root cause

When the Growatt HA integration is deleted and re-added, Home Assistant regenerates all entity IDs with a new device serial number prefix. BESS Manager's stored entity IDs in `/data/bess_settings.json` become stale, and every sensor read immediately fails with HTTP 404. The system never prompts the user to re-run the setup wizard because the wizard banner only appears when the sensors dict is completely empty — a re-install leaves the old (non-empty) IDs in place, so the banner is never shown. On top of that, even if a user manually re-runs the wizard, a guard in `settings_store.apply_discovered` silently discards the newly discovered Growatt device ID when one is already stored, leaving inverter control broken after the next restart.

Evidence

• `backend/api.py:2429` — wizard is suppressed whenever any sensor is configured, even if all are stale:

  "wizard_needed": configured == 0,

  After a reinstall, `configured > 0` (old entity IDs are still present), so the wizard banner never appears and the user has no in-product signal to act.

• `core/bess/ha_api_controller.py:416-422` — entity IDs are looked up exclusively from `self.sensors`; no fallback or re-discovery:

  if sensor_key in self.sensors:
      entity_id = self.sensors[sensor_key]
      if not entity_id or not entity_id.strip():
          raise ValueError(f"Empty entity ID configured for sensor '{sensor_key}'")
      return entity_id, "configured"

• `core/bess/ha_api_controller.py:590-600` — 404 responses are not retried; they fail immediately and propagate upward, taking down Battery Control, Battery Monitoring, and Energy Monitoring:

  if (
      hasattr(e, "response")
      and e.response is not None
      and e.response.status_code == 404
  ):
      logger.error("API request to %s failed: Sensor not found (404)...", url)
      raise  # Fail immediately on 404

• `backend/settings_store.py:168-173` — Growatt `device_id` is additive-only; re-running discovery silently discards the newly discovered ID if one is already stored:

  if growatt_device_id:
      growatt = dict(self.data.get("growatt", {}))
      if not growatt.get("device_id"):          # silently skipped when already set
          growatt["device_id"] = growatt_device_id
          self.data["growatt"] = growatt

  The same guard exists for `nordpool_config_entry_id` at line 163. Note that `app.py:283-285` (the in-memory path) correctly overwrites both fields, so they work until the next restart.

Proposed fix

Fix 1 — widen the wizard trigger (`api.py:2429`): Change `wizard_needed` to also return `true` when the system health check reports sensor errors (404s), not only when the sensors dict is completely empty. This gives the user a clear in-product signal immediately after a reinstall breaks entity IDs.

Fix 2 — remove the additive-only guards (`settings_store.py:163, 171`): Drop the `if not …get("device_id"):` guards so that re-running the setup wizard always overwrites `growatt_device_id` and `nordpool_config_entry_id` when discovery returns a new non-empty value. The docstring intent (don't erase on partial discovery) is already handled by the `if growatt_device_id:` outer check.

Both changes are small and isolated. No new classes or parallel implementations are needed.

Risks / open questions

• Wizard trigger threshold: Widening `wizard_needed` to fire on health errors could be noisy if sensors are temporarily unavailable (e.g. HA restart). The condition should probably be scoped to "404 errors specifically" rather than any health failure, to avoid false positives.
• Nordpool reinstall: The same additive guard at line 163 means a Nordpool reinstall has the same stale config_entry_id problem; Fix 2 covers both.
• Settings migration: Users who never re-run the wizard after this fix ships still need a path to recovery — the wizard trigger (Fix 1) handles this automatically.
• The InfluxDB-based energy sensors (historical data) are unaffected and remain functional; only real-time HA sensor reads are broken.

---

Reply `@claude-bot fix` to implement and open a draft PR.

[johanzander]

Actually my growatt integration does not work either so maybe this is a growatt cloud issue

[johanzander]

This is a deeper problem, same issue using the python package directly outside Home Assistant.

Looking into if it is library or growatt issue.

[hammarw]

It looks like they corrected the issue Growatt cloude for the API, it works now.