Spurious API failures cause loss of functionality

[lingwooc]

Version of the custom_component

1.6.2
Core 2025.1.2
Supervisor 2024.12.3
Operating System 14.1
Frontend 20250109.0

Bestway device

Lazy Spa Milan

Describe the bug

I don't suppose this is a bug in ha-bestway as such but it could handle it better.

It seems like the bestway api since about 1pm 9/1/25 has been wildly unreliable, throwing 400s for fun (see logs). Its possible its just me and I've a faulty pump but I can't tell that. The issue is that the integration interprets this as the device being unavailable which blocks automations when the post apis are probably up, or would be with a few retries.

I wonder if a bit of tollerance of the api being a bit flaky is in order. Maybe the BestwayUpdateCoordinator just keeps the current status unless there are 3 or 5 consecutive failures, then you pick up the genuine issues and longer lasting outages. If you still log the failed responsed then you still have visibility.

A don't know if the post api is having a moment too but a sensible retry approach to that might have value.

Logs

2025-01-10 09:53:44.546 DEBUG (MainThread) [custom_components.bestway.bestway.api] Device list refreshed: {"devices": [{"protoc": 3, "ws_port": 8080, "port_s": 8883, "gw_did": null, "host": "eum2m.gizwits.com", "sleep_duration": 3600, "port": 1883, "mcu_soft_version": "P151D102", "product_key": "********************************", "state_last_timestamp": 1736502644, "role": "owner", "is_sandbox": false, "type": "normal", "product_name": "Airjet", "is_disabled": false, "mcu_hard_version": "P150D100", "wifi_soft_version": "0402003A", "dev_alias": "Hot tub", "mesh_id": null, "is_online": true, "dev_label": [], "wss_port": 8880, "remark": "21", "did": "**********************", "mac": "************", "passcode": "**********", "wifi_hard_version": "00ESP826", "is_low_power": false}]}
2025-01-10 09:53:44.615 DEBUG (MainThread) [custom_components.bestway.bestway.api] New data received for device OkJNmZYM5wDNwE2AsyuTaJ
2025-01-10 09:53:44.616 DEBUG (MainThread) [custom_components.bestway.bestway.api] Status for device type 'Airjet' returned: {"system_err2": 0, "wave_appm_min": 59940, "heat_timer_min": 0, "heat_power": 1, "earth": 0, "wave_timer_min": 59940, "system_err6": 0, "system_err7": 0, "system_err4": 0, "system_err5": 0, "heat_temp_reach": 0, "system_err3": 0, "system_err1": 0, "system_err8": 0, "system_err9": 0, "filter_timer_min": 0, "heat_appm_min": 0, "power": 1, "temp_set_unit": "\u6444\u6c0f", "filter_appm_min": 0, "temp_now": 35, "wave_power": 0, "locked": 1, "filter_power": 1, "temp_set": 35}
2025-01-10 09:53:44.616 DEBUG (MainThread) [custom_components.bestway.coordinator] Finished fetching Bestway API data in 0.285 seconds (success: True)
2025-01-10 09:54:14.472 ERROR (MainThread) [custom_components.bestway.coordinator] Error requesting Bestway API data: 400, message='BAD REQUEST', url='https://euapi.gizwits.com/app/bindings'
2025-01-10 09:54:14.473 DEBUG (MainThread) [custom_components.bestway.coordinator] Finished fetching Bestway API data in 0.142 seconds (success: False)

[j0rdan]

Glad I saw this.
I'm experiencing the same.

Last weekend I had to reset my tub, and thought maybe that had something to do with it, but you having this issue confirms to me it not on my end.

My logs seem to cycle between connected and unnavailable every 40 seconds or so.
Any sort of retry would be appriciated to make it a bit more resiliant for automations.

I'm in the UK and on Octopus Agile, so I have automations based on electricity price, and todays price is crazy high, so I'm keeping an eye on the hottub to make sure it turns off when its supposed to :-D

[lingwooc]

I am doing exactly the same thing with octopus go and missed my window last night (unfortunate as it was -9!). I'll probably schdule it with the app tonight.

I didn't metion in the original report, the app is fine so I'm reasonably sure its not my hot tub which is at fault, moreso since @j0rdan is having the issue too.

[danielsmith89]

Having exactly this issue too, thought at first maybe the cold weather was affecting the WiFi to the pump, but clearly it looks like an API issue instead.

[NuttyNorm]

I have the same over the past few days, been rocksteady up till now, my log shows:
Error requesting Bestway API data: 400, message='BAD REQUEST', url='https://euapi.gizwits.com/app/bindings'

[davelegg83]

Just came to say i'm having the exact same issues as everyone above.

`This error originated from a custom integration.

Logger: custom_components.bestway.coordinator
Source: helpers/update_coordinator.py:392
integration: Bestway (documentation, issues)
First occurred: 10 January 2025 at 11:55:11 (526 occurrences)
Last logged: 07:33:34

Error requesting Bestway API data: 400, message='BAD REQUEST', url='https://euapi.gizwits.com/app/bindings'`

[leeshellard]

Just come to report the same. I’ve had to disable automations in the last few days as it’s become unreliable. Constant connecting and disconnecting. As reported still working great by the app seems to just be HA integration.

[lingwooc]

I have a hack but it involves editing the integration's source directly using the Studio Code Server addon.

The trouble seems to be here https://github.com/cdpuk/ha-bestway/blob/main/custom_components/bestway/coordinator.py. The fix is to replace the _async_update_data function in custom_components/bestway/coordinator.py (from line 28 onwards) with:

    async def _async_update_data(self) -> BestwayApiResults:
        """Fetch data from API endpoint.

        This is the place to pre-process the data to lookup tables
        so entities can quickly look up their data.
        """
        async with asyncio.timeout(10):
            try:
                await self.api.refresh_bindings()
            except Exception as e:
                pass  # Ignore failures on refresh_bindings

            return await self.api.fetch_data()

The troublesome endpoint seems to be https://euapi.gizwits.com/app/bindings which is called by await self.api.refresh_bindings(). This just wraps await self.api.refresh_bindings() with a try/except and ignored any errors. But as far as I can tell, apart from autodetection of new things, its not actually needed to run every poll. Its possible there is a new rate limit and they are sending the wrong code (should be a 429).

You could also just remove line 35 https://github.com/cdpuk/ha-bestway/blob/main/custom_components/bestway/coordinator.py#L35.

After making changes restart home assistant. If you get it wrong then the integration will break but HA will be just fine. If you really break it just reinstall the addon.

Early days but it seems stable now for me. I'm not doing a PR as I'm not used to HA integrations so I don't know best practice for fixing this (and this won't be that).

[danielsmith89]

I have a hack but it involves editing the integration's source directly using the Studio Code Server addon.

The trouble seems to be here https://github.com/cdpuk/ha-bestway/blob/main/custom_components/bestway/coordinator.py. The fix is to replace the _async_update_data function in custom_components/bestway/coordinator.py (from line 28 onwards) with:

    async def _async_update_data(self) -> BestwayApiResults:

        """Fetch data from API endpoint.



        This is the place to pre-process the data to lookup tables

        so entities can quickly look up their data.

        """

        async with asyncio.timeout(10):

            try:

                await self.api.refresh_bindings()

            except Exception as e:

                pass  # Ignore failures on refresh_bindings



            return await self.api.fetch_data()

The troublesome endpoint seems to be https://euapi.gizwits.com/app/bindings which is called by await self.api.refresh_bindings(). This just wraps await self.api.refresh_bindings() with a try/except and ignored any errors. But as far as I can tell, apart from autodetection of new things, its not actually needed to run every poll. Its possible there is a new rate limit and they are sending the wrong code (should be a 429).

You could also just remove line 35 https://github.com/cdpuk/ha-bestway/blob/main/custom_components/bestway/coordinator.py#L35.

After making changes restart home assistant. If you get it wrong then the integration will break but HA will be just fine. If you really break it just reinstall the addon.

Early days but it seems stable now for me. I'm not doing a PR as I'm not used to HA integrations so I don't know best practice for fixing this (and this won't be that).

Worked great for me thanks. Had to reload the integration once after HA restarted to get it working but after that all good.

[liam-london]

I have also been battling with this, and the solution above appears to fix it for me, at least for now.

[purp23]

Solution above fixed it for me as well, awesome thank you.

[j0rdan]

same here, the solution seems to have settled the 'unnavailable' and has allowed my automations to run as expected.

[normanhinks]

The solution above worked for me also. Many Thanks.

[lingwooc]

To make it easier here is a gist with the full contents of that file https://gist.github.com/lingwooc/17b6f3b0e57ad552b12dfd8f0e32e8eb This replaces custom_components/bestway/coordinator.py and has been working for a couple of days for me at least.

[Stealth100]

To make it easier here is a gist with the full contents of that file https://gist.github.com/lingwooc/17b6f3b0e57ad552b12dfd8f0e32e8eb This replaces custom_components/bestway/coordinator.py and has been working for a couple of days for me at least.

Hi all been trying to get rid of this error and get my automations back for a few days, just tried to copy and paste this over and restarted, it seemed to be fixed but now back as undefined in ha and automations not working again.

[j0rdan]

To make it easier here is a gist with the full contents of that file https://gist.github.com/lingwooc/17b6f3b0e57ad552b12dfd8f0e32e8eb This replaces custom_components/bestway/coordinator.py and has been working for a couple of days for me at least.

Hi all been trying to get rid of this error and get my automations back for a few days, just tried to copy and paste this over and restarted, it seemed to be fixed but now back as undefined in ha and automations not working again.

Did you also reload the device after restarting HA as suggested by some here?
Settings -> Devices & Services -> Bestway -> 3 dots and 'Reload' by you devide entry