Query on time registers

[stevebuk1]

As a followup to #62, I've now integrated your integration into Pv_opt (https://github.com/fboundy/pv_opt) as an option for controlling Solis inverters via SolisCloud and its under beta test.

My understanding of "After changing an inverter setting, the integration reads back the updated value from the inverter to verify that the change was applied successfully" means that if I use AppDaemons "Call_service" functionality (essentially the HA action "Text: Set value") then this will wait for inverter ringback before giving a response?

It appears that way as the response from Appdaemons "call_service" is the following:

Result of call_service is {'id': 7946, 'type': 'result', 'success': True, 'result': {'context': {'id': '01K1YE95W708QRHDB7B73ZVN5K', 'parent_id': None, 'user_id': '6d27b37a82da4e1f9c7f0be1f0e5610b'}}, 'ad_status': 'OK', 'ad_duration': 4.243782655335963}

The ad_duration field is the number of seconds it takes to return the response (I think!) which most of the time is around the 4-5 second mark. This feels right if solis-cloud-control writes to SolisCloud via API, then the inverter, then wait for the inverter to respond, then tell the API, then respond back to solis-cloud-control.

The issue however is that occasionally appdaemon times this out, as it has a limit of 10seconds for updating HA entities (which cannot be changed).
The message produced is

06/08, 01:20:58 WARNING HASS: Timed out [0:00:10] waiting for request: {'type': 'call_service', 'domain': 'text', 'service': 'set_value', 'service_data': {'value': '01:30-01:30'}, 'target': {'entity_id': 'text.solis_slot1_discharge_time'}, 'id': 6477}

It is a warning, and the write does go through to the inverter. My query is, if I were to access your integration via a custom service action as offered in #62, do you think this will improve things? (I've read that using the HA actions services are pretty slow). I guess my question really is, what is the average time delay in seconds that the verifies usually take?

[mkuthan]

The current algorithm of controlling inverter follows this:

• send control command and wait for the response up to 30 seconds
• if Solis API fails, retry after 5 seconds up to 2 times
• read the updated settings and wait up to 30 seconds
• if Solis API fails, retry after 5 seconds up to 2 times
• verify updated settings, if they don't match, wait 5 seconds and retry everything again one more time

As you see, in the worst case scenario it could take minutes to finish.

References:
https://github.com/mkuthan/solis-cloud-control/blob/main/custom_components/solis_cloud_control/coordinator.py#L72-L102
https://github.com/mkuthan/solis-cloud-control/blob/main/custom_components/solis_cloud_control/api/solis_api.py#L114-L145
https://github.com/mkuthan/solis-cloud-control/blob/main/custom_components/solis_cloud_control/api/solis_api.py#L114-L145

[stevebuk1]

Thanks for your reply. I tried doing a time register update manually via the HomeAssistant GUI using Developer Tools -> Actions -> Text: Set value and it does appear to take around 5 seconds to respond, and I'm assuming thats with a write going through with no delays. I think for now the rate of warnings experienced doesn't warrant any further action.

I've been running Pv_opt for a week now for controlling my Solis inverter with no issues, I'm on Octopus Intelligent Go and use Pv_opt to ensure the house battery doesnt discharge during EV charge, as well as topping up the house batttery based on forecast consumption and forecast solar for the next day.

For background, the main issue is an entity within Solis-sensor that reports house consumption in kWh is supposed to be incrementing and it doesnt always, meaning the forecast house consumptions are well off. This is apparently a Solis API issue rather than a solis-sensor issue, but this issue will seriously limit PV_opt functionality. The Solis app and downloading house consumption data from SolisCloud are perfectly fine (as is accessing the data directly from the inverter via Modbus).