Odd behaviour with web control

[Sylvia Else]

I've been controlling my powerwall 2 via its web interface for getting on for two years. Come the start of the off-peak tarrif period in the evening, the reserve gets set to the current charge, and grid power is used until the end of that period in the morning, whereapon the reserve is set to 50%, and the home is powered from the battery and solar. This way, I avoid the higher tarrifs, while not wearing out the battery by using it when its hardly worthwhile financially.

This has mostly worked fine.

But there have been a couple of hiccups recently, where the battery continued to discharge through the night, even though the gateway is showing that the correct reserve has been set. Even more oddly, this morning when the reserve was set back to 50%, the battery started charging from the grid, as if it had suddenly noticed the reserve set yesterday evening.

However, despite no further changes to the reserve setting during the day, it acted as expected this evening after the sun went down, and continued to power the home from the battery even as the charge level dropped below the reserve set yesterday evening, so it wasn't the case that it had somehow lost the 50% reserve setting.

The behaviour seems most odd. I"m almost wondering whether Tesla have been fiddling with it remotely.

Anyone had a similar experience?

I'm on version 1.43.3

 

[BJReplay]

Yep, there's a thread going here: Complete now returning 200 (1.43.3) · Issue #26 · vloschiavo/powerwall2.

It looks like calls to the API no longer stick. A second call might make the first call take effect, resulting in the behaviour you noticed.

 

[Sylvia Else]

Yep, there's a thread going here: Complete now returning 200 (1.43.3) · Issue #26

Thanks for the link.

 

[Sylvia Else]

Yep, there's a thread going here: Complete now returning 200 (1.43.3) · Issue #26 · vloschiavo/powerwall2.

It looks like calls to the API no longer stick. A second call might make the first call take effect, resulting in the behaviour you noticed.

I've found that repeating the /api/config/completed call as soon as the first one has responded seems to make things work. Quite what Tesla could have done to cause this behaviour is hard to imagine, unless it was actually deliberate.

 

[Sylvia Else]

I've found that repeating the /api/config/completed call as soon as the first one has responded seems to make things work.

Unfortunately, I seem to have spoken too soon. While doing it twice appears to give a greater chance of working, it's still not completely effective.