Inconsistent Load Shedding State Updates in v1.5.1

[CntrlAltDel45]

Home Assistant Version: 2025.3.1
Integration Version: v1.5.1

Issue Description
I'm experiencing inconsistent reporting of localized load shedding states with the ha.integration.load_shedding integration.

• The state correctly switches to "On" when load shedding begins.

• However, it does not reliably switch back to "Off", sometimes for hours or even days.

• Reloading the integration updates the state immediately to the correct value.

Additional Observations

• The history bar correctly indicates that the current state should be "Off", but the actual state remains "On".

• Initially, I suspected an ESP API quota issue, but this does not seem to be the cause.

• This issue affects automations that rely on state changes (On → Off and Off → On) as triggers.

Question
What could be causing this behaviour, and is there a known fix or workaround?

[tinuva]

Can confirm I have the same issue.

Essentially what happen is, when loadshedding start, the entries are added to the calender correcty.
When loadshedding ends, eg. both capetown and eskom states change the No Load Shedding the calendar entries are not removed. So automations go on, as if loadshedding is still active.

I guess to fix the automations, a simple fix is to check the current level of loadshedding etc. But the calendar really should update.

[CntrlAltDel45]

Can confirm I have the same issue.

Essentially what happen is, when loadshedding start, the entries are added to the calender correcty. When loadshedding ends, eg. both capetown and eskom states change the No Load Shedding the calendar entries are not removed. So automations go on, as if loadshedding is still active.

I guess to fix the automations, a simple fix is to check the current level of loadshedding etc. But the calendar really should update.

Have you tried reloading? Seems to be a workaround fix for me. I might just use a script to just reload every 5 minutes or so. Just worried it might cause an unexpected uptick in API queries.

[tinuva]

Can confirm I have the same issue.
Essentially what happen is, when loadshedding start, the entries are added to the calender correcty. When loadshedding ends, eg. both capetown and eskom states change the No Load Shedding the calendar entries are not removed. So automations go on, as if loadshedding is still active.
I guess to fix the automations, a simple fix is to check the current level of loadshedding etc. But the calendar really should update.

Have you tried reloading? Seems to be a workaround fix for me. I might just use a script to just reload every 5 minutes or so. Just worried it might cause an unexpected uptick in API queries.

Yeah that does work. I wouldnt do a reload every 5 minutes, sounds like a great way to deplete your 50 request limit for the day.

Rather, I would look at automating it to reload when the state goes to No Load Shedding.

[CntrlAltDel45]

Can confirm I have the same issue.
Essentially what happen is, when loadshedding start, the entries are added to the calender correcty. When loadshedding ends, eg. both capetown and eskom states change the No Load Shedding the calendar entries are not removed. So automations go on, as if loadshedding is still active.
I guess to fix the automations, a simple fix is to check the current level of loadshedding etc. But the calendar really should update.

Have you tried reloading? Seems to be a workaround fix for me. I might just use a script to just reload every 5 minutes or so. Just worried it might cause an unexpected uptick in API queries.

Yeah that does work. I wouldnt do a reload every 5 minutes, sounds like a great way to deplete your 50 request limit for the day.

Rather, I would look at automating it to reload when the state goes to No Load Shedding.

I was just treating the idea that reloading causes an API query uptick. I've not seen this behaviour when I've reloaded. But regardless, it would be better to use the "No Load Shedding" state for the city level (and not the neighbourhood level) as you've mentioned.

[wernerhp]

Yeah, I noticed there's an issue.
I'm testing a fix, but need load shedding to test 😂
Maybe downgrade to the previous version if it's critical, otherwise I'll post a beta with maybe a fix.

[tinuva]

Yeah, I noticed there's an issue. I'm testing a fix, but need load shedding to test 😂 Maybe downgrade to the previous version if it's critical, otherwise I'll post a beta with maybe a fix.

Sounds like, you need to build yourself a mock server, to allow you to do this testing without waiting for actual loadshedding.