Closed speedbird182 closed 1 week ago
Thanks for posting this issue, and I'm sorry that you're running into it! The good news here is that this sort of long term, periodic thing is exactly what I need to diagnose the problem, so it's very, very helpful.
I'll investigate and let you know, but I've got a theory about what's happening here
One interesting aspect to this is that it's occurring at 00:00 local time, not UTC like the API runs in. Do you mind letting me know what location you're querying data for?
One other helpful troubleshooting step here- would you be able to add in two hourly pressure sensors, for 0 and 1 hours? That way we could get a better handle about what the API is "seeing" at that moment
I'm querying the API from the UK (Currently UTC+1). I've set up the pressure sensors for hours 0 and 1 as I don't have them configured. I'll let you know how this goes next week (BST ends and goes back to being GMT/UTC this weekend)
Thanks for adding those! We'll keep an eye on it, since those two points will really help pinpoint what's happening
I am also encountering this same issue, but with all the live sensors I'm using from PirateWeather:
Also I cannot find a pattern to how often this is happening, except for the fact that it happens at midnight local time for me as well
Those 3 sensors above are the only ones exhibiting this issue. Here's a list of all the sensors I have enabled:
Let me know if there's anything more I can help with
I'm querying the API from the UK (Currently UTC+1). I've set up the pressure sensors for hours 0 and 1 as I don't have them configured. I'll let you know how this goes next week (BST ends and goes back to being GMT/UTC this weekend)
After the UK went back to GMT this weekend, this issue didn't happen overnight (at least for me it didn't).
This is still happening for me. Apparent temperature's last spike was 800k degrees celsius. HOT! :)
Edit: Romania also went back to GMT+2 but that didn't seem to have any effect.
What a strange bug! It looks like it's happening at 00:00:19 local- can you confirm that it's just for a single timestep, or does it show up for a while?
Ah, ha- got it!!! https://api.pirateweather.net/forecast/APIKEY/47.1756,27.594,1731880810 gives the incorrect values- it must have to do with my daily interpolation near zero. Fixing now
Got it! Float roundoff error in the currently calculation, fixed in version 2.4.0 (live on dev now)
It is happening at 00:00:19 local, just a single reading/timestep. Would you need me to test the dev version to see if it's fixed or should I wait for the official release?
Also, unsure what you're referencing by "version 2.4.0" as the last release today is v1.6.2 and from what I gather it doesn't contain this fix. Please correct me if I'm wrong and thanks for looking into this!
Should have clarified that- the version number I referenced is the API version number that provides the data for this integration, which was what the issue was with here.
I don't think there's any way to easily flip the HA integration over to the new source, but the new API version will go live in the next week, so it'll be fixed shortly
Describe the bug
Every week at midnight on Monday morning (00:00:xx) the pressure reading jumps to over 3000hPa before returning to normal. This reading of over 3000hPa then causes the display history to be completely unreadable.
Expected behavior
values with such a high jump (200% increase) should be ignored/dropped as being invalid.
Actual behavior
Any value is recorded and skews the data
Home Assistant version
2024.10.3
Integration version
1.6
Other details
No response
Troubleshooting steps