Looks like my Ecobee 3 Lite got a firmware update today. Since the update, the thermostat detail is not updating. Sensor detail is updating, so it looks like something in the new firmware is keeping the thermostat data from updating.
Same here… I’m missing a block of data. Started at 10am, but now it’s reporting again.
Same issue here: Over the last 48 hours, there have been two large gaps of data (each spanning hours) in the Beestat application for my Ecobee 3 Lite (firmware 220.127.116.11). The first gap occurred Wednesday 3/22 from 8:55 am to 3:30 pm CST. The second gap began Friday 3/24 at 2:55 am CST and it is still continuing as of 6:50 am CST on 3/24.
This is primarily an issue with ecobee. On Wednesday (and now this morning) they were having issues with their weather data. If you look on ecobee’s Home IQ portal, you can see the chunks of data where outdoor weather is missing.
For technical reasons, beestat throws away all thermostat data when the outdoor weather is missing. This is typically not an issue so I never spent the time and effort trying to account for it. Hopefully ecobee gets this resolved and then we can put the bad data behind us.
Sorry for any inconvenience!
Here’s ecobee’s status page: https://status.ecobee.com/
According to an Ecobee employee on Reddit, the root cause is the Dark Sky API outage. Also, since that API goes away soon, they’re in the process of transitioning to a new provider. It might be turbulent for a while.
@ziebelje By discarding the data, the heat/ac runtime numbers are skewed. Looking at the homeiq raw data, the heat runtime numbers are included.
If this data was omitted then I can see discarding the data entirely when the weather numbers are unavailable. However it doesn’t seem to make sense to do so if/when it is available.
Yes, it does skew the data. So far we’re up to ~10 hours this month which could, at most, miss 1% of runtime for the month. It’s not perfect, but given how few issues there have historically been, the simplicity was worth it.
If you’re curious, the all/none approach affects the thermostat summary aggregation. It was built in a way that allowed me to do a single query to populate that data for any date range. It would produce inaccurate results if you had some data but not others, so I made the decision to just throw the data away unless it was all present to make it simpler, given that there were generally very few problems with it.
Right or wrong that’s why it was built that way. If this is a persistent problem moving forward I’ll consider updating it.