openHAB 5.1: averageSince issue

, ,
1

Hi,

I’m running openhab 5.1 since a few days and it looks like averageSince is not working as excpected:

Script:

const past10Minute = time.ZonedDateTime.now().minusMinutes(10)
const itmSolar = items.getItem("mqtt_solar_energy_current")
const currentValue = itmSolar.rawState
const solarLastHourState = itmSolar.persistence.averageSince(past10Minute)
const solarLastHour = parseFloat(itmSolar.persistence.averageSince(past10Minute).rawState)
const solarCount = itmSolar.persistence.countSince(past10Minute)
const solarSum = parseFloat(itmSolar.persistence.sumSince(past10Minute).rawState)
const solarAvg = solarSum / solarCount

console.log("averageSince(", past15Minute, ") =", solarLastHourState)
console.log("averageSince(", past15Minute, ").rawState =", solarLastHour)
console.log("countSince(", past15Minute, ") =", solarCount)
console.log("sumSince(", past15Minute, ") =", solarSum)
console.log(" ", solarSum, "/", solarCount, "=" , solarAvg)

Console output:

averageSince( 2026-05-16T18:58:09.645+02:00[Europe/Berlin] ) = PersistedState (State=-730.9234180105466 W)
averageSince( 2026-05-16T18:58:09.645+02:00[Europe/Berlin] ).rawState = -730.9137161275914
countSince( 2026-05-16T18:58:09.645+02:00[Europe/Berlin] ) = 2
sumSince( 2026-05-16T18:58:09.645+02:00[Europe/Berlin] ) = 1393.9
1393.9 / 2 = 696.95

The corresponding data in the database (MariaDB)

image

The sum is correct: 671.7 + 722.2 = 1393.9. How could a average of only positive values be negative?

Has anyone any ideas?

2

What is the current state of your item? `averageSince` may also consider the current state if different from the last persisted state. `averageSince` also calculates a weighted average. So it will consider the 5 minutes between the 2 considered values for the first value and the time to now for the last value. `Sum` does not consider any time dimension (and therefore is of limited use in my view, the better methods are `RiemannSum`).

I still don’t understand why it would be negative though.

3

Thanks for your input. In my case it makes no difference between Sum and RiemannSum because the values are stored continuously every 5 minutes. Therefore the time a value is present is constant, ergo equals weights. I also intentionally pick a time where the current item state equals the last state stored. The only difference could be, that the current state is considered twice (last and current). But the average differs a lot even if the backend consider the last state twice.

I will play the next days a little bit more with the persistence functions und see what I could find out. It looks like, that I’m the first who run into this issue.

4

It will make a difference as the current time is never exactly 5 minutes after the last persisted value. And the last value will only be considered proportionally for the time between last persisted value and now. The same is true for averageSince.

5

From a mathematical POV, you are right, but let’s say it this way: In my case it will make no significantly difference for me. It doesn’t matter here for me if the value is 1234.12 or 1238.21. Because I want only get a rough estimate about the produced power from my little solar power plant for my own interest. But what matters is, that I got sometimes an average of 4000 where I would expect 1234.12 and sometimes -700, which makes no sense at all…

6

Can you switch on debug logging or org.openhab.persistence.jdbc and org.openhab.core.persistence? The retrieved values and calculations are not logged I believe, but at least the SQL query is and the number of values retrieved. Maybe there is something to be learned from that.

7

I’ve found the issue: For some reason the timezone of the OS (Ubuntu Server 24.04 LTS) was set to UTC. I set up openHAB with timezone UTC+1 (Berlin, Rome etc.). Therefore there exists a miss match in the set up timezones. After changing the OS timezone to UTC+1 and reboot the machine, everything worked fine.