Just wondering if there's been any movement on the delta statistics issue from a few months back?
Associated thread: https://forum.infiniteautomation.com/topic/4413/rollup-value-calculations-and-no-values-for-15m-intervals
Associated public github issue: https://github.com/infiniteautomation/ma-core-public/issues/1480
Unfortunately, this issue is causing us problems as our clients' primary need is for correct energy usage statistics/graphs.
Looking at the time period for your delta...
have you tried 901 seconds or even 900001 milliseconds for your roll-up period?
I had an issue with trying to obtain a single cumulative value for an entire year and doing days for some reason caused me to be a day short due to bookmarking. So I did it as milliseconds, ensuring it engulfed the time period I needed and nothing more.
Alternatively, I looked at the git repo, it looks like nothing has been done, and since no "checking is applied" for the previous time period, so perhaps "hacking" with the maMoment filter to subtract an additional 15 minutes/900 Seconds so you are able to get that first delta value...
from="from | maMoment : 'subtract' : 15 : 'minutes'"
Would you mind humouring me?
@mattfox Couldn't see much change unfortunately. When graphing at 1 hour deltas, at best it just brings the previous hour into the graph.
@terrypacker you're up!
Yes there are some issues with the statistic, I confirmed that today. I've just escalated this for Mango 4.0 but I didn't have much luck getting it working today and I'm out of time for now. The problem is how we handle the change between statistic periods as previously mentioned. However we do now track the last value from the previous period to feed into the next but this does not seem to be enough to solve the problem.
@terrypacker Hi Terry, just wondering if there's something we can do from our end to help solve the issue. We're currently in the process of introducing new clients but as our main interest is consumption, the incorrect deltas are detrimental to us.
You could use meta data points to compute the delta. This would mean you will have 2 data points for every consumption you want to track. The real data coming in and the processed data that you can chart. I'm not sure of your proficiency with meta data points but the basic idea is:
- Setup meta point to have your consumption raw data as a context point that updates the context
- Create a script that will compute the delta the way you want it, you can access previous values by sample number by using the
lastValue(...)style methods on the context points. This way you could get the previous value for your raw data by using
lastValue(1,true)where the second variation will use cached values which may be what you want depending on your logging settings.
- Just return the computed delta.
Would that sort of concept work?
@terrypacker Hi Terry, yup that's exactly what we've done in the meantime. Thanks for your help.
I'm running into trouble when new data gets imported though. See https://forum.infiniteautomation.com/topic/4867/meta-point-classcastexception