We now try switching to InfluxDB v2 with Flux query language. It looks to my like the perfomance is dramatically low in comparison to the old access method. OpenEMS UI History views take much longer to build up. So right now I am not so happy with this decision. I am curious:
- Is someone else experiencing this performance issues?
- What is the reason for the switch to the Flux query language?
Another Background for my questions (also a hint for others):
I can’t find the other post, but someone was complaining about wrong energy values a few weeks ago. The reason for this could be the Influx “GROUP BY” command. A short time ago we solved this problem in Influx 1.6 by adding a resolution offset (see Time Boundary in InfluxDB Group by Time Statement · KK's Blog (fromkk) for more): We wanted to do a pull request. But due to the switch to Flux query language this pull request is not needed anymore.
But the problem with the monthly energy values is the same for flux query language. So a resolution offset is needed here too. Unfortunately there seems to be no easy solution now. Because in the flux java library no “resolution offset” method is included. We are still investigating this.