svalouch
commented
2 years ago Thank you for this PR, I've added some comments.
I am not using influx so I don't know if that works. Did I see correctly that influx is just re-using the prometheus metrics?
Two-fold:
- Tables are filled if
collect_influx()is implemented and wired in, so it doesn't happen automatically, as what belongs together can't be deduced from the names alone in some cases. I'm not using those tables myself, and haven't implement most of them, actually. - All values that are correctly received are immediately added to the internal buffer and end up in a "raw" table. This includes string, float, enum, etc. types, and was originally meant for debugging, but I've found it to be very convenient to work with in Grafana: Simply request the points where the name matches one of the internal names to populate a panel in Grafana.→ done. So even if
collect_influx()is not wired in, all the values are automatically pushed to the database if it is enabled. I use this as long-term storage, and Prometheus for alarming.
This PR adds two new prometheus metrics: a) Add total energy counters for inverter, grid feed, grid load, household, generators b) Add insulation metrics (total insulation, DC+ and DC-)
Tested successfully against my inverter.
I am not using influx so I don't know if that works. Did I see correctly that influx is just re-using the prometheus metrics?