Edit: for clarification: downsampling is done server-side, in timed intervals. We keep separate datasets with downsampled data, and downsampling is done using a cascading approach.
Zooming in is one of these things that’s technically possible, but not yet made user-friendly. The downsampled result set is selected automatically, based on how many points would be returned for your query. Zooming in can be done by narrowing the shown period in your graph.
As reference, you can select the “Combined Influx Points” metric. This keeps track of the downsample rate (X points downsampled to 1).
I can see about adding some settings for overriding the current sanity checks (disable downsampling, disable truncating result set at 2000 points), but the alternative is probably preferable anyway.
We use influxDB as the backing database for history data. The UI uses the history service as gateway / query builder, but nothing’s stopping you from directly accessing the database itself.
Some options available are:
- Publishing the InfluxDB port (8086) in docker-compose.yml for access to the HTTP API.
- Running the influx CLI client inside the docker container:
docker-compose exec influx influx
- This can be used to export to CSV
- Connecting Grafana, a big and shiny suite of data visualization tools.