[Feature Request]: Support the Flux query interface for InfluxDB

[mosche]

What would you like to happen?

Unfortunately InfluxQL is very limited. Many, rather basic types of queries can only be done using the Flux query interface. E.g. extracting parts from a string for readability or also grouping by a dynamic set of columns as required to handle Java microbenchmark (JMH) params.

Grafana supports Flux running on InfluxDB 1.8+. See 1.8 compatibility for more information and connection details.
https://grafana.com/docs/grafana/v8.1/datasources/influxdb/influxdb-flux/

For 1.8 it has to be enabled using INFLUXDB_HTTP_FLUX_ENABLED=true
Though, it looks like the versions are incompatible. I wasn't able to successfully test it without upgrading InfluxDB to 2.x. But it might have been misconfiguration:

panic: column retentionPeriod:int is not of type string

Here's an example:

import "regexp"

from(bucket: "beam_test_metrics/a_year")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> truncateTimeColumn(unit: 1w)
  |> filter(fn: (r) => r["_measurement"] == "java_jmh_thrpt" and r["benchmark"] =~ /ByteStringOutputStreamBenchmark/ and r["_field"] == "score")
  |> map(fn: (r) => ({r with name: regexp.findString(v: r.benchmark, r:/\w+\.\w+$/)}))
  |> drop(columns: ["benchmark", "scoreUnit", "_field", "_measurement"])
  |> group(columns: ["_time", "_value"], mode: "except")
  |> aggregateWindow(every: 1w, fn: mean, createEmpty: true)
  |> fill(value:0.0)

Issue Priority

Priority: 2

Issue Component

Component: community-metrics