In particular, queries to retrieve datapoints from InfluxDB have some subtleties:
- must use InfluxDB, not rootbench-InfluxDB as source (unless you want old data)
- must select a single nodelabel and build configuration in the WHERE clause
- python benchmarks store runtimes in the
duration field, C++ benchmarks use real_time
Oh and you need to press "Save" or lose all your changes :)