I’m trying to use InfluxDB 1.5 on a Raspberry Pi3 (quad-core ARMv7 1.0GHz, 1GB RAM) to store high-frequency sensor data (5 sensors * 200 sample/s * 3 measurements/sensor * 16-bit/measurement).
So I’m writing about 3000 points/sec on the /write endpoint, in 5 batches/s with 200 rows/batch, which Influx seems to be handling fine (except for some high-figure virtual memory usage, ~900MB and some high CPU load).
When I try to run even the simplest query trough the Influx CLI (trying to get 3 minutes worth of the above data):
influx -database sensors -execute “select x,y,z from meas where time > ‘2018-07-25T13:53:00Z’ and time < ‘2018-07-25T13:56:00Z’ group by name”
it takes several (30-60) minutes…
Any idea if there’s anything I can tweak? Am I pushing this too hard?