resin-influx

A basic example of running InfluxDB via resin.io.

Logs various data (load averages, memory usage, /data partition usage) via Python -script to the database. This database can then be queried via the built-in webinterface, at http://<resin-ip>:8083

Performance

This example was mainly done as a test of InfluxDB performance on RPi, so here:

Raspberry Pi B+:

• writing a single measurement (9 values): ~20-30 milliseconds

• querying ~130 000 load averages (last 12 hours) and calculating 15 minute averages: ~1.2 seconds select mean(value) from load_avg where time > now() - 12h group by avg_time, time(15m)

• with 1 minute averages: ~2.3 seconds select mean(value) from load_avg where time > now() - 12h group by avg_time, time(1m)

• with 15 minute averages, last 2 hours: ~0.25 seconds select mean(value) from load_avg where time > now() - 2h group by avg_time, time(15m)

• with 1 minute averages, last 2 hours: ~0.5 seconds select mean(value) from load_avg where time > now() - 2h group by avg_time, time(1m)

• querying free memory, load average, and free disk space for the last 2 hours, 15 minute averages: ~0.25 Note: 1 query!

  select mean(value) / 1024 / 1024 from memory where type = 'available' and time > now() - 2h group by time(5m);
  select mean(value) from load_avg where avg_time = '1m' and time > now() - 2h group by time(5m);
  select mean(value) / 1024 / 1024 from disk_usage where type = 'free' and time > now() - 2h group by time(5m)
  

• memory usage: ~53 % -> 255 MB

• ~260 000 points stored in the database, in 9 series