rpi-tig

Dockerized Telegraf/InfluxDB/Grafana for Rpi

I just wanted a fire-and-forget way to quickly monitor a new Pi, this is it. On a Raspbian Buster image, simply:

apt-get install docker.io
# If your distro does not ship the plugin, install docker-compose-plugin

... clone this repository, and do:

./reset.sh
./start.sh

You should be set. After a couple of minutes (it's a Raspberry Pi after all), you will get the following:

• A telegraf instance gathering system stats, with statsd available on the local host for external log input
• InfluxDB 2.x storing the telegraf data
• Grafana 11.x already configured to talk to the InfluxDB instance, and preconfigured with a default system dashboard

I made a moderate effort to at least randomize the Influx database password at first run, and Grafana will remind you to change the admin password. That said, if you put this in production as is, it's really on you. There are many things that should still be done before considering this secure in any way...

Included dashboards

The grafana instance is automatically provisioned with a couple of dashboard that are useful for me, and in any case are hopefully interesting examples of what grafana + influx can do.

• System stats: a Raspberry Pi centric system monitoring dashboard
• goestools: monitors the statistics of a goestools receiver
• ax25 clients: traffic statistics of an Amateur Radio AX25 stack (see additional info below)

AX25 dashboard

Look in the utils directory for the files you should install on your Pi to spawn a listen process that is logged to /var/log and picked by Telegraf.

Notes

• telegraf inside a Docker container will not be able to monitor your network interfaces, obviously
• This stack works remarkably well on a Pi 3B+ (and 4, of course). I've had it monitor devices for months at a time with zero issues
• Thanks to docker-compose/Docker, the stack will survive a reboot
• Provisioned dashboards on Grafana cannot be saved, so they really should be used as templates, and saved as a new Dashboard if you want to edit/modify them
• Feel free to contribute PRs with useful dashboards if you have some

Troubleshooting

• Telegraf "Unauthorized" every ~10s in InfluxDB logs: Telegraf is writing without valid credentials. This repo wires Telegraf to the same env variables as InfluxDB and Grafana via env.influxdb.

  • Ensure env.influxdb exists (created from env.influxdb.default by start.sh).
  • DOCKER_INFLUXDB_INIT_ORG, DOCKER_INFLUXDB_INIT_BUCKET, and DOCKER_INFLUXDB_INIT_ADMIN_TOKEN are injected into the Telegraf container and referenced by telegraf/telegraf.conf.
  • If you see authorization not found, run ./reset.sh and then ./start.sh to regenerate and propagate a valid token.

• Repeated SHOW MEASUREMENTS queries in logs: These are read queries from clients (e.g., Grafana/CLI). Normal and unrelated to Telegraf write authorization.