How can I populate /var/prometheus/*.prom file(s) to graph out disk metrics? Is there some documentation I can review?
- Storage Used
- blks in/out
- IO Wait
- ...
Maybe I have not understood well, but you're using your hardware/router also as/for a Prometheus database?
Anyway you need to install the exporters/collectors via opkg, edit/configure the node exporter
root@R4S:/etc/config# cat prometheus-node-exporter-lua
config prometheus-node-exporter-lua 'main'
option listen_interface '*'
option listen_port '9100'
option listen_ipv6 '0'
The collectors are here:
root@R4S:~# ls /usr/lib/lua/prometheus-collectors/
conntrack.lua filefd.lua nat_traffic.lua netstat.lua uname.lua
cpu.lua loadavg.lua netclass.lua time.lua wireguard.lua
entropy.lua meminfo.lua netdev.lua uci_dhcp_host.lua
And after you will find the DB in your location
And remember to specify in prometheus.yml the correct targets/IP of the router.
Giuliomagnifico, thank you for replying.. I have the packages you specify installed, sending data to prometheus and dashboards created in Grafana.
My dashboards are missing disk/storage metrics. How can I use prometheus-*-textfile to collect disk data? I want to create a bash script to generate metrics and dump them into a text file, but I'm not sure how to begin..
Oh okay, only these are missing, unfortunately (as you can see from my folder above) they aren't availabe on opkg, you have to write them, or... I use collectd for those metrics, you can build exactly the same graphs in Grafana, the packages are
collectd-mod-disk
collectd-mod-df
(I'm not using the mod-df collector).
I'm using collect for almost all metrics
No I'm using only Prometheus DB, sorry I forgot to mention the Collectd Exporter fo Prometheus, my mistake. I wrote some time ago how to achieve it, here.
Download the Collectd Exporter fo Prometheus package for your architecture, place the collect_exporter where you want, then enable it in your server/router/whatever. (I'm doing it in a non optimal way in crontab)
@reboot cd /home/dietpi/prometheus && ./collectd_exporter --collectd.listen-address=":25827"
And you have also to specify to the collectd network plugin on OpenWrt, where to write the metrics:
(never have a crash or anything in about one year).