GOAL
Implement Pingdom type functionality with https://prometheus.io/ with least LOC possible.
wc -l *.{yml,service} prometheus/* | tail -n1
129 total
- Monitor lots of Websites: foo.example.com, bar.example.com, google.com ....
- Email when site goes down with alertmanager & AWS SES
- Graph history of outages ... http://%H:9090/graph?g0.range_input=1h&g0.expr=probe_success&g0.tab=0
Features:
- Keeps upto date between reboots
- Containerized
Setup
Assuming you are sshing to a CoreOS machine. "pingprom" requires systemd & Docker.
ssh core@ip
# docker network create pingprom
# git clone https://github.com/kaihendry/pingprom.git
# export PINGPROM=$(readlink -f pingprom) # make a note where the checkout is
# echo PINGPROM=$PINGPROM > /etc/default/pingprom
# cd /etc/systemd/system
# for i in $PINGPROM/*.service; do ln $i; done
# systemctl daemon-reload
# systemctl start $(for i in $PINGPROM/*.service; do echo $(basename $i); done)
# systemctl enable $(for i in $PINGPROM/*.service; do echo $(basename $i); done)
Now you need to systemctl status or journalctl -b -f to debug the failing ones and once everything looks OK.
Caddy configuration for nicer URLs
prom.example.com {
tls youremail@example.com
proxy / prom:9090
}
alerts.example.com {
tls youremail@example.com
proxy / alertmanager:9093
}