Prometheus ping metrics

This system measures network latency between its instances. You can deploy it in multiple servers or data centers and collect the latency and packet loss information via Prometheus.

Each instance is given a source name (reported in the metrics), a UDP port (for other instances to ping), and a list of targets (that will be pinged every second).

For example:

cargo build --release
target/release/prometheus_ping --source spire:5000 --listen-port 5000 database:5555
target/release/prometheus_ping --source database:5555 --listen-port 5555 spire:5000

The resulting metrics look like:

ping_packet_loss_total{source="spire:5000",target="database:5555"} 123.0
ping_packet_loss_total{source="database:5555",target="spire:5000"} 27.0
ping_latency_average_30s{source="spire:5000",target="database:5555"} 0.00123
ping_latency_average_30s{source="database:5555",target="spire:5000"} 0.00130

Deploying with Helm

If you have 3 zones alpha, bravo, and charlie, you can deploy like this:

helm install prometheus-ping \
    oci://ghcr.io/remram44/prometheus-ping/helm-charts/prometheus-ping \
    --namespace default \
    --set-json 'locations.alpha={"topology.kubernetes.io/zone": "alpha"}' \
    --set-json 'locations.bravo={"topology.kubernetes.io/zone": "bravo"}' \
    --set-json 'locations.charlie={"topology.kubernetes.io/zone": "charlie"}'