kube-network-monitor

A Kubernetes DaemonSet that continuously pings any other node in the cluster and exposes the reachability and latency as Prometheus metrics.

Metrics

• network_monitor_reachable: Gauge that is either 1 (at least some packets make it through) or 0
• network_monitor_latency: Histogram of round trip times
• network_monitor_packets_sent: Counter of sent packets
• network_monitor_packets_received: Counter of received packets
• network_monitor_packets_lost: Counter of lost packets

Labels

• source: Node name of source
• target: Node name of target

Query

histogram_quantile(0.5, sum(rate(network_monitor_latency_bucket[5m])) by (le, source, target))