For monitoring purposes it's nice to have a Prometheus exporter endpoint (/metrics) to expose things like daemon health and also wg counters, babel states etc.
Since the dynamc nature of the config, you also can implement Prometheus HTTP Service Discovery, so instead of hardcoding addresses, the monitoring stack could automatically find and scrape your nodes (i've done that for Nebula).
For the container setup it's also nice to have /healthz endpoint (and possibly /readyz).
I'm running mesh of P2P AWG links + BGP+BFD (FRR) on top of OpenWrt, orchestrated by Ansible.
I found /127 464xlat quite useful for that p2p links because it allows me to choose easily decodable addresses like (prefix::<node_a_id>:<node_b_id>:<0/1>/127).
For monitoring purposes it's nice to have a Prometheus exporter endpoint (
/metrics) to expose things like daemon health and also wg counters, babel states etc.Since the dynamc nature of the config, you also can implement Prometheus HTTP Service Discovery, so instead of hardcoding addresses, the monitoring stack could automatically find and scrape your nodes (i've done that for Nebula).
For the container setup it's also nice to have
/healthzendpoint (and possibly/readyz).I'm running mesh of P2P AWG links + BGP+BFD (FRR) on top of OpenWrt, orchestrated by Ansible.
I found /127 464xlat quite useful for that p2p links because it allows me to choose easily decodable addresses like (
prefix::<node_a_id>:<node_b_id>:<0/1>/127).