Documentation
¶
Overview ¶
Command wg-monitor checks the platform and files what it finds.
One pass per invocation, driven by a systemd timer rather than an internal loop. A timer is externally visible (systemctl list-timers says when it last ran and when it runs next), it restarts after a crash without supervision logic here, and a missed run is a fact the operating system records rather than something this process would have to notice about itself.
The blind spot is stated rather than hidden: this runs ON the node and reports THROUGH the database. If the whole node is down, or the database is unreachable, it cannot deliver an alert about that — the case where alerting is most needed is the one an in-band monitor cannot cover. It writes those failures to the journal at a distinguishable priority so a log-based external alarm can find them, and closing the gap properly needs an external prober (tracked separately).
Source Files