Guides

Practical guides to monitoring your sites, servers, homelab, and the monitoring stack itself.

• AI agent uptime monitoring

  Monitoring built so an agent can operate it, not just read it: an MCP server for conversational agents, a versioned REST API for scripts, one set of scoped tokens.

• A dead man's switch for cron jobs

  Get alerted when a scheduled job does not run. Working recipes for cron, shell scripts, systemd timers, and GitHub Actions, plus how to pick a grace period.

• External monitoring for self-hosted services

  A decision framework for watching self-hosted services from outside: active checks when exposed, heartbeats when not, authenticated checks for everything gated.

• Monitor Docker containers externally

  Per container heartbeats from cron and HEALTHCHECK, without exposing the Docker socket. Detect the container that died and stayed dead, from off site.

• Monitor Home Assistant externally

  Watch Home Assistant from outside your LAN: the real /api/ health endpoint, token headers, and a cron heartbeat for instances you do not expose.

• Monitor Jellyfin externally with /health

  Jellyfin ships a real /health endpoint on port 8096. Here is how to check it from outside your LAN, plus a heartbeat recipe for private installs.

• Monitor Pi-hole externally the safe way

  Pi-hole v6 replaced the old API. Here is the current /api auth flow, why you must not expose port 53, and a heartbeat gated on a real DNS lookup.

• Monitor Plex from outside your network

  Plex has no health endpoint. Here is what to check instead: the unauthenticated /identity response on port 32400, token headers, and a cron heartbeat.

• Monitor Proxmox VE from outside

  Check Proxmox from off site: the /api2/json/version endpoint on port 8006, PVEAPIToken headers, PVEAuditor scope, and heartbeats for private clusters.

• Monitor a Synology NAS externally

  Watch DSM from off site: the unauthenticated SYNO.API.Info query, ports 5000 and 5001, and a heartbeat from Task Scheduler for NAS boxes you keep private.

• Monitor TrueNAS from outside your LAN

  Check TrueNAS off site: the unauthenticated core/ping route, Bearer API keys on /api/v2.0, the REST deprecation, and heartbeats tied to pool health.

• How to monitor Uptime Kuma from outside

  Two working recipes for watching Uptime Kuma from outside your network: an active check against its status page, and heartbeat pings that need no open ports.

• Uptime monitoring MCP server

  Connect Claude or any MCP client to Undownable and let it create monitors, read incidents, and schedule maintenance windows with a scoped API token.

• Who monitors the monitor?

  Your self-hosted monitoring cannot tell you that it is dead. The structural blind spot, the usual DIY fixes and what they cost, and the backstop pattern.