Watchdog

fleet watchdog

Check all registered apps and the docker-databases service for health issues. If any failures are found, send an alert via the configured notification adapters (Telegram).

Designed to be run on a cron schedule (e.g. every 5 minutes). Exits with code 1 if any services are unhealthy.

Usage

fleet watchdog [--motd]

Flags

Flag	Description
`--motd`	Display failures only, do not send alerts. Always exits 0. Useful for SSH login banners.

Examples

All healthy:

$ fleet watchdog
✓ All 4 services healthy

With failures:

$ fleet watchdog
! 2 service(s) unhealthy
✗   worker: down (systemd: failed)
✗   api: degraded (containers down: api-web-1; http check failed)
✓ Alert sent

With --motd (no alerts):

$ fleet watchdog --motd
! 1 service(s) unhealthy
✗   worker: down (systemd: failed)

What watchdog checks

docker-databases systemd unit status
For each registered app:
- systemd unit state (must be active)
- Container running status
- HTTP health endpoint (if healthPath is set in the registry)

Notification configuration

Watchdog reads notify config from /etc/fleet/notify.json. It supports Telegram and BlueBubbles iMessage via an adapters array:

{
  "adapters": [
    {
      "type": "telegram",
      "botToken": "123456:ABC-DEF...",
      "chatId": "-100..."
    },
    {
      "type": "bluebubbles",
      "serverUrl": "https://your-bluebubbles-server",
      "password": "your-password",
      "chatGuid": "iMessage;-;+15551234567",
      "cfAccessClientId": "...",
      "cfAccessClientSecret": "..."
    }
  ]
}

cfAccessClientId and cfAccessClientSecret are optional — only needed if the BlueBubbles server sits behind Cloudflare Access.

Running on a cron schedule

*/5 * * * * root /usr/local/bin/fleet watchdog

Running as an MOTD script

#!/bin/bash
/usr/local/bin/fleet watchdog --motd