shekse
esc
    ← home

    post · Jun 7, 2026 · 4 min read

    Homelab Alerts on Your Phone: Integrating Beszel and Uptime Kuma with ntfy

    Wire up Beszel and Uptime Kuma to send push notifications through ntfy. Configure Shoutrrr URLs for Beszel, native ntfy integration for Uptime Kuma, set meaningful alert thresholds, and harden everything with access tokens instead of passwords.


    This is part two of the homelab notifications series. In part one we set up ntfy on LXC 158 and apprise on LXC 159. Now we wire up the two monitoring tools already running on the homelab — Beszel and Uptime Kuma — so alerts land on your phone the moment something goes wrong.

    Both tools connect directly to ntfy. No apprise in the middle — they both speak ntfy natively, so there’s no reason to add an extra hop.

    Beszel ──────────────────────→ ntfy → phone
    Uptime Kuma ─────────────────→ ntfy → phone
    Sonarr / Radarr / scripts ───→ apprise → ntfy → phone

    apprise earns its place with services that don’t have native ntfy support. For monitoring tools that do, go direct.


    System Reference

    PropertyValue
    ntfy LXC IP192.168.0.158
    ntfy alert topichomelab-alerts
    Beszel LXC IP192.168.0.157
    Uptime Kuma LXC IP192.168.0.154

    Part 1 — Beszel

    Beszel uses Shoutrrr under the hood for webhook notifications. Shoutrrr has native ntfy support, so the integration is a single URL.

    Add the notification URL

    Go to Beszel → Settings → Notifications → + Add URL and enter:

    ntfy://publisher:<publisher-password>@192.168.0.158/homelab-alerts?scheme=http

    Why ?scheme=http: Shoutrrr’s ntfy scheme defaults to HTTPS. Since ntfy is on an internal LAN IP without TLS, you must explicitly pass ?scheme=http — otherwise Shoutrrr tries to connect on port 443 and fails.

    Hit Test — you should receive a notification on your phone. Save Settings.

    Set alert thresholds

    Click the bell icon next to your Proxmox host in the systems table. Enable these alerts:

    AlertThresholdDuration
    Status—2 minutes down
    CPU Usage90%10 minutes
    Memory Usage85%10 minutes
    Disk Usage85%10 minutes
    Temperature80°C10 minutes
    Load Average 5m610 minutes

    Why these thresholds:

    • Status at 2 minutes — don’t wait 10 minutes to know your server is down
    • Load Average 5m at 6 — the i7-8700T has 6 cores; sustained load above core count is a red flag
    • Temperature at 80°C — the 8700T in a mini PC enclosure can get warm under sustained load; 80°C gives you warning before thermal throttling kicks in
    • Skip GPU Usage — iGPU passthrough to Immich and Jellyfin will spike this during transcoding constantly, too noisy
    • Skip Bandwidth — not critical for a homelab


    Part 2 — Uptime Kuma

    Uptime Kuma has ntfy as a built-in notification type — no Shoutrrr URL needed.

    Go to Settings → Notifications → Add Notification:

    • Notification Type: ntfy
    • Name: ntfy-homelab-alerts
    • Server URL: http://192.168.0.158
    • Topic: homelab-alerts
    • Username: publisher
    • Password: your publisher password
    • Priority: High

    Hit Test, confirm the notification lands, then Save.

    Then assign this notification to your monitors — open each monitor, scroll to Notifications and toggle on ntfy-homelab-alerts.


    Part 3 — Hardening with Access Tokens

    Using passwords in service URLs is fine for a homelab, but access tokens are better practice — each service gets its own token, and if one is ever compromised you revoke just that token without affecting anything else.

    Generate tokens

    Jump into LXC 158:

    pct exec 158 -- bash

    Create a dedicated token for each service:

    ntfy token add publisher --label beszel
    ntfy token add publisher --label uptime-kuma

    List to verify:

    ntfy token list

    Expected output:

    user publisher
    - tk_owy8d8x8p5ddx7gwdwi7g72n4yypi, never expires, label: beszel
    - tk_irmbr8hvgitzipx5ec2byc01ormzj, never expires, label: uptime-kuma

    Update Beszel

    Go to Settings → Notifications, remove the old URL and replace with the token format. Per the Beszel ntfy docs, the token format uses a colon directly before the token with no username:

    ntfy://:tk_owy8d8x8p5ddx7gwdwi7g72n4yypi@192.168.0.158/homelab-alerts?scheme=http

    Test and save.

    Update Uptime Kuma

    Go to Settings → Notifications, edit the ntfy notification. Clear the username and password fields and put the token in the Token field:

    • Token: tk_irmbr8hvgitzipx5ec2byc01ormzj
    • Username: (clear this)
    • Password: (clear this)

    Test and save.

    Revoke unused tokens

    If you created any test tokens during setup, clean them up:

    ntfy token list
    ntfy token remove publisher <token>

    Keep only the labelled tokens — one per service, nothing anonymous.


    What You End Up With

    ServiceIntegrationToken
    BeszelShoutrrr ntfy URLbeszel token
    Uptime KumaNative ntfyuptime-kuma token

    Both monitoring tools now send alerts directly to ntfy on homelab-alerts. Beszel fires on CPU, memory, disk, temperature, load average, and status changes. Uptime Kuma fires when any monitor goes down or recovers.


    What’s Next

    Media stack notifications

    Sonarr, Radarr, Jellyfin, and Seerr don’t have native ntfy support — they’ll route through apprise on LXC 159 to a separate homelab-media topic. That keeps download completions and library updates out of the alerts feed and lets you set different notification priorities per topic on your phone.