← All guides

Uptime Kuma without the server: self-hosted vs hosted uptime monitoring

Updated 26 August 2026 · by Cassian Wei · 3 min read

Every time a monitoring service raises prices, a wave of people discover Uptime Kuma — the self-hosted, open-source uptime monitor — and it deserves the enthusiasm. This guide is the conversation to have before you docker run: what running your own monitoring actually costs, the structural blind spot it creates, and what to do about it whichever way you go. (Spoiler: even if you keep Kuma, you should add one external check.)

What self-hosting your monitoring really costs

  • A host. A small VPS (~$4–6/month) or a corner of an existing box. If it's an existing box, note which one — it matters below.
  • Maintenance. Kuma ships updates regularly; you're now on the hook for image upgrades, Node/OS patches, disk space for its history database, and the reverse proxy + TLS in front of it if you want it reachable and not world-writable.
  • Backups. Kuma's monitor configs, notification setups and history live in a SQLite file. If the host dies un-backed-up, you rebuild every check from memory.
  • Deliverability. Email alerts from a residential IP or fresh VPS land in spam surprisingly often. (Most people wire Kuma to Telegram/ntfy instead — that part works great.)

None of this is hard for someone who already runs services. It's just real, recurring work attached to the one tool whose entire job is to be reliable when everything else isn't.

The blind spot: who monitors the monitor?

The failure modes that hurt most are the correlated ones:

  • Kuma runs on the same box (or same provider, or behind the same router) as the services it watches. The box dies, the provider has an outage, the home connection drops — your monitoring dies at the same instant as your services and nothing alerts, because the thing that alerts is gone too.
  • Kuma itself breaks quietly — a failed container restart after reboot, a full disk, a crashed update. Everything looks fine, because "no alerts" and "no monitoring" are indistinguishable from the inside.

The principle (vendor-neutral, applies to any self-hosted monitor): at least one check must run from infrastructure you don't operate.

If you keep Kuma: add one external check (do this today)

Point a free hosted monitor at your Kuma instance itself, from the outside. Two minutes with Watchpup (free, no card):

curl -X POST https://watchpup.watchpup.workers.dev/api/signup \
  -H 'content-type: application/json' \
  -d '{"email":"you@example.com","password":"at-least-8-chars"}'
# → {"ok":true,"api_key":"wp_..."}

curl -X POST https://watchpup.watchpup.workers.dev/api/monitors \
  -H 'authorization: Bearer wp_...' -H 'content-type: application/json' \
  -d '{"name":"uptime-kuma itself","url":"https://status.example.com","interval":60}'

Now "Kuma is down" is an alert instead of a silence. Two upgrades worth making:

  • Mirror your 2–3 most critical checks externally too — the ones where you'd want to know even if your whole rack vanished.
  • Heartbeat the box itself: a cron line on the host pinging a dead-man's-switch monitor catches "the machine is gone" even for hosts with no public HTTP service.

If you'd rather not run a server at all

The hosted trade: someone else carries the ops, and the checks come from outside your infrastructure by construction. Watchpup's free tier is built to be enough for real use, not a teaser: 15 monitors at 60-second intervals — HTTP(S) with keyword/status assertions, TCP ports, TLS certificate expiry, domain expiry, and cron-job heartbeats — plus public status pages, and alerts by email, signed webhook, Slack, Discord, Teams, Google Chat, ntfy or Telegram. Everything is driveable by API, CLI and MCP, and there's a live demo status page to judge it by.

Where Uptime Kuma wins — honestly

  • Unlimited monitors. Hundreds of checks cost you nothing but disk. Watchpup free caps at 15.
  • Internal networks. Kuma can watch 192.168.x.x, LAN NAS boxes, and services never exposed to the internet. A hosted monitor can't reach those — Watchpup deliberately refuses private and internal addresses.
  • Notification breadth. Kuma speaks to ~90 notification providers; Watchpup covers the common eight.
  • Data ownership. Your history sits in your SQLite file, forever, under your rules.
  • No third-party risk. Kuma can't shut down, get acquired, or change its pricing.

If those matter to you, self-host — genuinely. Just don't let the monitor be the one service nobody is watching. The external check is free either way; there's a wider free-tier comparison if you're weighing hosted options against each other.

Watchpup is free uptime & cron monitoring — 1-minute checks, heartbeats, TLS/domain expiry, status pages, alerts everywhere. Sign up, try the live demo, or read more guides.