Start here
Quickstart — account to first alert in about five minutes.
Concepts — monitors, checks, thresholds, incidents, and why a failing
monitor sits in pending before it goes down. Ten minutes here saves an hour of confusion
later.
Monitor types
Eleven types is enough to be worth a decision column. The middle column is the question the type answers; the third is when to reach for it rather than something adjacent.
| Type | Answers | Reach for it when | Reference |
|---|---|---|---|
| HTTP | Does this URL respond correctly? | Anything served over HTTP. Start here | docs |
| TCP | Does this port accept a connection? | A port with no protocol you can assert on — a broker, a game server | docs |
| Heartbeat | Did my cron job run? | The thing has no address to poll. Backups, cron, CI | docs |
| Certificate | How long until my TLS certificate expires? | Anything on 443. Renewal should be routine, not an incident | docs |
| Domain | How long until my registration expires? | Every domain you own. Nothing else predicts this failure | docs |
| DNS | Does this record still say what it should? | A record you would notice being wrong only after users did | docs |
| Email auth | Are SPF and DMARC still published? | You send mail — including your own alert emails | docs |
| ICMP | Is this host reachable, and how lossy is the path? | You want latency history, or the host serves nothing to poll | docs |
| UDP | Does this DNS, NTP or STUN service answer? | A resolver, time server or STUN/TURN endpoint | docs |
| gRPC | Does the health service report SERVING? | gRPC, where a bound port is not a ready service | docs |
| SMTP / IMAP | Does the mail server greet, and does STARTTLS work? | A mail server — and its certificate, which is not on 443 | docs |
Two common mistakes. A TCP monitor on a mail server or a gRPC port reports up while the service is broken — use the specific type. And an ICMP monitor is not a substitute for an HTTP one: a machine with a dead application answers pings perfectly.
HTTP, TCP, ICMP, UDP, gRPC, SMTP and IMAP run at your chosen interval, from 20 seconds. Certificate, domain, DNS and email auth default to once a day, because certificates and registrations change on a scale of months.
Alerting
Channels — email, webhook, ntfy and Telegram, with setup for each. Includes the Telegram trap that makes a working bot token look broken.
Reminders while still down — repeat the alert while an incident stays open. Off by default.
External second opinion — how WhatPing decides between “the target is down” and “we cannot reach the target”.
Reference
API reference — provision monitors and read state from Terraform, CI or your own dashboard. Bearer key auth, cursor pagination, idempotent creates.
Webhook payload — the full JSON schema, with a real example.
Heartbeat ping endpoint — URL format, methods, responses, and snippets for cron, systemd and GitHub Actions.
Limits and defaults — every bound in one table.
Security model — what is validated, what is redacted, what is stored.
Troubleshooting — the questions people actually ask.
Conventions in these docs
- Every number stated is the value enforced by the product. If a document and the interface disagree, the interface is right and the document is a bug — please report it.
- Alert text shown in examples is the literal text WhatPing sends, emoji included.
- Anything not built is not documented. See the roadmap for what is planned.