Under the hood

The technical detail, for whoever looks after the technical side.

Everywhere else we keep things in everyday words. This page doesn't: it's for IT staff, agencies and developers who want to know exactly what Pingerer checks, how it decides something is down, and what you can build on top of it.

How each check works

HTTP(S)

GET by default (any method, custom headers). Passes on status 200–399 unless you set your own ranges, with optional keyword present/absent matching on the body. Follows redirects unless told not to. Each response is timed in phases: DNS, TCP connect, TLS handshake and time to first byte. 30-second timeout by default; slowness alerts from a response-time threshold sustained over several checks or an average.

TCP, ICMP and SMTP

TCP connects to any public port. Ping sends ICMP echo. SMTP connects to your mail server, reads the banner and can negotiate STARTTLS or implicit TLS, optionally verifying the certificate. Same intervals, timeouts and confirmation as HTTP.

Multi-step (synthetic)

A sequence of HTTP requests (GET, POST, PUT, PATCH, DELETE; form or JSON bodies) with assertions on status, headers and JSON fields, and values extracted from one step into the next (a login token, an ID). Secret variables are encrypted at rest and never shown back.

TLS certificates

Every 6 hours: full chain validation against public roots, hostname match (including SANs and wildcards), and expiry. Warnings at 30, 14 and 3 days by default, configurable per check.

Domain registration

Daily, over RDAP, found through IANA's bootstrap registry: expiry date, registrar and EPP statuses (transfer lock). For registries without RDAP or that redact the expiry (most GCC ccTLDs: .kw, .sa, .ae, .qa, .om; also .io, .co), you enter the renewal date and we warn from that.

DNS changes

Every 15 minutes: A, AAAA, CNAME, MX, NS and TXT (you choose which) compared against the last known answer, so an unexpected nameserver, MX or record change alerts you.

Email authentication

Daily: SPF, DKIM at your selectors (common ones are checked by default), and the DMARC policy. Problems that existed when monitoring started are shown as advice rather than alerting; anything that gets worse alerts.

Heartbeats

For cron jobs, backups and workers: each gets a unique URL (GET or POST) and an email address to ping. Miss the expected interval plus grace period and it alerts. Add /fail to report a failed run straight away.

"It's your host" detection

Each site is fingerprinted (hosting network by ASN/AS org, CDN, platform). When several unrelated, established customers on the same verified network go down within the window, alerts say it's probably the provider. Header-only claims aren't trusted unless the network backs them up. Public summary: Internet weather.

How we decide something is down

Uptime checks rotate through the locations you pick, one per interval. A failure is never alerted on its own: the check is immediately re-run from other locations, and an incident opens only when at least two locations agree (configurable per check, up to all of them). A failure caused by our side, such as a resolver or registry timeout, is marked inconclusive and never opens an incident.

Once open, an incident sends one alert per channel, escalates if nobody acknowledges it, and resolves when checks pass again. Maintenance windows keep checks running but suppress alerts, and the time doesn't count against uptime. Quiet hours hold non-urgent alerts (TLS, domain, DNS, email, slowness); downtime and missed heartbeats always go through.

We refuse to check private, loopback, link-local and other non-public addresses, on the first request and after every redirect, so Pingerer can't be pointed at anyone's internal network.

When Pingerer itself is down

The two things customers' own users and jobs depend on don't depend on our main server being up. Both sit behind Cloudflare Workers that pass every request straight through and only step in when the app can't answer.

Status pages. The worker keeps the last good copy of every page, its history, RSS and Atom feeds, JSON summary, badges and stylesheet in a Durable Object, refreshed from live traffic at most every 2 minutes and for every published page every 10 minutes. If the app doesn't answer within 8 seconds, visitors get the copy with a dated "live updates are paused" note. Custom domains are covered too. Subscribe, confirm and unsubscribe requests made meanwhile go on a Cloudflare Queue and are delivered once the app is back.

Heartbeat pings. If the app doesn't answer within 6 seconds, the worker answers the job itself and queues the ping with the time it arrived. Queued pings are delivered in batches once the app is back (kept for up to a day) and count at the time they were sent; each carries an id, so a ping that was also answered counts once. The sweep that marks heartbeats missed skips any heartbeat with pings still waiting, so an outage on our side doesn't turn into false alerts.

Where checks come from

To allowlist us in a firewall or WAF, allow these addresses. Requests identify as Pingerer/1.0 (+https://pingerer.com/bot).

LocationCodeIPv4IPv6
Falkenstein fsn 46.224.96.215 2a01:4f8:c016:908a::1
New York iad 165.227.209.205 2604:a880:800:14:0:3:b978:d000
Singapore sgp 167.71.213.96 2400:6180:0:d2:0:3:442f:b000
Toronto yyz 134.122.34.63 2604:a880:cad:d0:0:1:c251:7001

More on IP addresses and allowlisting →

Build with Pingerer

Everything in the dashboard is also available from code.

Available

REST API

JSON over HTTPS with bearer tokens, read-only or read-write, optionally limited to one account. Sites, checks, results, incidents, heartbeats, alert channels and rules, status pages and their updates. OpenAPI 3 spec included.

curl https://pingerer.com/api/v1/sites \
  -H "Authorization: Bearer $PINGERER_TOKEN"

API reference · openapi.yaml

Available

Signed webhooks

Every alert event (opened, confirmed, escalated, update, recovered) as a JSON POST, signed with HMAC-SHA256 over the timestamp and body so you can verify it came from us.

X-Pingerer-Event: opened
X-Pingerer-Signature: t=1760000000,v1=<hex>

Webhook details

Available

Heartbeat URLs

Ping from any script, scheduler or CI job. No token needed: the URL is the secret.

0 3 * * * /usr/local/bin/backup.sh && \
  curl -fsS https://pingerer.com/hb/YOUR-TOKEN

Heartbeats

Available

Status pages and badges

Public status pages on your own domain, with SVG badges for live status, uptime and "Monitored by Pingerer" to embed anywhere.

Status pages

Available

Command-line tool and GitHub Action

Check status, add sites and checks, acknowledge and resolve incidents from your terminal or CI; pause alerts or open a maintenance window around a deploy and smoke-test afterwards.

Command-line tool

Available

MCP server for AI assistants

A remote MCP server with one-click OAuth sign-in for Claude, ChatGPT, Cursor and Claude Code. Every tool goes through the REST API, so roles, plan limits and Undo apply, and nothing deletes.

claude mcp add --transport http pingerer https://pingerer.com/mcp

AI assistants

Coming soon

Terraform

Manage sites, checks and alerting as code.

Compare: Pingerer vs UptimeRobot · Pingerer vs Better Stack · Pingerer vs Pingdom

Add your website. We'll do the watching.

It takes about a minute, it's free for one domain, and we don't ask for a card.

Start free Read the help center