Docs / MonitoringBrowse

Monitors and check types

A monitor is one thing Heedery checks: a website, an API, a server, a port or a scheduled job. Each one has a check type, an interval and its own alert settings.

On this page
  1. Pick a check type
  2. Websites and keywords
  3. APIs
  4. Browser flows
  5. Ping, ports, DNS and mail servers
  6. Scheduled jobs (heartbeats)
  7. How often checks run
  8. Regions
  9. Alerts, categories and pausing

Pick a check type

Press Add monitor. A website needs only its address. For anything else, press Something else? and pick what you want to watch:

Website
A site or online shop people open in a browser.
APIHTTP
An address your app calls, such as /health. Answers in JSON.
Scheduled jobHeartbeat
A backup or cron job that should run on time. It pings us; silence means trouble.
Browser flowChromium
A robot signs in or goes through checkout the way a visitor would.
Server is onlinePing
Any server or router with an address: we check that it answers.
Server portTCP
A database, SSH or game server: we check that its port lets us in.
Domain recordsDNS
Your domain still points where it should and nobody changed it.
Mail serverSMTP, IMAP, POP3
The server that sends or receives your mail is answering.
UDP port
Game servers, time servers, syslog: we send a packet and wait for a reply.
The list of monitor types: website, API, scheduled job and more
Add monitor → Something else?

You can change the type later: open the site, then Settings, then Change address.

Websites and keywords

A website check requests the page and times each part: the name lookup, the connection, TLS, the first byte and the transfer. It counts as down when:

  • nothing answers within 10 seconds;
  • the status code is not one you expect. By default any 2xx or 3xx passes. Set your own list under Expected status codes, such as 200-299,404;
  • a keyword rule fails.

To check that the page shows the right thing, open Also make sure the page shows certain text and choose Must contain or Must not contain. The match ignores case and reads the first 1 MB of the HTML. Redirects are followed for keyword checks.

APIs

An API check sends the request you describe and reads the reply. If you add a site whose address answers JSON, Heedery switches it to an API for you.

Method
GET, HEAD, POST, PUT, PATCH or DELETE
Headers
Up to 20, such as an Authorization header.
Body
Any text, for POST, PUT and PATCH.
JSON value
A path into the reply and the value it must have.
Down when slower than
A limit in milliseconds. New APIs start at 2000 ms; websites have none.

Browser flows

Plan/Pro and up. Pro includes 3

A browser flow runs your scenario in a real Chromium browser: sign in, add to cart, check out. Start from a Start page, then add steps with Add a step…: Go to URL, Click, Fill field, Press key, Wait for, Expect text or Expect URL.

  • Up to 20 steps.
  • One step may take 15 seconds, the whole run 60.
  • It runs every 5 minutes at most.
  • Tick Secret on a password field. It is stored encrypted, never shown again and masked in screenshots.

When a step fails, the run keeps the step, its error and a screenshot of the page.

Ping, ports, DNS and mail servers

Server is online (Ping)
Sends 3 pings. Under Up with, pick any reply, all 3 replies, or at most 1 lost.
Server port (TCP)
Up when a connection to the host and port opens within 10 seconds.
UDP port
Sends one packet, as text or hex, up to 1024 bytes. With Reply contains set, it waits for that reply. Without it, only “port unreachable” counts as down.
Domain records (DNS)
Looks up the A, AAAA, CNAME, MX or TXT records of a hostname. Add Contains to require a value.
Mail server
SMTP, IMAP or POP3, with STARTTLS, TLS or neither. It waits for the greeting and never logs in.

Scheduled jobs (heartbeats)

A heartbeat works the other way round: your job calls us. Use it for backups, cron jobs, queues and anything inside a private network.

  1. 1.
    Press Add monitor, pick Scheduled job, and set Expect a ping to how often the job runs.
  2. 2.
    Set Grace: how late a ping may be before we call it missed (up to 24 hours).
  3. 3.
    Copy Your ping URL and call it at the end of the job.
At the end of your script
curl -fsS -m 10 --retry 3 https://heedery.com/api/heartbeat/<your-token>

GET, POST and HEAD all work. When no ping arrives for the period plus the grace, an incident opens. To get a new URL, open the site’s Settings and press New URL.

How often checks run

Choose the interval per monitor. Your plan sets the shortest one:

PlanCheck every
Starter
10, 15, 30 or 60 minutes
Pro
1, 5, 10, 15, 30 or 60 minutes
Team
1, 5, 10, 15, 30 or 60 minutes
Agency
1, 5, 10, 15, 30 or 60 minutes
Enterprise
1, 5, 10, 15, 30 or 60 minutes

A site that fails is checked again after 1 minute, whatever its interval. 2 failed checks in a row open an incident; 2 good ones close it. See Incidents.

Regions

On every plan, while more than one check region runs, an outage is confirmed before anyone is alerted. Every other region checks the site at once. The incident opens only if one of them sees the failure too. A region that does not answer within 20 seconds never holds an alert back.

The map on the site page shows each region and where the site’s server is. Behind Cloudflare or another CDN, the arcs end at the CDN point that answered each region most often over the last 24 hours, read from the response headers. The CDN picks the point for each request, so it can change from one check to the next; the lines under the map give each point’s share, such as FRA 71%, IAD 29%. To show where the site really runs, fill in Origin server in the site’s settings.

Plan/Team and up. Response time from each region

With regional checks, each region also measures the response time every 5 minutes. The site’s Performance tab shows it on a map and in a By region table. Pick the regions under Check regions in the site’s settings.

If a region is unavailable for more than 15 minutes, a notice on the Overview and on the site page says so. Your sites are still checked from the main region; until the region is back, it measures no response times and takes no part in confirming outages.

NoteWhich regions exist depends on where checkers run today. How we check lists them live.

Alerts, categories and pausing

Open a site and go to Settings. Changes wait until you press Save changes.

When to alert
Open the incident at once or after 1 to 30 minutes down. Short blips then never alert.
Email alerts
On or off for this site. Also send to adds up to 10 more addresses, people outside the workspace too.
Chat channels, escalation
Which channels and which policy this site uses. See Alerts and channels.
Monitoring on
Off pauses checks and alerts. The history stays. An open incident closes at the pause, marked Monitoring paused, with no recovery alert (status page subscribers told of the outage get a short email that monitoring is paused), and paused time is not downtime; status pages show the site as paused. The Pause button on the site page does the same.
Organise
A category and the status page the site appears on. A new site is on no page until you pick one.

Categories group sites in lists, filters and status pages. Create them on Sites → Categories. To move many sites into one, open the category and press Add sites.

Check now on the site page runs a check at once and reads the certificate and domain dates again.

Still stuck? Write to us.