Guide

Watching a website

A monitor loads the address your users type, from our points of presence, on a schedule — so a broken site reaches you before it reaches a customer.

Step by step

1. Go to Website monitoring and press + New monitor.
2. Name and URL: the public address, the one users type. Not the origin, not an internal name.
3. Interval is how often each point of presence runs it. For is how long a verdict has to hold before it counts. Confirmations is how many points of presence have to agree — two by default, and two is the right answer.
4. Points of presence: leave them all unticked and we choose. Pick them when you have a reason, such as customers in one region.
5. Behind Cloudflare, Akamai or AWS WAF? Open Behind a WAF? before you save. It mints a secret and writes out the rule you paste into your WAF, so it lets our probes through and changes nothing else.
6. Press Create monitor. The first result arrives within one interval, and the monitor’s own page shows what each point of presence saw.

Why it is worth more than a ping

A monitor that only knows up or down tells you the site is broken and nothing else. This one compares what the CDN served with what your origin serves and answers in words: the CDN is at fault, the origin is down and the cache is still covering for it, everything is out, or the target blocked us.

The second of those is the one worth having. It is critical and nobody has noticed yet, which is the only window in which a fix costs nothing.

That comparison is off until you turn it on, and turning it on means our probes talk to your origin directly — which is exactly the protection you pay a CDN for. It is a real decision, not a default, and the form says so where you make it.

The one thing people get wrong

Where to go next

Then the wait that keeps a rule honest, and how to put a monitor on a page your customers can read.

• A rule, and its wait
• The status page
• All the guides