New in FlowAlert: Uptime Checks — We Now Watch Your Site from the Outside
Every site in your dashboard can now be probed from the outside: three failed checks page whoever's on call, recovery closes the loop, and WordPress check-ins got a lot smarter about quiet nights.
New in FlowAlert: Uptime Checks — We Now Watch Your Site from the Outside
Until now, FlowAlert monitored your site the way a doctor reads a pulse: your site checked in with us every ten minutes, and silence meant trouble. That dead-man's switch catches a whole class of failures other monitors miss — broken cron, dying PHP, a site that's "up" but internally wedged.
Today we're adding the other direction. Uptime checks are live for every site in your dashboard: FlowAlert visits your site from the outside, every few minutes, and pages the right person when it stops answering.
What shipped
- One toggle per site. Open Dashboard → Sites, flip on Uptime check. No plugin required, no code, works for any site — WordPress or not.
- Three strikes, then a page. A single failed probe is noise; three consecutive failures is an outage. You get a
site.downalert with the reason (timeout, connection refused, HTTP status), routed through your rules like any other event — push, acknowledgement, escalation and all. - Recovery closes the loop. The first successful probe after an outage sends
site.recovered, with the response time, so the all-clear reaches the same phones the alarm did. - False-positive protection. When most probes in a sweep fail at once, that's our vantage point having a bad minute, not sixty customer sites dying simultaneously — the checker recognizes this and holds fire.
Two directions, one picture
The two signals are complementary, and the difference is exactly the information you want at 3 a.m.:
| Uptime check | Site check-in | |
|---|---|---|
| Who initiates | We visit your site | Your site reports to us |
| Catches | Site unreachable: host down, DNS, SSL, crash | Internal silence: WP-Cron starved, PHP broken, backup never ran |
| Needs the plugin | No | The WordPress plugin (or any cron that can ping a URL) |
| Reads as | "Customers can't reach you" | "The machinery inside stopped" |
Down + not checking in: real outage. Up but not checking in: the site serves pages but WordPress is broken behind a cache — a failure most uptime tools happily sleep through.
Check-ins got smarter too: verify-on-miss
The check-in side learned a trick from the new prober. WordPress check-ins ride WP-Cron, and WP-Cron only runs when the site gets traffic — so on a quiet night, a healthy-but-sleepy site used to look exactly like a dead one.
Now, when a site-linked check-in goes overdue, FlowAlert probes the site first. If it answers, we wait, quietly re-checking, and only page you if the silence outlasts a generous tolerance — worded honestly: site serves HTTP, WordPress hasn't checked in. If the probe fails too, you're paged immediately with both facts. Quiet nights stopped false-alarming; real failures page faster.
Alerts that refuse to die quietly
Like check-in alerts, site.down is never swallowed by an exhausted monthly quota. A dead-man's signal that goes quiet exactly when nobody is watching would be worse than none at all — so these alerts bypass the cap, on every plan.
Turn it on in two minutes
- Dashboard → Sites — every site you've connected is already listed (events auto-register them).
- Flip Uptime check on the sites that matter.
- Make sure something routes
site.*— the dashboard will offer to pre-fill the routing rule for you, with acknowledgement on, so a down site always has an owner.
That's it. From today, FlowAlert watches from both sides of the wall: your site tells us when something inside breaks, and we tell you when the outside world can't get in. Details and the full behavior live in the monitoring docs.
