What changed before it broke
- DNS: an A record changed
- Region A: no answer
- Both regions fail: down
Demo data, made up. The change is marked: it came 6 minutes before the incident.
A site is down only when two regions agree.
One failed check from one place is often the network between you and the site. Beamfar asks a second region before it wakes anyone, and keeps what both saw.
- Every site checked every minute, or every 30 seconds from Team up (Free: every 5 minutes)
- From two regions
- Nothing to install
Two regions, four outcomes
the rule
| Region A | Region B | What Beamfar says |
|---|---|---|
| Passed | Passed | Steady |
| Failed | Passed | Regional issue, no alert |
| Failed | Failed | Down: one alert |
| No reading | No reading | Data unavailable |
The two-region rule, step by step
Every check runs from every enabled region. A site is down when two regions both fail to get a valid answer. One region failing is a regional issue: it is recorded, shown, and sends no alert.
- Step 1
Each region checks the site
Every minute (every 30 seconds on Team and up when you choose it, every 5 minutes on the Free plan), each region requests the page, reads the certificate, and looks up DNS. Nothing is installed on the client site.
- Step 2
A failure in one region is a regional issue
It appears on the timeline and in the report, with what that region saw. No alert is sent, because the fault may be the route to the site and not the site.
- Step 3
Two regions agree: the site is down
The incident opens and the alert goes out by email, Slack, Teams, Discord, PagerDuty or another channel you added.
- Step 4
The alert carries the evidence
It lists what each region saw (result and time) and links to the incident page with the full evidence: status code, redirects, certificate, DNS answer and timings. A check that gave no valid measurement says data unavailable. It is never counted as up and never shown as zero.
- Step 5
The timeline shows what changed just before
DNS records, the certificate, the final redirect address and the SEO signals are recorded between checks. A change that came within the hour before an incident is marked next to it.
What you see when it happens
The incident page of the app, on made-up demo data: both regions failed, so the incident is open, with what each one saw and why it most likely happened.
- Both regions saw it. An incident opens when both regions saw the failure, and keeps what each one saw.
- Why this happened. The likely cause with its confidence, the evidence behind it and who has to act.
- Nothing is guessed. A check without a valid measurement says data unavailable, never up.
One region against two
Thirty made-up checks per region. Switch the rule and count the alerts: with one region every blip wakes you, with two only the real outage does.
Demo data
Two regions agree: 1 alert, for the real outage.
- Steady
- Regional issueno alert
- Downtwo regions agree
- Data unavailablenever counted as up
The evidence you can forward
After an outage the client asks what happened and the host asks for proof. The incident page holds both regions’ results, the change timeline and the times, taken from stored checks.
What each region saw
eu-1
eu-2
Downboth regions agree
The result and the time from each region, then the status code, redirects, certificate, DNS answer and timings on the incident page.
One source for the three
- Incident page
- Monthly report
- Status page
The same stored checks feed the monthly report and the status page, so the three never disagree.
- See a sample monthly report (demo data).
- See a sample status page (demo data).
What this does not do
Two regions remove a common false alarm and leave a record. They are an outside view of the public page, from the regions Beamfar runs in.
- It does not make a site faster
- Two regions confirm an outage; they do not change how the site answers.
- It does not see behind a login
- A problem that only shows after signing in is not caught by an outside check.
- It describes today’s code
- Beamfar is new and has no paying customers yet. Everything on this page describes what the code does today.
Questions
How does Beamfar decide a site is down?
Two regions must both fail to get a valid answer. One region failing is shown as a regional issue and sends no alert.
What is in an alert?
What each region saw: the result and the time, a link to the incident page for the full evidence, and a note when something changed in the hour before.
What happens when a check gives no valid result?
It is shown as data unavailable. It is never counted as up and never shown as zero.
Guides that go with this page
Longer reads on the same ground, in plain words, each with its sources.
- Two-region confirmation and false alarmsWhy single-location checks raise false alarms, how re-checks and multi-region confirmation differ and what each costs.
- How often should you check a website?How the check interval sets the delay before you learn of an outage, with the formulas and what to choose per site type.
- What a 200 OK response hidesWhy a 200 status can hide soft 404s, error pages, broken checkouts and stale caches, and how to check content too.
- Website down: a first-15-minutes triage checklistA step-by-step triage for a site-down alert: confirm it, find the broken layer and decide what to tell the client.
Know before your clients do.
Add your client sites and both regions start checking. Judge the evidence yourself.
No card for the trial.