Monitoring
A request every 60 seconds, from four regions.
Tenvey asks for your pages the way a visitor would and records what came back. When one region sees something the other three do not, that difference is the finding.
What one check does.
- 01Resolve and connectDNS lookup and the TLS handshake are timed separately, so a slow resolver never gets blamed on your server.
- 02Request the URLNot only the homepage. Every path you name gets its own check — the pricing page, the login form, the checkout.
- 03Read the responseStatus code, time to first byte, and a string you expect to find in the body. A 200 that renders an empty page still fails.
- 04Rotate the regionThe next check leaves from the next region, so each of the four sees your site every four minutes.
Check log
every 60s
- Zurich200186 ms
- Frankfurt200201 ms
- Virginia200243 ms
- Singapore500—
- Zurich200192 ms
Five consecutive checks. Singapore returned a 500 at 14:29:04 and Virginia answered 200 a minute later — one failure, not an outage.
One failed check is not an incident.
A single 500 from Singapore is as likely to be a network hiccup as a broken server. Tenvey re-runs the check immediately from the other three regions and only opens an incident when at least two of them agree.
That is also why the uptime figure is worth reading. It counts confirmed failures, not every dropped packet.
- Interval
- 60 sone request per minute, per URL
- Regions
- 4Frankfurt, Zurich, Virginia, Singapore
- Confirmation
- 3re-checks from the other regions, within 20 s
- Uptime, 90 days
- 99.94%129,600 checks, 77 of them failed
Incident · /checkout
CriticalCheckout returns a server error
128 requests to /checkout returned a 500 in the last 24 hours, starting 14 minutes after deployment v2.4.1.
POST /checkout → 500 Internal Server Error · 128 occurrences · first at 16:38, last at 23:51
Next step
Roll back v2.4.1 or fix the payment handler it changed. This blocks every purchase.
An outage arrives with its cause attached.
128 of the 433 checks on /checkout since 16:38 came back 500. The failure is intermittent, which is why the homepage still looks fine and why a person clicking around would probably have missed it.
Tenvey knows deployment v2.4.1 went out at 16:24, so the incident opens with that release named in the first line rather than leaving you to line up two timestamps yourself.
Also watched