Les articles d'aide existent en allemand et en anglais. Vous les lisez ici en anglais.
Limites et périmètre
How do I tell there is an incident?
The status page at /status shows measured availability for four components over the last 90 days, calculated from samples. We measure twice: from the inside through our own processing, and from the outside through an independent service that calls our addresses from several parts of the world. Both figures sit side by side, along with since when we have been measuring and how many samples each is based on.
Where do I see whether Metrunio is available, and how reliable is that figure?
What is measured
- Event intake, the critical part. If it fails, measurement time is lost.
- Processing and attribution, measured by the age of the oldest waiting event, not by whether the process is running.
- Customer portal.
- Read API.
Measured twice, and why
The large figure is measured from the inside: our own processing checks the other components every minute. If it fails too, no sample is written at all, not even an “unavailable” one. That is why we calculate against the expected number of samples, so a gap counts as downtime. Otherwise a total outage would look like a perfect week.
The smaller figure below it is measured from the outside: an independent service calls our addresses at short intervals from several parts of the world. A minute counts as available only once the majority of locations got through. That way a network fault somewhere in the world does not become an outage at our end, while an outage stays one even if a single location still gets through.
Neither replaces the other. The inside measurement sees the backlog in the queue, which no call from outside can detect. That is why “processing and attribution” carries no second figure. The outside measurement sees everything between you and our application: network, load balancing and certificate.
Note
What the outside measurement still does not answer: whether the path out of your own network was clear. If nothing arrives at your end while both figures are green, the problem is probably between us. Support can help from there. Where the two figures differ noticeably, the cause lies between the services rather than inside them; that too is worth telling support.
What happens during an incident
An outage in processing causes delay, not loss: events wait in a queue and are caught up. Measurement time is only lost if intake itself fails, which is why it runs separately from the portal and reporting.
Legal note
The figures on the status page are a measurement, not a guaranteed level of availability. A contractual commitment (SLA) is negotiated separately.
Dernière vérification le 13 oct. 2026
Également demandé
- Do you have a status page?
- Is Metrunio down right now?
- What is your availability?
- How do you measure uptime?
- Do you measure from outside or only internally?
- Why are there two availability figures next to each other?
- How do I know the figures are real?
Articles connexes
- How is Metrunio built technically?
Which technical building blocks does Metrunio run on today?
Data is stored and evaluated in PostgreSQL. A BigQuery analytics mirror runs for all accounts; reports are calculated from it once the data volume warrants it. We do not announce future features: what is listed here exists.
- How do I reach support?
Who do I turn to with a question, and how fast will I get an answer?
Through the chat on the website, by email to support@metrunio.com or in the portal under “Help”; anything our help centre covers, the chat answers immediately and names the articles. On plans from Pro, we get back to you by the next working day if your measurement fails, and answer all other questions in the order they arrive. On the free Start plan, the help centre and chat answer your questions, and a person answers security, data protection and legal questions.
- How long until data appears?
When do I see the first numbers after installing?
Individual events appear in the live test within seconds. Journeys and conversions are created during processing and usually appear within minutes; attribution shares are calculated afterwards.