Service status
These checks run every minute from a host outside our network, so they see what you see rather than what our own servers believe. The alerts API and the live feed are independent, so one can be down while the other keeps serving.
Loading current status…
Each block is one day, oldest on the left, going back three months. Blocks are shaded by how much of the day was down, so a single bad minute and a six-hour outage are not the same day, so they are not the same colour. Grey blocks are days before we started watching. Hover a block for that day’s figures.
Where our alerts come from, and how fast
The same warning reaches us over three independent feeds. We keep all three because they fail differently, since the satellite feed keeps working when our internet does not, and we take whichever copy arrives first. Below is how they actually compare, measured on hazards more than one feed carried.
Loading feed timings…
Time from our server publishing an alert to it arriving on this page, measured live on your own connection.
What these checks actually test
- Website
- Loads wxalerts.org and confirms real content came back, not an error page served with a success code.
- Alerts API
- Checks the API, its database and the geometry engine, and confirms alert data is still arriving, because an API that answers instantly with hours-old data counts as degraded, not healthy.
- Live alert feed (MQTT)
- Opens a real connection to the broker, signs in and reads a live message, the same path the apps and integrations use. A port check would pass while the feed itself was silent.
- Data pipeline
- The ingest and forecast processors have no public address and check in on their own schedule instead. Until that reporting is switched on they show as not yet reporting, which is not the same as down.