Application health
The Health card on an application's page shows whether the application answers from outside. The platform requests each public address of the running revision once a minute and records the result. The card shows what was true when the page loaded.
Reading the card
| Figure | Meaning |
|---|---|
| Badge | The worst state across the application's addresses |
| Uptime (24 hours, 7 days, 30 days) | The share of checks that got a healthy answer. A dash means no data yet |
| Response time p50 and p95 | The median and the slow tail, over the last 24 hours |
| Address rows | The state, the HTTP status, the response time, and the 24-hour uptime of each address |
| Likely cause | A hint under an address that failed |
| Outages | Each period when an address was down, with its start and end. An open outage reads "still down" |
Each address has one of these states:
| State | Meaning |
|---|---|
| up | A healthy status code, at normal speed |
| degraded | A healthy status code, but slower than 3 seconds |
| down | The request failed or returned a status code you didn't list as healthy |
| gated | An access gate guards the address. The platform sends no request, and doesn't bypass the gate. A gated check counts in neither the up time nor the down time |
| not checked | Nothing checked the application in the last 15 minutes |
Check settings
An owner or admin sees Check settings in the same card.
| Field | Default | Meaning |
|---|---|---|
| Path to request | / |
The path the platform requests on each address |
| Healthy status codes | 200-399 |
Codes and ranges that count as healthy. Write them as a list, for example 200-399,401 |
Select Save to apply the change. The next check uses it.
Email alerts
The platform emails the owner of the organization about an outage. It sends:
- One down email after two failed checks in a row.
- One recovered email when the address answers again.
One email covers every address of one application in one pass, and one outage sends one down email. A gated result, a stopped application, or an address that left the route set sends no recovery email.
Alerts are on by default. An owner or admin can select Turn off email alerts in the card. While alerts are off, nobody gets an email. If you turn them on during an outage, the down email goes out then.
The platform sends no Telegram or chat message for this. See Notifications for the in-app centre.