Incidents
Apply.Build watches every app for problems and groups related signals into an incident. There is nothing to install or configure, and your logs can be in any format.
What is watched
- Logs. Error lines are picked out of your app's logs and grouped into patterns, so the same error with a different ID or timestamp counts as one.
- Request metrics. The server error (5xx) rate and response times, compared with what is usual for your app.
- Traces. Which endpoints are failing or slow.
- Lifecycle. Unexpected restarts, crash loops and running out of memory.
Signals
An incident is made of one or more signals:
| Signal | Meaning |
|---|---|
| New error | An error pattern that hasn't been seen before |
| Error spike | Far more errors in the last 5 minutes than usual |
| More errors than usual | A smaller rise in errors |
| Server error (5xx) rate | A larger share of requests failing than usual |
| Slower responses than usual | The 95th percentile response time is above its baseline |
| Restarted unexpectedly | The app stopped and was restarted outside a deployment |
| Crashing repeatedly | The app is in a crash loop |
| Ran out of memory | The app was stopped for using more memory than its size |
Each incident has a severity of Minor, Major or Critical.
Viewing incidents
Open an app's page and find the Incidents card. Each incident shows when it opened, its signals, the error lines behind it and the affected endpoints. An incident resolves itself once its signals clear.
Notifications
To get incident alerts in Slack or Discord, turn on App status for a channel under Notifications. Incidents are sent alongside the app's other status changes.
Ignoring errors
Some errors are expected, like a failed login. Click Ignore next to an error line and it stops causing incidents. Errors that look like obvious noise are ignored automatically.
Ignored errors are listed on the Ignored errors tab of the Incidents card. Click Undo to start tracking an error again.