Settings
These settings look across the project.
| Setting | What it means in the project | Default |
|---|---|---|
| Project-wide error spike | Email when total error volume across the project crosses the count and time window you choose. | Off, 100 events in 15 minutes |
| Project p95 latency | Email when recent transaction latency is above the millisecond limit you choose. | Off, 1,000 ms |
| Check-in monitors | Email when scheduled work is missed, then again when it recovers. | On |
Project-wide error spike is not the same as the single-error volume threshold. This setting is about the whole project becoming noisy, even if no single grouped issue crosses its own threshold.
Verify
Health alerts are not always instant.
These alerts are evaluated in the background, so expect a short delay. To test, temporarily lower the threshold, generate matching activity, and wait for the project health email instead of refreshing the settings page immediately.
Troubleshooting
Make sure the project-level condition actually happened.
| Symptom | What to check |
|---|---|
| No spike alert | The total project error count may not have crossed the threshold in the selected window. |
| No latency alert | Transactions may be missing durations, or p95 may still be below the configured millisecond limit. |
| No missed monitor alert | The check-in may not have an expected interval, or the monitor sweep may not have run yet. |
| Only one group is noisy | Use Error triage alerts for a single grouped issue; this page is for the whole project. |
Keep Sidekiq running for the sweeps that evaluate health alerts.