# Monitoring

**Available in:** Sites, for supported production monitoring configurations. Projects use Logs and development diagnostics; Chats and Agents do not have this Site Monitoring dashboard.

Monitoring helps you understand failures in an operating Site. It connects exception evidence with occurrence history and release context so you can investigate the affected code and check a correction.

## Set up exception reporting

Open **Monitoring** on the runtime Site. Under **Monitoring configuration**, choose an available provider, select **Save configuration**, and release an update to apply it. On a restricted example Site, trigger a handled test scenario that reports a uniquely named exception through the configured integration. Find that issue in Monitoring and compare its release and occurrence time with the test. Remove the example trigger before making the Site public.

## Investigate an exception

- Identify the affected Site and the observed release.
- Review the exception, available stack details, and occurrence history.
- Use related Logs to understand what happened around the failure.
- Reproduce the behavior with an appropriate, authorized setup.
- Release a correction and check for recurrence.

Stack interpretation depends on the code and mapping available for that exact release. Missing attribution should remain an explicit gap; do not assume a trace from one revision describes another.

## Distinguish monitoring from health

| Signal | Question |
| --- | --- |
| Analytics | How did measured request behavior change? |
| Exception evidence | What application failure was observed? |
| Logs | What happened around the event? |
| Runtime health | Can the process run and serve traffic? |

A healthy process can still contain a broken workflow. A quiet exception list can also reflect missing instrumentation or unavailable collection rather than a problem-free application.

## Configure responsibly

Review the Site's available monitoring controls and any provider setup required for the workflow. Keep diagnostic data proportionate and sanitize secrets and private content. Access to production diagnostics requires the appropriate resource permission.

Collection coverage and provider integration vary by the configured capability. Verify a representative event before relying on monitoring for an important workflow. See [Logs](/docs/services/logs) and [Security](/docs/services/security).

## Example: an exception follows an update

If exceptions increase after a release, compare their occurrence times and available release attribution. Use Logs to inspect surrounding activity and determine whether the failure belongs to new code, configuration, or a dependency.

Verify the corrected workflow after release; absence of another collected exception is not enough on its own. See [Releases](/docs/sites/releases) and [Testing](/docs/services/testing).