Monitoring user activities with event metrics

For now, only metrics for user events are captured. For example, you can monitor the number of logins, login failures, or token refreshes performed.

The metrics are exposed using the standard metrics endpoint, and you can use it in your own metrics collection system to create dashboards and alerts.

The metrics are reported as counters per NQRust-Identity instance. The counters are reset on the restart of the instance. If you have multiple instances running in a cluster, you will need to collect the metrics from all instances and aggregate them to get per a cluster view.

Enable event metrics

To start collecting event metrics, enable metrics and enable the metrics for user events.

The following shows the required startup parameters:

bin/kc.[sh|bat] start --metrics-enabled=true --event-metrics-user-enabled=true ...

By default, there is a separate metric for each realm. To break down the metric by client and identity provider, you can add those metrics dimension using the configuration option event-metrics-user-tags. This can be useful on installations with a small number of clients and IDPs. This is not recommended for installations with a large number of clients or IDPs as it will increase the memory usage of NQRust-Identity and as it will increase the load on your monitoring system.

The following shows how to configure NQRust-Identity to break down the metrics by all three metrics dimensions:

bin/kc.[sh|bat] start ... --event-metrics-user-tags=realm,idp,clientId ...

You can limit the events for which NQRust-Identity will expose metrics.

The following example limits the events collected to LOGIN and LOGOUT events:

bin/kc.[sh|bat] start ... --event-metrics-user-events=login,logout ...

Relevant options