What's new in Grafana alerting

Grafana alerting has several enhancements over classic dashboard alerting.

Create multidimensional alerting

You can now create a single alerting rule that gives you system-wide visibility, generating multiple alert instances from a single alert rule. For example, you can create a rule to monitor the disk usage of multiple mount points on a single host. The evaluation engine returns multiple time series from a single query, with each time series identified by its label set.

Create alerts outside of dashboards

Unlike classic dashboard alerts, with Grafana alerting you can create queries and expressions that combine data from multiple sources in unique ways. You can still link dashboards and panels to alerting rules using their ID and quickly troubleshoot the system under observation.

Since unified alerts are no longer directly tied to panel queries, they do not include images or query values in the notification email. You can use customized notification templates to view query values.

Create Loki and Cortex alerting rules

In Grafana alerting, you can manage Loki and Cortex alerting rules using the same UI and API as your Grafana managed alerts.

View and search alerts from Amazon Managed Service for Prometheus and other Prometheus compatible data sources

Alerts for Amazon Managed Service for Prometheus and Prometheus compatible data sources are now listed in the Alerting interface. You can search for labels across multiple data sources to quickly find relevant alerts.

Special alerts for alert state NoData and Error

Grafana alerting generates special alerts that have the following labels, when evaluation of an alerting rule produces state NoData or Error:

You can handle these alerts the same as regular alerts, for example, by adding a silence, or routing to a contact point.