How do you configure the rule evaluation interval in Prometheus?

Prometheus evaluates alerting and recording rules at a regular cadence determined by theevaluation_intervalsetting. This can be defined globally in the mainPrometheus configuration file (prometheus.yml)under the global: section or overridden for specific rule groups in therule configuration files.

The global evaluation_interval specifies how frequently Prometheus should execute all configured rules, while rule-specific intervals can fine-tune evaluation frequency for individual groups. For instance:

global:

evaluation_interval: 30s

This means Prometheus evaluates rules every 30 seconds unless a rule file specifies otherwise.

This parameter is distinct from scrape_interval, which governs metric collection frequency from targets. It has no relation to TSDB, service discovery, or command-line flags.