Interactive observability

Prometheus Rules vs Mimir Ruler

Switch rule engines, choose a synthetic probe, click a node, and inspect the exact YAML, PromQL, and alert payload involved.

Read blog

Clicked node

Prometheus rules

Prometheus evaluates rule_files against its local TSDB and sends firing alerts to Alertmanager.

Best for local scrape truth
Survives central metrics outages
Recording output stays local
Uses alerting.alertmanagers

Examples

groups:
  - name: synthetic-alerts
    interval: 30s
    rules:
      - alert: SyntheticUrlDown
        expr: probe_success{job="synthetics", module="http_2xx"} == 0
        for: 2m
        labels:
          severity: page
          team: platform
          source: prometheus
        annotations:
          summary: "URL monitor failed"
          description: "{{ $labels.instance }} failed module {{ $labels.module }}"

Invocation flow

Local rule path for HTTP URL

Prometheus scrapes Blackbox Exporter /probe.
probe_success lands in the local Prometheus TSDB.
Prometheus rule_files evaluate the alert expression.
Prometheus sends firing alerts to Alertmanager.
Alertmanager routes SyntheticUrlDown to platform-pagerduty.

Prometheus rules

Best for local truth

Evaluates rule files against local TSDB data. Use it for scrape health, local Blackbox probes, cluster health, and alerts that should still fire if the central metrics path breaks.

rule_files + alerting.alertmanagers

Mimir Ruler

Best for global truth

Evaluates tenant PromQL against Mimir. Use it for global SLOs, multi-cluster synthetic ratios, long-retention queries, and central rule management.

ruler.alertmanager_url + Mimir tenant data

Alertmanager

Notification truth

Receives alert objects from either engine. It routes by labels, groups related alerts, deduplicates repeated alerts, applies silences, and sends notifications.

labels -> route -> receiver