{ lib, pkgs, config, ... }: { services.grafana.provision.alerting.rules.settings.groups = [ { name = "CPUUsageAlerts"; folder = "System Alerts"; interval = "1m"; rules = [ { uid = "high-cpu-usage-alert-uid"; title = "HighCPUUsage"; condition = "D"; # Condition is now D data = [ # Query A: Calculate CPU usage percentage { refId = "A"; datasourceUid = "vm-datasource-uid"; queryType = "prometheus"; relativeTimeRange = { from = 60; to = 0; }; # Query over the last minute model = { # Calculate average CPU usage over 1m, grouped by instance and job expr = ''(1 - avg by (instance, job) (rate(node_cpu_seconds_total{mode="idle"}[1m]))) * 100''; legendFormat = "CPU usage on {{instance}} ({{job}})"; instant = false; # This is a range query }; } # Expression C: Reduce Query A to its last value, preserving labels { refId = "C"; datasourceUid = "__expr__"; model = { type = "reduce"; expression = "A"; # Input is Query A reducer = "last"; # Get the last value of each series in A }; } # Expression D: Apply math condition to the reduced values from C { refId = "D"; datasourceUid = "__expr__"; model = { type = "math"; expression = "$C > 90"; # Alert if CPU usage from C is > 90% }; } ]; for = "5m"; # Duration the condition must be met noDataState = "NoData"; execErrState = "Error"; annotations = { summary = "High CPU usage on {{ $labels.instance }}"; description = ''CPU usage on {{ $labels.instance }} (job: {{ $labels.job }}) has been above 90% for more than 5 minutes. Current value: {{ if $values.C }}{{ $values.C | humanizePercentage }}{{ else }}N/A{{ end }}%.''; }; labels = { severity = "warning"; category = "performance"; }; } ]; } ]; }