ansible-ffibk/roles/service-prometheus/files/alert.rules

groups:
- name: blackbox
  rules:
  - alert: Icmp4Timeout
    expr: (probe_success{job="icmp4"} == 0 and on(hostname) ALERTS{alertname="InstanceDown"} == 0) or (probe_success{job="icmp4",hostname=~"zuckerwatte|aubergine|glueckskeks"} == 0)
    for: 5m
    annotations:
      description: 'ICMP requests to the primary IPv4 address timed out'
      summary: 'Instance {{ $labels.instance }} does not respond to ICMPv4 echo requests'

  - alert: Icmp6Timeout
    expr: (probe_success{job="icmp6"} == 0 and on(hostname) ALERTS{alertname="InstanceDown"} == 0) or (probe_success{job="icmp6",hostname=~"zuckerwatte|aubergine|glueckskeks"} == 0)
    for: 5m
    annotations:
      description: 'ICMP requests to the primary IPv6 address timed out'
      summary: 'Instance {{ $labels.instance }} does not respond to ICMPv6 echo requests'

- name: node
  rules:
  - alert: InstanceDown
    expr: up{job="node"} == 0
    for: 5m
    annotations:
      description: 'The instance is down for more than 5 minutes'
      summary: 'Instance {{ $labels.instance }} is down'

  - alert: ExporterDown
    expr: up{job!="node"} == 0 and ON(hostname) ALERTS{alertname="InstanceDown"} == 0
    for: 5m
    annotations:
      description: 'An exporter is down for more than 5 minutes'
      summary: 'Exporter {{ $labels.instance }} is down'

  - alert: InstanceHighCpu
    expr: 100 - (avg(rate(node_cpu_seconds_total{mode="idle"}[5m])) BY (instance) * 100) > 90
    for: 5m
    annotations:
      description: 'CPU usage above 90% for more than 5m'
      summary: 'Instance {{ $labels.instance }}: cpu usage at {{ $value }}'
      value: '{{ $value }}'

  - alert: InstanceHighCpuLong
    expr: 100 - (avg(rate(node_cpu_seconds_total{mode="idle"}[5m])) BY (instance) * 100) > 90
    for: 30m
    annotations:
      description: 'CPU usage above 90% for more than 30m'
      summary: 'Instance {{ $labels.instance }}: persistent cpu usage at {{ $value }}'
      value: '{{ $value }}'

  - alert: InstanceLowMem
    expr: node_memory_MemAvailable_bytes / 1024 / 1024 < node_memory_MemTotal_bytes / 1024 / 1024 / 10
    for: 3m
    annotations:
      description: 'Less than 10% of free memory'
      summary: 'Instance {{ $labels.instance }}: {{ $value }}MB of free memory'
      value: '{{ $value }}'

  - alert: InstanceLowDiskPrediction4Hours
    expr: predict_linear(node_filesystem_free_bytes{device=~"/dev/.*",job="node"}[1h], 4 * 3600) < 0
    for: 30m
    annotations:
      description: 'Disk {{ $labels.mountpoint }} ({{ $labels.device }}) will be full in less than 4 hours'
      summary: 'Instance {{ $labels.instance }}: Disk {{ $labels.mountpoint }} ({{ $labels.device }}) will be full in less than 4 hours'

  - alert: InstanceLowDiskPrediction12Hours
    expr: predict_linear(node_filesystem_free_bytes{device=~"/dev/.*",job="node"}[3h], 12 * 3600) < 0
    for: 60m
    annotations:
      description: 'Disk {{ $labels.mountpoint }} ({{ $labels.device }}) will be full in less than 12 hours'
      summary: 'Instance {{ $labels.instance }}: Disk {{ $labels.mountpoint }} ({{ $labels.device }}) will be full in less than 12 hours'

  - alert: InstanceLowDiskAbs
    expr: node_filesystem_avail_bytes{mountpoint="/"} / 1024 / 1024 < 1024
    for: 1m
    annotations:
      description: 'Less than 1GB of free disk space left on the root filesystem'
      summary: 'Instance {{ $labels.instance }}: {{ $value }}MB free disk space on {{ $labels.device }}'
      value: '{{ $value }}'

  - alert: InstanceLowDiskPerc
    expr: 100 * (node_filesystem_free_bytes / node_filesystem_size_bytes) < 10
    for: 1m
    annotations:
      description: 'Less than 10% of free disk space left on a device'
      summary: 'Instance {{ $labels.instance }}: {{ $value }}% free disk space on {{ $labels.device }}'
      value: '{{ $value }}'

  - alert: ServiceFailed
    expr: node_systemd_unit_state{state="failed"} > 0
    for: 1m
    annotations:
      description: 'A systemd unit went into failed state'
      summary: 'Instance {{ $labels.instance }}: Service {{ $labels.name }} failed'
      value: '{{ $labels.name }}'

  - alert: ServiceFlapping
    expr: changes(node_systemd_unit_state{state="failed"}[5m]) > 5 or
     (changes(node_systemd_unit_state{state="failed"}[1h]) > 15 unless changes(node_systemd_unit_state{state="failed"}[30m]) < 7)
    annotations:
      description: 'A systemd service changed its state more than 5x/5min or 15x/1h'
      summary: 'Instance {{ $labels.instance }}: Service {{ $labels.name }} is flapping'
      value: '{{ $labels.name }}'

- name: fastd
  rules:
  - alert: FastdNoTraffic
    expr: irate(fastd_tx_bytes[5m]) == 0 and on(hostname,interface) (fastd_uptime_seconds / 60) > 15
    for: 5m
    annotations:
      description: 'No TX data was seen on a fastd interface for more than 5 minutes'
      summary: 'Instance {{ $labels.instance }}: No traffic on {{ $labels.interface }}'