# services/monitoring/grafana-alerting-config.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: grafana-alerting
  namespace: monitoring
  labels:
    grafana_alerting: "1"
data:
  alerting.yaml: |
    apiVersion: 1
    contactPoints:
      - orgId: 1
        name: email-admins
        receivers:
          - uid: email-admins
            type: email
            settings:
              addresses: ${GRAFANA_ALERT_EMAILS}
              singleEmail: true
    policies:
      - orgId: 1
        receiver: email-admins
        group_by:
          - alertname
  rules.yaml: |
    apiVersion: 1
    groups:
      - orgId: 1
        name: atlas-disk
        folder: Alerts
        interval: 1m
        rules:
          - uid: disk-pressure-root
            title: "Node rootfs high (>80%)"
            condition: C
            for: "10m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 600
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: avg by (node) ((avg by (instance) ((1 - (node_filesystem_avail_bytes{mountpoint="/",fstype!~"tmpfs|overlay"} / node_filesystem_size_bytes{mountpoint="/",fstype!~"tmpfs|overlay"})) * 100)) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=""}, "node", "$1", "nodename", "(.*)"))
                  legendFormat: '{{node}}'
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [80]
                        type: gt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: Error
            annotations:
              summary: "{{ $labels.node }} rootfs >80% for 10m"
            labels:
              severity: warning
          - uid: disk-growth-1h
            title: "Node rootfs growing fast (>1Gi in 1h)"
            condition: C
            for: "10m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 3600
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: increase((node_filesystem_size_bytes{mountpoint="/",fstype!~"tmpfs|overlay"} - node_filesystem_free_bytes{mountpoint="/",fstype!~"tmpfs|overlay"})[1h]) / 1024 / 1024 / 1024
                  legendFormat: '{{instance}}'
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [1]
                        type: gt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: Error
            annotations:
              summary: "{{ $labels.instance }} rootfs grew >1Gi in the last hour"
            labels:
              severity: warning
      - orgId: 1
        name: atlas-cpu
        folder: Alerts
        interval: 1m
        rules:
          - uid: cpu-high-10m
            title: "Node CPU high (>90% for 10m)"
            condition: C
            for: 10m
            data:
              - refId: A
                relativeTimeRange:
                  from: 600
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: avg_over_time((1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m]))) * 100)[10m:1m] * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")
                  legendFormat: '{{instance}}'
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [90]
                        type: gt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: NoData
            annotations:
              summary: "{{ $labels.node }} CPU >90% for 10m"
            labels:
              severity: warning
      - orgId: 1
        name: atlas-metrics
        folder: Alerts
        interval: 1m
        rules:
          - uid: victoria-metrics-down
            title: "VictoriaMetrics unavailable (>30m)"
            condition: C
            for: "30m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 600
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: sum(up{job="victoriametrics"})
                  legendFormat: victoriametrics
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [1]
                        type: lt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: Alerting
            execErrState: Alerting
            annotations:
              summary: "VictoriaMetrics is unavailable for >30m"
            labels:
              severity: critical
      - orgId: 1
        name: maintenance
        folder: Alerts
        interval: 1m
        rules:
          - uid: maint-sweeper
            title: "Maintenance sweeper not ready"
            condition: C
            for: "5m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 300
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: kube_daemonset_status_number_ready{namespace="maintenance",daemonset="node-image-sweeper"} / on(namespace,daemonset) kube_daemonset_status_desired_number_scheduled{namespace="maintenance",daemonset="node-image-sweeper"}
                  legendFormat: '{{daemonset}}'
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [1]
                        type: lt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: Error
            annotations:
              summary: "node-image-sweeper not fully ready"
            labels:
              severity: warning
          - uid: maint-cron-stale
            title: "Maintenance CronJobs stale (>3h since success)"
            condition: C
            for: "5m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 300
                  to: 0
                datasourceUid: atlas-vm
                model:
                  expr: time() - max by (cronjob) (kube_cronjob_status_last_successful_time{namespace="maintenance",cronjob="image-sweeper"}) and on(cronjob) (kube_cronjob_spec_suspend{namespace="maintenance",cronjob="image-sweeper"} == 0)
                  intervalMs: 60000
                  maxDataPoints: 43200
                  legendFormat: '{{cronjob}}'
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [10800]
                        type: gt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: Error
            annotations:
              summary: "Maintenance cronjob stale >3h since last success"
            labels:
              severity: warning
      - orgId: 1
        name: postmark
        folder: Alerts
        interval: 1m
        rules:
          - uid: postmark-bounce
            title: "Postmark bounce rate high"
            condition: C
            for: "10m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 300
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: POSTMARK_OUTBOUND_BOUNCE_RATE{window="1d"}
                  legendFormat: bounce 1d
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [5]
                        type: gt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: Error
            annotations:
              summary: "Postmark 1d bounce rate >5%"
            labels:
              severity: warning
          - uid: postmark-api-down
            title: "Postmark exporter down"
            condition: C
            for: "5m"
            data:
              - refId: A
                relativeTimeRange:
                  from: 300
                  to: 0
                datasourceUid: atlas-vm
                model:
                  intervalMs: 60000
                  maxDataPoints: 43200
                  expr: POSTMARK_API_UP
                  legendFormat: api up
                  datasource:
                    type: prometheus
                    uid: atlas-vm
              - refId: B
                datasourceUid: __expr__
                model:
                  expression: A
                  intervalMs: 60000
                  maxDataPoints: 43200
                  reducer: last
                  type: reduce
              - refId: C
                datasourceUid: __expr__
                model:
                  expression: B
                  intervalMs: 60000
                  maxDataPoints: 43200
                  type: threshold
                  conditions:
                    - evaluator:
                        params: [1]
                        type: lt
                      operator:
                        type: and
                      reducer:
                        type: last
                      type: query
            noDataState: NoData
            execErrState: Error
            annotations:
              summary: "Postmark exporter reports API down"
            labels:
              severity: critical