oglog/etc/grafana/provisioning/alerting/alerts.yaml

133 lines
4.3 KiB
YAML

apiVersion: 1
groups:
- orgId: 1
name: root-disk-usage-check
folder: Node Exporter Alerts
interval: 1m
rules:
- uid: deknkrm5dz400b
title: Root FS Used
condition: C
data:
- refId: A
relativeTimeRange:
from: 86400
to: 0
datasourceUid: PBFA97CFB590B2093
model:
adhocFilters: []
datasource:
type: prometheus
uid: PBFA97CFB590B2093
editorMode: code
exemplar: false
expr: 100 - ((node_filesystem_avail_bytes{instance="ogserver.opengnsys:9100",job="ogserver",mountpoint="/",fstype!="rootfs"} * 100) / node_filesystem_size_bytes{instance="ogserver.opengnsys:9100",job="ogserver",mountpoint="/",fstype!="rootfs"})
format: time_series
instant: true
interval: ""
intervalFactor: 1
intervalMs: 15000
maxDataPoints: 43200
range: false
refId: A
step: 240
- refId: C
datasourceUid: __expr__
model:
conditions:
- evaluator:
params:
- 90
type: gt
operator:
type: and
query:
params:
- C
reducer:
params: []
type: last
type: query
datasource:
type: __expr__
uid: __expr__
expression: A
intervalMs: 1000
maxDataPoints: 43200
refId: C
type: threshold
dashboardUid: rYdddlPWk
panelId: 154
noDataState: NoData
execErrState: Error
annotations:
__dashboardUid__: rYdddlPWk
__panelId__: "154"
description: |-
Porcentaje de uso actual: ${A.Value}%
Punto de montaje: /
Job: ${__field.labels.job}
summary: 'Alerta: Uso alto del sistema de archivos raíz en ${__field.labels.instance}'
isPaused: false
notification_settings:
receiver: lgromero-email
- orgId: 1
name: server-host-down
folder: Node Exporter Alerts
interval: 10s
rules:
- uid: deknvl4y9svlsb
title: Server Down
condition: C
data:
- refId: A
relativeTimeRange:
from: 600
to: 0
datasourceUid: PBFA97CFB590B2093
model:
disableTextWrap: false
editorMode: builder
expr: up{job="ogserver"} == bool 0
fullMetaSearch: false
includeNullMetadata: true
instant: true
intervalMs: 1000
legendFormat: __auto
maxDataPoints: 43200
range: false
refId: A
useBackend: false
- refId: C
datasourceUid: __expr__
model:
conditions:
- evaluator:
params:
- 1
type: lt
operator:
type: and
query:
params:
- C
reducer:
params: []
type: last
type: query
datasource:
type: __expr__
uid: __expr__
expression: A
intervalMs: 1000
maxDataPoints: 43200
refId: C
type: threshold
noDataState: NoData
execErrState: Error
annotations:
description: Servidor ogcore caido
summary: Servidor ogcore caido
isPaused: false
notification_settings:
receiver: lgromero-email