Deploy dashboards

This commit is contained in:
Rodrigue Koffi
2022-08-04 15:19:46 +02:00
parent d09f679ef5
commit cd827f6fb9
35 changed files with 37766 additions and 66 deletions
+1 -1
View File
@@ -43,7 +43,7 @@ module "eks_observability_accelerator" {
# }
enable_infra_metrics = true
#infra_metrics_config = {}
# infra_metrics_config = {}
tags = local.tags
}
+10 -21
View File
@@ -65,6 +65,12 @@ resource "grafana_data_source" "amp" {
}
}
# dashboards
resource "grafana_folder" "this" {
title = "Observability Accelerator Dashboards"
}
module "java" {
count = var.enable_java ? 1 : 0
source = "./modules/workloads/java"
@@ -77,6 +83,8 @@ module "java" {
enable_recording_rules = var.enable_java_recording_rules
dashboards_folder_id = grafana_folder.this.id
depends_on = [
module.operator
]
@@ -91,28 +99,9 @@ module "infra" {
amp_endpoint = local.amp_ws_endpoint
amp_id = local.amp_ws_id
amp_region = local.amp_ws_region
config = var.infra_metrics_config
config = var.infra_metrics_config
# enable_kube_state_metrics = var.infra_metrics_config.enable_kube_state_metrics
# kms_create_namespace = var.infra_metrics_config.kms_create_namespace
# ksm_k8s_namespace = var.infra_metrics_config.ksm_k8s_namespace
# ksm_helm_chart_name = var.infra_metrics_config.ksm_helm_chart_name
# ksm_helm_chart_version = var.infra_metrics_config.ksm_helm_chart_version
# ksm_helm_release_name = var.infra_metrics_config.ksm_helm_release_name
# ksm_helm_repo_url = var.infra_metrics_config.ksm_helm_repo_url
# ksm_helm_settings = var.infra_metrics_config.ksm_helm_settings
# ksm_helm_values = var.infra_metrics_config.ksm_helm_values
# enable_node_exporter = var.infra_metrics_config.enable_node_exporter
# ne_create_namespace = var.infra_metrics_config.ne_create_namespace
# ne_helm_chart_name = var.infra_metrics_config.ne_helm_chart_name
# ne_helm_chart_version = var.infra_metrics_config.ne_helm_chart_version
# ne_helm_release_name = var.infra_metrics_config.ne_helm_release_name
# ne_helm_repo_url = var.infra_metrics_config.ne_helm_repo_url
# ne_helm_settings = var.infra_metrics_config.ksm_helm_settings
# ne_helm_values = var.infra_metrics_config.ksm_helm_values
dashboards_folder_id = grafana_folder.this.id
depends_on = [
module.operator
+1 -6
View File
@@ -98,12 +98,7 @@ groups:
EOF
}
# dashboards
resource "grafana_folder" "this" {
title = "Observability Accelerator Dashboards"
}
resource "grafana_dashboard" "this" {
folder = grafana_folder.this.id
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/default.json")
}
+4
View File
@@ -26,6 +26,10 @@ variable "amp_region" {
default = null
}
variable "dashboards_folder_id" {
type = string
}
variable "addon_context" {
description = "Input configuration for the addon"
type = object({
@@ -0,0 +1,155 @@
resource "grafana_dashboard" "alertmanager" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/alertmanager.json")
}
resource "grafana_dashboard" "workloads" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/workloads.json")
}
resource "grafana_dashboard" "scheduler" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/scheduler.json")
}
resource "grafana_dashboard" "proxy" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/proxy.json")
}
resource "grafana_dashboard" "prometheus" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/prometheus.json")
}
resource "grafana_dashboard" "podnetwork" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/pods-networking.json")
}
resource "grafana_dashboard" "pods" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/pods.json")
}
resource "grafana_dashboard" "pv" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/pesistentvolumes.json")
}
resource "grafana_dashboard" "nodes" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/nodes.json")
}
resource "grafana_dashboard" "necluster" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/nodeexpoter-use-cluster.json")
}
resource "grafana_dashboard" "nenodeuse" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/nodeexporter-use-node.json")
}
resource "grafana_dashboard" "nenode" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/nodeexporter-nodes.json")
}
resource "grafana_dashboard" "nwworload" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/networking-workloads.json")
}
resource "grafana_dashboard" "nsworkload" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/namespace-workloads.json")
}
resource "grafana_dashboard" "nspods" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/namespace-pods.json")
}
resource "grafana_dashboard" "nsnwworkload" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/namespace-nw-workloads.json")
}
resource "grafana_dashboard" "nsnw" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/namespace-networking.json")
}
resource "grafana_dashboard" "macos" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/macos.json")
}
resource "grafana_dashboard" "kubelet" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/kubelet.json")
}
resource "grafana_dashboard" "grafana" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/grafana.json")
}
resource "grafana_dashboard" "etcd" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/etcd.json")
}
resource "grafana_dashboard" "coredns" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/coredns.json")
}
resource "grafana_dashboard" "controller" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/controller.json")
}
resource "grafana_dashboard" "clusternw" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/cluster-networking.json")
}
resource "grafana_dashboard" "cluster" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/cluster.json")
}
resource "grafana_dashboard" "apis" {
count = var.config.enable_dashboards ? 1 : 0
folder = var.dashboards_folder_id
config_json = file("${path.module}/dashboards/apiserver.json")
}
@@ -0,0 +1,626 @@
{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": {
"type": "grafana",
"uid": "-- Grafana --"
},
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"target": {
"limit": 100,
"matchAny": false,
"tags": [],
"type": "dashboard"
},
"type": "dashboard"
}
]
},
"editable": false,
"fiscalYearStartMonth": 0,
"graphTooltip": 1,
"id": 22,
"iteration": 1656210900773,
"links": [],
"liveNow": false,
"panels": [
{
"collapsed": false,
"datasource": {
"type": "prometheus",
"uid": "prometheus"
},
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 0
},
"id": 6,
"panels": [],
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "prometheus"
},
"refId": "A"
}
],
"title": "Alerts",
"type": "row"
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 7,
"w": 12,
"x": 0,
"y": 1
},
"hiddenSeries": false,
"id": 2,
"legend": {
"alignAsTable": false,
"avg": false,
"current": false,
"max": false,
"min": false,
"rightSide": false,
"show": false,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(alertmanager_alerts{namespace=~\"$namespace\",service=~\"$service\"}) by (namespace,service,instance)",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}}",
"refId": "A"
}
],
"thresholds": [],
"timeRegions": [],
"title": "Alerts",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"format": "none",
"logBase": 1,
"show": true
},
{
"format": "none",
"logBase": 1,
"show": true
}
],
"yaxis": {
"align": false
}
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 7,
"w": 12,
"x": 12,
"y": 1
},
"hiddenSeries": false,
"id": 3,
"legend": {
"alignAsTable": false,
"avg": false,
"current": false,
"max": false,
"min": false,
"rightSide": false,
"show": false,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(rate(alertmanager_alerts_received_total{namespace=~\"$namespace\",service=~\"$service\"}[$__rate_interval])) by (namespace,service,instance)",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} Received",
"refId": "A"
},
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(rate(alertmanager_alerts_invalid_total{namespace=~\"$namespace\",service=~\"$service\"}[$__rate_interval])) by (namespace,service,instance)",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} Invalid",
"refId": "B"
}
],
"thresholds": [],
"timeRegions": [],
"title": "Alerts receive rate",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"format": "ops",
"logBase": 1,
"show": true
},
{
"format": "ops",
"logBase": 1,
"show": true
}
],
"yaxis": {
"align": false
}
},
{
"collapsed": false,
"datasource": {
"type": "prometheus",
"uid": "prometheus"
},
"gridPos": {
"h": 1,
"w": 24,
"x": 0,
"y": 8
},
"id": 7,
"panels": [],
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "prometheus"
},
"refId": "A"
}
],
"title": "Notifications",
"type": "row"
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 7,
"w": 6,
"x": 0,
"y": 9
},
"hiddenSeries": false,
"id": 4,
"legend": {
"alignAsTable": false,
"avg": false,
"current": false,
"max": false,
"min": false,
"rightSide": false,
"show": false,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 5,
"points": false,
"renderer": "flot",
"repeat": "integration",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(rate(alertmanager_notifications_total{namespace=~\"$namespace\",service=~\"$service\", integration=\"$integration\"}[$__rate_interval])) by (integration,namespace,service,instance)",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} Total",
"refId": "A"
},
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(rate(alertmanager_notifications_failed_total{namespace=~\"$namespace\",service=~\"$service\", integration=\"$integration\"}[$__rate_interval])) by (integration,namespace,service,instance)",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} Failed",
"refId": "B"
}
],
"thresholds": [],
"timeRegions": [],
"title": "$integration: Notifications Send Rate",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"format": "ops",
"logBase": 1,
"show": true
},
{
"format": "ops",
"logBase": 1,
"show": true
}
],
"yaxis": {
"align": false
}
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 7,
"w": 6,
"x": 0,
"y": 16
},
"hiddenSeries": false,
"id": 5,
"legend": {
"alignAsTable": false,
"avg": false,
"current": false,
"max": false,
"min": false,
"rightSide": false,
"show": false,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 5,
"points": false,
"renderer": "flot",
"repeat": "integration",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "histogram_quantile(0.99,\n sum(rate(alertmanager_notification_latency_seconds_bucket{namespace=~\"$namespace\",service=~\"$service\", integration=\"$integration\"}[$__rate_interval])) by (le,namespace,service,instance)\n) \n",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} 99th Percentile",
"refId": "A"
},
{
"datasource": {
"uid": "$datasource"
},
"expr": "histogram_quantile(0.50,\n sum(rate(alertmanager_notification_latency_seconds_bucket{namespace=~\"$namespace\",service=~\"$service\", integration=\"$integration\"}[$__rate_interval])) by (le,namespace,service,instance)\n) \n",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} Median",
"refId": "B"
},
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(rate(alertmanager_notification_latency_seconds_sum{namespace=~\"$namespace\",service=~\"$service\", integration=\"$integration\"}[$__rate_interval])) by (namespace,service,instance)\n/\nsum(rate(alertmanager_notification_latency_seconds_count{namespace=~\"$namespace\",service=~\"$service\", integration=\"$integration\"}[$__rate_interval])) by (namespace,service,instance)\n",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "{{instance}} Average",
"refId": "C"
}
],
"thresholds": [],
"timeRegions": [],
"title": "$integration: Notification Duration",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"format": "s",
"logBase": 1,
"show": true
},
{
"format": "s",
"logBase": 1,
"show": true
}
],
"yaxis": {
"align": false
}
}
],
"refresh": "30s",
"schemaVersion": 36,
"style": "dark",
"tags": [
"alertmanager-mixin"
],
"templating": {
"list": [
{
"current": {
"selected": false,
"text": "Prometheus",
"value": "Prometheus"
},
"hide": 0,
"includeAll": false,
"label": "Data Source",
"multi": false,
"name": "datasource",
"options": [],
"query": "prometheus",
"refresh": 1,
"regex": "",
"skipUrlSync": false,
"type": "datasource"
},
{
"current": {
"selected": false,
"text": "default",
"value": "default"
},
"datasource": {
"type": "prometheus",
"uid": "$datasource"
},
"definition": "",
"hide": 0,
"includeAll": false,
"label": "namespace",
"multi": false,
"name": "namespace",
"options": [],
"query": {
"query": "label_values(alertmanager_alerts, namespace)",
"refId": "Prometheus-namespace-Variable-Query"
},
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 1,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
},
{
"current": {
"selected": false,
"text": "kube-prometheus-stack-alertmanager",
"value": "kube-prometheus-stack-alertmanager"
},
"datasource": {
"type": "prometheus",
"uid": "$datasource"
},
"definition": "",
"hide": 0,
"includeAll": false,
"label": "service",
"multi": false,
"name": "service",
"options": [],
"query": {
"query": "label_values(alertmanager_alerts, service)",
"refId": "Prometheus-service-Variable-Query"
},
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 1,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
},
{
"current": {
"selected": false,
"text": "All",
"value": "$__all"
},
"datasource": {
"type": "prometheus",
"uid": "$datasource"
},
"definition": "",
"hide": 2,
"includeAll": true,
"multi": false,
"name": "integration",
"options": [],
"query": {
"query": "label_values(alertmanager_notifications_total{integration=~\".*\"}, integration)",
"refId": "Prometheus-integration-Variable-Query"
},
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 1,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
}
]
},
"time": {
"from": "now-1h",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"5s",
"10s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
],
"time_options": [
"5m",
"15m",
"1h",
"6h",
"12h",
"24h",
"2d",
"7d",
"30d"
]
},
"timezone": "utc",
"title": "Alertmanager / Overview",
"uid": "alertmanager-overview",
"version": 1,
"weekStart": ""
}
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,563 @@
{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": {
"type": "datasource",
"uid": "grafana"
},
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"target": {
"limit": 100,
"matchAny": false,
"tags": [],
"type": "dashboard"
},
"type": "dashboard"
}
]
},
"editable": true,
"fiscalYearStartMonth": 0,
"graphTooltip": 0,
"id": 15,
"iteration": 1656211285882,
"links": [],
"liveNow": false,
"panels": [
{
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"mappings": [],
"noValue": "0",
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 80
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 5,
"w": 6,
"x": 0,
"y": 0
},
"id": 6,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"mean"
],
"fields": "",
"values": false
},
"text": {},
"textMode": "auto"
},
"pluginVersion": "9.0.1",
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "grafana_alerting_result_total{job=~\"$job\", instance=~\"$instance\", state=\"alerting\"}",
"instant": true,
"interval": "",
"legendFormat": "",
"refId": "A"
}
],
"title": "Firing Alerts",
"type": "stat"
},
{
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 80
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 5,
"w": 6,
"x": 6,
"y": 0
},
"id": 8,
"options": {
"colorMode": "value",
"graphMode": "area",
"justifyMode": "auto",
"orientation": "auto",
"reduceOptions": {
"calcs": [
"mean"
],
"fields": "",
"values": false
},
"text": {},
"textMode": "auto"
},
"pluginVersion": "9.0.1",
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum(grafana_stat_totals_dashboard{job=~\"$job\", instance=~\"$instance\"})",
"interval": "",
"legendFormat": "",
"refId": "A"
}
],
"title": "Dashboards",
"type": "stat"
},
{
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"custom": {
"displayMode": "auto",
"inspect": false
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 80
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 5,
"w": 12,
"x": 12,
"y": 0
},
"id": 10,
"options": {
"footer": {
"fields": "",
"reducer": [
"sum"
],
"show": false
},
"showHeader": true
},
"pluginVersion": "9.0.1",
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "grafana_build_info{job=~\"$job\", instance=~\"$instance\"}",
"instant": true,
"interval": "",
"legendFormat": "",
"refId": "A"
}
],
"title": "Build Info",
"transformations": [
{
"id": "labelsToFields",
"options": {}
},
{
"id": "merge",
"options": {}
},
{
"id": "organize",
"options": {
"excludeByName": {
"Time": true,
"Value": true,
"branch": true,
"container": true,
"goversion": true,
"namespace": true,
"pod": true,
"revision": true
},
"indexByName": {
"Time": 7,
"Value": 11,
"branch": 4,
"container": 8,
"edition": 2,
"goversion": 6,
"instance": 1,
"job": 0,
"namespace": 9,
"pod": 10,
"revision": 5,
"version": 3
},
"renameByName": {}
}
}
],
"type": "table"
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"links": []
},
"overrides": []
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 5
},
"hiddenSeries": false,
"id": 2,
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 2,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum by (status_code) (irate(grafana_http_request_duration_seconds_count{job=~\"$job\", instance=~\"$instance\"}[1m])) ",
"interval": "",
"legendFormat": "{{status_code}}",
"refId": "A"
}
],
"thresholds": [],
"timeRegions": [],
"title": "RPS",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"$$hashKey": "object:157",
"format": "reqps",
"logBase": 1,
"show": true
},
{
"$$hashKey": "object:158",
"format": "short",
"logBase": 1,
"show": false
}
],
"yaxis": {
"align": false
}
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"links": []
},
"overrides": []
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 5
},
"hiddenSeries": false,
"id": 4,
"legend": {
"avg": false,
"current": false,
"max": false,
"min": false,
"show": true,
"total": false,
"values": false
},
"lines": true,
"linewidth": 1,
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 2,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"exemplar": true,
"expr": "histogram_quantile(0.99, sum(irate(grafana_http_request_duration_seconds_bucket{instance=~\"$instance\", job=~\"$job\"}[$__rate_interval])) by (le)) * 1",
"interval": "",
"legendFormat": "99th Percentile",
"refId": "A"
},
{
"datasource": {
"uid": "$datasource"
},
"exemplar": true,
"expr": "histogram_quantile(0.50, sum(irate(grafana_http_request_duration_seconds_bucket{instance=~\"$instance\", job=~\"$job\"}[$__rate_interval])) by (le)) * 1",
"interval": "",
"legendFormat": "50th Percentile",
"refId": "B"
},
{
"datasource": {
"uid": "$datasource"
},
"exemplar": true,
"expr": "sum(irate(grafana_http_request_duration_seconds_sum{instance=~\"$instance\", job=~\"$job\"}[$__rate_interval])) * 1 / sum(irate(grafana_http_request_duration_seconds_count{instance=~\"$instance\", job=~\"$job\"}[$__rate_interval]))",
"interval": "",
"legendFormat": "Average",
"refId": "C"
}
],
"thresholds": [],
"timeRegions": [],
"title": "Request Latency",
"tooltip": {
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"$$hashKey": "object:210",
"format": "ms",
"logBase": 1,
"show": true
},
{
"$$hashKey": "object:211",
"format": "short",
"logBase": 1,
"show": true
}
],
"yaxis": {
"align": false
}
}
],
"schemaVersion": 36,
"style": "dark",
"tags": [],
"templating": {
"list": [
{
"current": {
"selected": false,
"text": "Prometheus",
"value": "Prometheus"
},
"hide": 0,
"includeAll": false,
"multi": false,
"name": "datasource",
"options": [],
"query": "prometheus",
"queryValue": "",
"refresh": 1,
"regex": "",
"skipUrlSync": false,
"type": "datasource"
},
{
"allValue": ".*",
"current": {
"selected": false,
"text": "All",
"value": "$__all"
},
"datasource": {
"uid": "$datasource"
},
"definition": "label_values(grafana_build_info, job)",
"hide": 0,
"includeAll": true,
"multi": true,
"name": "job",
"options": [],
"query": {
"query": "label_values(grafana_build_info, job)",
"refId": "Billing Admin-job-Variable-Query"
},
"refresh": 1,
"regex": "",
"skipUrlSync": false,
"sort": 0,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
},
{
"allValue": ".*",
"current": {
"selected": false,
"text": "All",
"value": "$__all"
},
"datasource": {
"uid": "$datasource"
},
"definition": "label_values(grafana_build_info, instance)",
"hide": 0,
"includeAll": true,
"multi": true,
"name": "instance",
"options": [],
"query": {
"query": "label_values(grafana_build_info, instance)",
"refId": "Billing Admin-instance-Variable-Query"
},
"refresh": 1,
"regex": "",
"skipUrlSync": false,
"sort": 0,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
}
]
},
"time": {
"from": "now-6h",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"10s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
]
},
"timezone": "utc",
"title": "Grafana Overview",
"uid": "6be0s85Mk",
"version": 1,
"weekStart": ""
}
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,550 @@
{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": {
"type": "grafana",
"uid": "-- Grafana --"
},
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"target": {
"limit": 100,
"matchAny": false,
"tags": [],
"type": "dashboard"
},
"type": "dashboard"
}
]
},
"editable": false,
"fiscalYearStartMonth": 0,
"graphTooltip": 0,
"id": 18,
"iteration": 1656212469966,
"links": [],
"liveNow": false,
"panels": [
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 7,
"w": 18,
"x": 0,
"y": 0
},
"hiddenSeries": false,
"id": 2,
"interval": "1m",
"legend": {
"alignAsTable": true,
"avg": true,
"current": true,
"max": true,
"min": true,
"rightSide": true,
"show": true,
"total": false,
"values": true
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "(\n sum without(instance, node) (topk(1, (kubelet_volume_stats_capacity_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})))\n -\n sum without(instance, node) (topk(1, (kubelet_volume_stats_available_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})))\n)\n",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Used Space",
"refId": "A"
},
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum without(instance, node) (topk(1, (kubelet_volume_stats_available_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})))\n",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Free Space",
"refId": "B"
}
],
"thresholds": [],
"timeRegions": [],
"title": "Volume Space Usage",
"tooltip": {
"shared": false,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"format": "bytes",
"logBase": 1,
"min": 0,
"show": true
},
{
"format": "bytes",
"logBase": 1,
"min": 0,
"show": true
}
],
"yaxis": {
"align": false
}
},
{
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "rgba(50, 172, 45, 0.97)",
"value": null
},
{
"color": "rgba(237, 129, 40, 0.89)",
"value": 80
},
{
"color": "rgba(245, 54, 54, 0.9)",
"value": 90
}
]
},
"unit": "percent"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 6,
"x": 18,
"y": 0
},
"id": 3,
"interval": "1m",
"links": [],
"maxDataPoints": 100,
"options": {
"orientation": "horizontal",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showThresholdLabels": false,
"showThresholdMarkers": true
},
"pluginVersion": "9.0.1",
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "max without(instance,node) (\n(\n topk(1, kubelet_volume_stats_capacity_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})\n -\n topk(1, kubelet_volume_stats_available_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})\n)\n/\ntopk(1, kubelet_volume_stats_capacity_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})\n* 100)\n",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "",
"refId": "A"
}
],
"title": "Volume Space Usage",
"type": "gauge"
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": {
"uid": "$datasource"
},
"fill": 1,
"fillGradient": 0,
"gridPos": {
"h": 7,
"w": 18,
"x": 0,
"y": 7
},
"hiddenSeries": false,
"id": 4,
"interval": "1m",
"legend": {
"alignAsTable": true,
"avg": true,
"current": true,
"max": true,
"min": true,
"rightSide": true,
"show": true,
"total": false,
"values": true
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null",
"options": {
"alertThreshold": true
},
"percentage": false,
"pluginVersion": "9.0.1",
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": true,
"steppedLine": false,
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "sum without(instance, node) (topk(1, (kubelet_volume_stats_inodes_used{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})))\n",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": "Used inodes",
"refId": "A"
},
{
"datasource": {
"uid": "$datasource"
},
"expr": "(\n sum without(instance, node) (topk(1, (kubelet_volume_stats_inodes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})))\n -\n sum without(instance, node) (topk(1, (kubelet_volume_stats_inodes_used{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})))\n)\n",
"format": "time_series",
"intervalFactor": 1,
"legendFormat": " Free inodes",
"refId": "B"
}
],
"thresholds": [],
"timeRegions": [],
"title": "Volume inodes Usage",
"tooltip": {
"shared": false,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"mode": "time",
"show": true,
"values": []
},
"yaxes": [
{
"format": "none",
"logBase": 1,
"min": 0,
"show": true
},
{
"format": "none",
"logBase": 1,
"min": 0,
"show": true
}
],
"yaxis": {
"align": false
}
},
{
"datasource": {
"uid": "$datasource"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "thresholds"
},
"mappings": [
{
"options": {
"match": "null",
"result": {
"text": "N/A"
}
},
"type": "special"
}
],
"max": 100,
"min": 0,
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "rgba(50, 172, 45, 0.97)",
"value": null
},
{
"color": "rgba(237, 129, 40, 0.89)",
"value": 80
},
{
"color": "rgba(245, 54, 54, 0.9)",
"value": 90
}
]
},
"unit": "percent"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 6,
"x": 18,
"y": 7
},
"id": 5,
"interval": "1m",
"links": [],
"maxDataPoints": 100,
"options": {
"orientation": "horizontal",
"reduceOptions": {
"calcs": [
"lastNotNull"
],
"fields": "",
"values": false
},
"showThresholdLabels": false,
"showThresholdMarkers": true
},
"pluginVersion": "9.0.1",
"targets": [
{
"datasource": {
"uid": "$datasource"
},
"expr": "max without(instance,node) (\ntopk(1, kubelet_volume_stats_inodes_used{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})\n/\ntopk(1, kubelet_volume_stats_inodes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\", persistentvolumeclaim=\"$volume\"})\n* 100)\n",
"format": "time_series",
"intervalFactor": 2,
"legendFormat": "",
"refId": "A"
}
],
"title": "Volume inodes Usage",
"type": "gauge"
}
],
"refresh": "10s",
"schemaVersion": 36,
"style": "dark",
"tags": [
"kubernetes-mixin"
],
"templating": {
"list": [
{
"current": {
"selected": false,
"text": "default",
"value": "default"
},
"hide": 0,
"includeAll": false,
"label": "Data Source",
"multi": false,
"name": "datasource",
"options": [],
"query": "prometheus",
"refresh": 1,
"regex": "",
"skipUrlSync": false,
"type": "datasource"
},
{
"current": {
"isNone": true,
"selected": false,
"text": "None",
"value": ""
},
"datasource": {
"type": "prometheus",
"uid": "$datasource"
},
"definition": "",
"hide": 2,
"includeAll": false,
"label": "cluster",
"multi": false,
"name": "cluster",
"options": [],
"query": {
"query": "label_values(kubelet_volume_stats_capacity_bytes{job=\"kubelet\", metrics_path=\"/metrics\"}, cluster)",
"refId": "Prometheus-cluster-Variable-Query"
},
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 1,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
},
{
"current": {
"isNone": true,
"selected": false,
"text": "None",
"value": ""
},
"datasource": {
"type": "prometheus",
"uid": "$datasource"
},
"definition": "",
"hide": 0,
"includeAll": false,
"label": "Namespace",
"multi": false,
"name": "namespace",
"options": [],
"query": {
"query": "label_values(kubelet_volume_stats_capacity_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\"}, namespace)",
"refId": "Prometheus-namespace-Variable-Query"
},
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 1,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
},
{
"current": {
"isNone": true,
"selected": false,
"text": "None",
"value": ""
},
"datasource": {
"type": "prometheus",
"uid": "$datasource"
},
"definition": "",
"hide": 0,
"includeAll": false,
"label": "PersistentVolumeClaim",
"multi": false,
"name": "volume",
"options": [],
"query": {
"query": "label_values(kubelet_volume_stats_capacity_bytes{cluster=\"$cluster\", job=\"kubelet\", metrics_path=\"/metrics\", namespace=\"$namespace\"}, persistentvolumeclaim)",
"refId": "Prometheus-volume-Variable-Query"
},
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 1,
"tagValuesQuery": "",
"tagsQuery": "",
"type": "query",
"useTags": false
}
]
},
"time": {
"from": "now-7d",
"to": "now"
},
"timepicker": {
"refresh_intervals": [
"5s",
"10s",
"30s",
"1m",
"5m",
"15m",
"30m",
"1h",
"2h",
"1d"
],
"time_options": [
"5m",
"15m",
"1h",
"6h",
"12h",
"24h",
"2d",
"7d",
"30d"
]
},
"timezone": "utc",
"title": "Kubernetes / Persistent Volumes",
"uid": "919b92a8e8041bd567af9edab12c840c",
"version": 1,
"weekStart": ""
}
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+10
View File
@@ -2,6 +2,16 @@ locals {
name = "adot-collector-kubeprometheus"
namespace = try(var.config.helm_config.namespace, local.name)
}
terraform {
required_providers {
grafana = {
source = "grafana/grafana"
version = "1.24.0"
}
}
}
resource "helm_release" "kube_state_metrics" {
count = var.config.enable_kube_state_metrics ? 1 : 0
chart = var.config.ksm_helm_chart_name
@@ -24,6 +24,10 @@ variable "amp_region" {
default = null
}
variable "dashboards_folder_id" {
type = string
}
variable "addon_context" {
description = "Input configuration for the addon"
type = object({
@@ -64,11 +68,14 @@ variable "config" {
ne_helm_repo_url = string
ne_helm_settings = map(string)
ne_helm_values = map(any)
enable_dashboards = bool
})
default = {
enable_kube_state_metrics = true
enable_node_exporter = true
enable_dashboards = true
helm_config = {}
kms_create_namespace = true
@@ -1,14 +0,0 @@
terraform {
required_version = ">= 1.0.0"
required_providers {
aws = {
source = "hashicorp/aws"
version = ">= 3.72"
}
kubernetes = {
source = "hashicorp/kubernetes"
version = ">= 2.10"
}
}
}
+2 -24
View File
@@ -141,31 +141,9 @@ variable "infra_metrics_config" {
ne_helm_repo_url = string
ne_helm_settings = map(string)
ne_helm_values = map(any)
enable_dashboards = bool
})
default = null
# default = {
# enable_kube_state_metrics = true
# enable_node_exporter = true
# helm_config = {}
# kms_create_namespace = true
# ksm_helm_chart_name = "kube-state-metrics"
# ksm_helm_chart_version = "4.9.2"
# ksm_helm_release_name = "kube-state-metrics"
# ksm_helm_repo_url = "https://prometheus-community.github.io/helm-charts"
# ksm_helm_settings = {}
# ksm_helm_values = {}
# ksm_k8s_namespace = "kube-system"
# ne_create_namespace = true
# ne_k8s_namespace = "prometheus-node-exporter"
# ne_helm_chart_name = "prometheus-node-exporter"
# ne_helm_chart_version = "2.0.3"
# ne_helm_release_name = "prometheus-node-exporter"
# ne_helm_repo_url = "https://prometheus-community.github.io/helm-charts"
# ne_helm_settings = {}
# ne_helm_values = {}
# }
}