diff --git a/Makefile b/Makefile index 5d4b5c4..71ed8f9 100644 --- a/Makefile +++ b/Makefile @@ -46,3 +46,10 @@ ifeq (, $(shell which jsonnet)) @go get github.com/google/go-jsonnet/cmd/jsonnet @go get github.com/brancz/gojsontoyaml endif + +change_suffix: + @perl -p -i -e 's/^(\s*)\-\ host:.*/\1- host: alertmanager.${IP}.nip.io/g' manifests/ingress-alertmanager-main.yaml manifests/ingress-prometheus-k8s.yaml manifests/ingress-grafana.yaml + @echo "Ingress IPs changed to [service].${IP}.nip.io" + ${K3S} kubectl apply -f manifests/ingress-alertmanager-main.yaml + ${K3S} kubectl apply -f manifests/ingress-grafana.yaml + ${K3S} kubectl apply -f manifests/ingress-prometheus-k8s.yaml diff --git a/Readme.md b/Readme.md index d8a0b79..923f1b4 100644 --- a/Readme.md +++ b/Readme.md @@ -55,6 +55,32 @@ $ until kubectl get servicemonitors --all-namespaces ; do date; sleep 1; echo "" $ kubectl apply -f manifests/ # This command sometimes may need to be done twice (to workaround a race condition). ``` +If you get an error from applying the manifests, run the `make deploy` or `kubectl apply -f manifests/` again. Sometimes the resources required to apply the CRDs are not deployed yet. + +## Customizing for K3s + +To have your [K3s](https://github.com/rancher/k3s) cluster and the monitoring stack on it, deploy K3s with `curl -sfL https://get.k3s.io | sh -`. + +Now to deploy the monitoring stack on your K3s cluster, there are three parameters to be configured on `vars.jsonnet`: + +1. Set `k3s.enabled` to `true`. +2. Change your K3s master node IP(your VM or host IP) on `k3s.master_ip`. +3. Edit `suffixDomain` to have your node IP with the `.nip.io` suffix. This will be your ingress URL suffix. + +After changing these values, run `make` to build the manifests and `k3s kubectl apply -f manifests/` to apply the stack to your cluster. In case of errors on some resources, re-run the command. + +Now you can open the applications: + +* Grafana on [https://grafana.[your_node_ip].nip.io](https://grafana.[your_node_ip].nip.io), +* Prometheus on [https://prometheus.[your_node_ip].nip.io](https://prometheus.[your_node_ip].nip.io) +* Alertmanager on [https://alertmanager.[your_node_ip].nip.io](https://alertmanager.[your_node_ip].nip.io) + +There are some dashboards that shows no values due to some cadvisor metrics not having the complete metadata. Check the open issues for more information. + +## Updating the ingress suffixes + +To avoid rebuilding all manifests, there is a make target to update the Ingress URL suffix to a different suffix (using nip.io) to match your host IP. Run `make change_suffix IP="[IP-ADDRESS]"` to change the ingress route IP for Grafana, Prometheus and Alertmanager and reapply the manifests. If you have a K3s cluster, run `make change_suffix IP="[IP-ADDRESS] K3S=k3s`. + ## Customizing The content of this project consists of a set of jsonnet files making up a library to be consumed. diff --git a/jsonnetfile.lock.json b/jsonnetfile.lock.json index b292f34..9514157 100644 --- a/jsonnetfile.lock.json +++ b/jsonnetfile.lock.json @@ -8,7 +8,7 @@ "subdir": "jsonnet/kube-prometheus" } }, - "version": "4adb70b017e9a4ecb884a636dfef6fcae7d4bed8" + "version": "da959c643657c7d2aac6f5ddd68582a949283c49" }, { "name": "ksonnet", @@ -28,7 +28,7 @@ "subdir": "" } }, - "version": "0afc72e70df6048c6b65fd3e4968e53b0812b30c" + "version": "193d4934f85c9ff596d1f3e4dce7bd2da62a4d5e" }, { "name": "grafonnet", @@ -48,7 +48,7 @@ "subdir": "grafana-builder" } }, - "version": "3c44dfa9bfe2b66985733d4b16e0afd29094b4a0" + "version": "565bf6b51d636e0efe4add39f2ab8e2b1abb731f" }, { "name": "grafana", @@ -58,7 +58,7 @@ "subdir": "grafana" } }, - "version": "c27d2792764867cdaf6484f067cc875cb8aef2f6" + "version": "7fadaf2274d5cbe4ac6fbaf8786e4b7ecf3c1713" }, { "name": "prometheus-operator", @@ -78,7 +78,7 @@ "subdir": "Documentation/etcd-mixin" } }, - "version": "43ce2eefaa0a4bdd5c1e825ff08a32e6e46f3343" + "version": "8037e6e08727d4a17649f782cb4dbc482b8fe780" }, { "name": "prometheus", @@ -88,7 +88,7 @@ "subdir": "documentation/prometheus-mixin" } }, - "version": "ff40de7ca6084f5aab1f3971025c00c217615589" + "version": "f0bb8129c3e6ffc6906bdc130f5625110643f168" } ] } diff --git a/k3s-overrides.jsonnet b/k3s-overrides.jsonnet new file mode 100644 index 0000000..8d8672e --- /dev/null +++ b/k3s-overrides.jsonnet @@ -0,0 +1,250 @@ +local k = import 'ksonnet/ksonnet.beta.4/k.libsonnet'; +local vars = import 'vars.jsonnet'; +local service = k.core.v1.service; +local servicePort = k.core.v1.service.mixin.spec.portsType; + +{ + prometheus+:: { + kubeControllerManagerPrometheusDiscoveryService: + service.new('kube-controller-manager-prometheus-discovery', { 'k8s-app': 'kube-controller-manager' }, servicePort.newNamed('http-metrics', 10252, 10252)) + + service.mixin.metadata.withNamespace('kube-system') + + service.mixin.metadata.withLabels({ 'k8s-app': 'kube-controller-manager' }) + + service.mixin.spec.withClusterIp('None'), + kubeControllerManagerPrometheusDiscoveryEndpoints: + local endpoints = k.core.v1.endpoints; + local endpointSubset = endpoints.subsetsType; + local endpointPort = endpointSubset.portsType; + + local Port = endpointPort.new() + + endpointPort.withName('http-metrics') + + endpointPort.withPort(10252) + + endpointPort.withProtocol('TCP'); + + local subset = endpointSubset.new() + + endpointSubset.withAddresses([ + { ip: vars.k3s.master_ip }]) + + endpointSubset.withPorts(Port); + + endpoints.new() + + endpoints.mixin.metadata.withName('kube-controller-manager-prometheus-discovery') + + endpoints.mixin.metadata.withNamespace('kube-system') + + endpoints.mixin.metadata.withLabels({ 'k8s-app': 'kube-controller-manager' }) + + endpoints.withSubsets(subset), + + kubeSchedulerPrometheusDiscoveryService: + service.new('kube-scheduler-prometheus-discovery', { 'k8s-app': 'kube-scheduler' }, servicePort.newNamed('http-metrics', 10251, 10251)) + + service.mixin.metadata.withNamespace('kube-system') + + service.mixin.metadata.withLabels({ 'k8s-app': 'kube-scheduler' }) + + service.mixin.spec.withClusterIp('None'), + + kubeSchedulerPrometheusDiscoveryEndpoints: + local endpoints = k.core.v1.endpoints; + local endpointSubset = endpoints.subsetsType; + local endpointPort = endpointSubset.portsType; + + local Port = endpointPort.new() + + endpointPort.withName('http-metrics') + + endpointPort.withPort(10251) + + endpointPort.withProtocol('TCP'); + + local subset = endpointSubset.new() + + endpointSubset.withAddresses([ + { ip: vars.k3s.master_ip }]) + + endpointSubset.withPorts(Port); + + endpoints.new() + + endpoints.mixin.metadata.withName('kube-scheduler-prometheus-discovery') + + endpoints.mixin.metadata.withNamespace('kube-system') + + endpoints.mixin.metadata.withLabels({ 'k8s-app': 'kube-scheduler' }) + + endpoints.withSubsets(subset), + + serviceMonitorKubelet+: + { + spec+: { + endpoints: [ + { + port: 'https-metrics', + scheme: 'https', + interval: '30s', + honorLabels: true, + tlsConfig: { + insecureSkipVerify: true, + }, + bearerTokenFile: '/var/run/secrets/kubernetes.io/serviceaccount/token', + }, + { + port: 'https-metrics', + scheme: 'https', + path: '/metrics/cadvisor', + interval: '30s', + honorLabels: true, + tlsConfig: { + insecureSkipVerify: true, + }, + bearerTokenFile: '/var/run/secrets/kubernetes.io/serviceaccount/token', + metricRelabelings: [ + // Drop a bunch of metrics which are disabled but still sent, see + // https://github.com/google/cadvisor/issues/1925. + { + sourceLabels: ['__name__'], + regex: 'container_(network_tcp_usage_total|network_udp_usage_total|tasks_state|cpu_load_average_10s)', + action: 'drop', + }, + ], + }, + ], + }, + }, + }, + + nodeExporter+:: { + daemonset+: { + spec+: { + template+: { + spec+: { + containers: + std.filterMap( + function(c) std.startsWith(c.name, 'kube-rbac') != true, + function(c) + if std.startsWith(c.name, 'node-exporter') then + c { + args: [ + '--web.listen-address=:' + $._config.nodeExporter.port, + '--path.procfs=/host/proc', + '--path.sysfs=/host/sys', + '--path.rootfs=/host/root', + // The following settings have been taken from + // https://github.com/prometheus/node_exporter/blob/0662673/collector/filesystem_linux.go#L30-L31 + // Once node exporter is being released with those settings, this can be removed. + '--collector.filesystem.ignored-mount-points=^/(dev|proc|sys|var/lib/docker/.+)($|/)', + '--collector.filesystem.ignored-fs-types=^(autofs|binfmt_misc|cgroup|configfs|debugfs|devpts|devtmpfs|fusectl|hugetlbfs|mqueue|overlay|proc|procfs|pstore|rpc_pipefs|securityfs|sysfs|tracefs)$', + ], + ports: [ + { + containerPort: 9100, + name: 'http' + }], + + } + else + c, + super.containers, + ), + }, + }, + }, + }, + + service+: + { + spec+: { + ports: [{ + name: 'http', + port: 9100, + targetPort: 'http' + }] + } + }, + + serviceMonitor+: + { + spec+: { + endpoints: [ + { + port: 'http', + scheme: 'http', + interval: '30s', + relabelings: [ + { + action: 'replace', + regex: '(.*)', + replacment: '$1', + sourceLabels: ['__meta_kubernetes_pod_node_name'], + targetLabel: 'instance', + }, + ], + }, + ], + }, + }, + }, + + + kubeStateMetrics+:: { + deployment+: { + spec+: { + template+: { + spec+: { + containers: + std.filterMap( + function(c) std.startsWith(c.name, 'kube-rbac') != true, + function(c) + if std.startsWith(c.name, 'kube-state-metrics') then + c { + args: [ + '--port=8080', + '--telemetry-port=8081', + ], + ports: [ + { + containerPort: 8080, + name: 'http-main' + }, + { + containerPort: 8081, + name: 'http-self' + }], + } + else + c, + super.containers, + ), + }, + }, + }, + }, + + service+: + { + spec+: { + ports: [{ + name: 'http-main', + port: 8080, + targetPort: 'http-main' + }, + { + name: 'http-self', + port: 8081, + targetPort: 'http-self' + }] + } + }, + + serviceMonitor+: + { + spec+: { + endpoints: [ + { + port: 'http-main', + scheme: 'http', + interval: $._config.kubeStateMetrics.scrapeInterval, + scrapeTimeout: $._config.kubeStateMetrics.scrapeTimeout, + honorLabels: true, + tlsConfig: { + insecureSkipVerify: true, + }, + }, + { + port: 'http-self', + scheme: 'http', + interval: '30s', + tlsConfig: { + insecureSkipVerify: true, + }, + }, + ], + }, + }, + }, + +} \ No newline at end of file diff --git a/main.jsonnet b/main.jsonnet index 5d0b23f..094a403 100644 --- a/main.jsonnet +++ b/main.jsonnet @@ -15,13 +15,17 @@ local kp = (import 'kube-prometheus/kube-prometheus.libsonnet') + (import 'kube-prometheus/kube-prometheus-kubeadm.libsonnet') // Use http Kubelet targets. Comment to revert to https + (import 'kube-prometheus/kube-prometheus-insecure-kubelet.libsonnet') - + (import 'base_operator_stack.jsonnet') + (import 'smtp_server.jsonnet') // Additional modules are loaded dynamically from vars.jsonnet + join_objects([module.file for module in vars.modules if module.enabled]) + // Load K3s customized modules + + join_objects([m for m in [import 'k3s-overrides.jsonnet'] if vars.k3s.enabled]) + // Base stack is loaded at the end to override previous definitions + + (import 'base_operator_stack.jsonnet') // Load image versions last to override default from modules + (import 'image_sources_versions.jsonnet'); + // Generate core modules { ['00namespace-' + name]: kp.kubePrometheus[name] for name in std.objectFields(kp.kubePrometheus) } { ['0prometheus-operator-' + name]: kp.prometheusOperator[name] for name in std.objectFields(kp.prometheusOperator) } diff --git a/manifests/grafana-dashboardDefinitions.yaml b/manifests/grafana-dashboardDefinitions.yaml index f61c7ac..48bcf8b 100644 --- a/manifests/grafana-dashboardDefinitions.yaml +++ b/manifests/grafana-dashboardDefinitions.yaml @@ -3883,1919 +3883,6 @@ items: metadata: name: grafana-dashboard-coredns-dashboard namespace: monitoring -- apiVersion: v1 - data: - k8s-cluster-rsrc-use.json: |- - { - "annotations": { - "list": [ - - ] - }, - "editable": true, - "gnetId": null, - "graphTooltip": 0, - "hideControls": false, - "links": [ - - ], - "refresh": "10s", - "rows": [ - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 1, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:cluster_cpu_utilisation:ratio{cluster=\"$cluster\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "CPU Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": 1, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 2, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_cpu_saturation_load1:{cluster=\"$cluster\"} / scalar(sum(min(kube_pod_info{cluster=\"$cluster\"}) by (node)))", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "CPU Saturation (Load1)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": 1, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "CPU", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 3, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:cluster_memory_utilisation:ratio{cluster=\"$cluster\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Memory Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": 1, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 4, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_memory_swap_io_bytes:sum_rate{cluster=\"$cluster\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Memory Saturation (Swap I/O)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "Bps", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Memory", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 5, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_disk_utilisation:avg_irate{cluster=\"$cluster\"} / scalar(:kube_pod_info_node_count:{cluster=\"$cluster\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk IO Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": 1, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 6, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_disk_saturation:avg_irate{cluster=\"$cluster\"} / scalar(:kube_pod_info_node_count:{cluster=\"$cluster\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk IO Saturation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": 1, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Disk", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 7, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_net_utilisation:sum_irate{cluster=\"$cluster\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Net Utilisation (Transmitted)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "Bps", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 8, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_net_saturation:sum_irate{cluster=\"$cluster\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Net Saturation (Dropped)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "Bps", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Network", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 10, - "id": 9, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 0, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 12, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "sum(max(node_filesystem_size_bytes{fstype=~\"ext[234]|btrfs|xfs|zfs\", cluster=\"$cluster\"} - node_filesystem_avail_bytes{fstype=~\"ext[234]|btrfs|xfs|zfs\", cluster=\"$cluster\"}) by (device,pod,namespace)) by (pod,namespace)\n/ scalar(sum(max(node_filesystem_size_bytes{fstype=~\"ext[234]|btrfs|xfs|zfs\", cluster=\"$cluster\"}) by (device,pod,namespace)))\n* on (namespace, pod) group_left (node) node_namespace_pod:kube_pod_info:{cluster=\"$cluster\"}\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{node}}", - "legendLink": "/d/4ac4f123aae0ff6dbaf4f4f66120033b/k8s-node-rsrc-use", - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk Capacity", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": 1, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Storage", - "titleSize": "h6" - } - ], - "schemaVersion": 14, - "style": "dark", - "tags": [ - "kubernetes-mixin" - ], - "templating": { - "list": [ - { - "current": { - "text": "Prometheus", - "value": "Prometheus" - }, - "hide": 0, - "label": null, - "name": "datasource", - "options": [ - - ], - "query": "prometheus", - "refresh": 1, - "regex": "", - "type": "datasource" - }, - { - "allValue": null, - "current": { - "text": "prod", - "value": "prod" - }, - "datasource": "$datasource", - "hide": 2, - "includeAll": false, - "label": "cluster", - "multi": false, - "name": "cluster", - "options": [ - - ], - "query": "label_values(kube_node_info, cluster)", - "refresh": 1, - "regex": "", - "sort": 2, - "tagValuesQuery": "", - "tags": [ - - ], - "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": "", - "title": "Kubernetes / USE Method / Cluster", - "uid": "a6e7d1362e1ddbb79db21d5bb40d7137", - "version": 0 - } - kind: ConfigMap - metadata: - name: grafana-dashboard-k8s-cluster-rsrc-use - namespace: monitoring -- apiVersion: v1 - data: - k8s-node-rsrc-use.json: |- - { - "annotations": { - "list": [ - - ] - }, - "editable": true, - "gnetId": null, - "graphTooltip": 0, - "hideControls": false, - "links": [ - - ], - "refresh": "10s", - "rows": [ - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 1, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_cpu_utilisation:avg1m{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Utilisation", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "CPU Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 2, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_cpu_saturation_load1:{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Saturation", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "CPU Saturation (Load1)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "CPU", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 3, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_memory_utilisation:{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Memory", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Memory Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 4, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_memory_swap_io_bytes:sum_rate{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Swap IO", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Memory Saturation (Swap I/O)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "Bps", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Memory", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 5, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_disk_utilisation:avg_irate{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Utilisation", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk IO Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 6, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_disk_saturation:avg_irate{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Saturation", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk IO Saturation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Disk", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 7, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_net_utilisation:sum_irate{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Utilisation", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Net Utilisation (Transmitted)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "Bps", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 8, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_net_saturation:sum_irate{cluster=\"$cluster\", node=\"$node\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "Saturation", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Net Saturation (Dropped)", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "Bps", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Net", - "titleSize": "h6" - }, - { - "collapse": false, - "height": "250px", - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "id": 9, - "legend": { - "avg": false, - "current": false, - "max": false, - "min": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null as zero", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 12, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_filesystem_usage:{cluster=\"$cluster\"}\n* on (namespace, pod) group_left (node) node_namespace_pod:kube_pod_info:{cluster=\"$cluster\", node=\"$node\"}\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{device}}", - "legendLink": null, - "step": 10 - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk Utilisation", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": 0, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": false - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": true, - "title": "Disk", - "titleSize": "h6" - } - ], - "schemaVersion": 14, - "style": "dark", - "tags": [ - "kubernetes-mixin" - ], - "templating": { - "list": [ - { - "current": { - "text": "Prometheus", - "value": "Prometheus" - }, - "hide": 0, - "label": null, - "name": "datasource", - "options": [ - - ], - "query": "prometheus", - "refresh": 1, - "regex": "", - "type": "datasource" - }, - { - "allValue": null, - "current": { - "text": "prod", - "value": "prod" - }, - "datasource": "$datasource", - "hide": 2, - "includeAll": false, - "label": "cluster", - "multi": false, - "name": "cluster", - "options": [ - - ], - "query": "label_values(kube_node_info, cluster)", - "refresh": 1, - "regex": "", - "sort": 2, - "tagValuesQuery": "", - "tags": [ - - ], - "tagsQuery": "", - "type": "query", - "useTags": false - }, - { - "allValue": null, - "current": { - "text": "prod", - "value": "prod" - }, - "datasource": "$datasource", - "hide": 0, - "includeAll": false, - "label": "node", - "multi": false, - "name": "node", - "options": [ - - ], - "query": "label_values(kube_node_info{cluster=\"$cluster\"}, node)", - "refresh": 1, - "regex": "", - "sort": 2, - "tagValuesQuery": "", - "tags": [ - - ], - "tagsQuery": "", - "type": "query", - "useTags": false - } - ] - }, - "time": { - "from": "now-1h", - "to": "now" - }, - "timepicker": { - "refresh_intervals": [ - "5s", - "10s", - "30s", - "1m", - "5m", - "15m", - "30m", - "1h", - "2h", - "1d" - ], - "time_options": [ - "5m", - "15m", - "1h", - "6h", - "12h", - "24h", - "2d", - "7d", - "30d" - ] - }, - "timezone": "", - "title": "Kubernetes / USE Method / Node", - "uid": "4ac4f123aae0ff6dbaf4f4f66120033b", - "version": 0 - } - kind: ConfigMap - metadata: - name: grafana-dashboard-k8s-node-rsrc-use - namespace: monitoring - apiVersion: v1 data: k8s-resources-cluster.json: |- @@ -7684,7 +5771,7 @@ items: "steppedLine": false, "targets": [ { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}) by (pod)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}) by (pod)", "format": "time_series", "intervalFactor": 2, "legendFormat": "{{pod}}", @@ -7967,7 +6054,7 @@ items: ], "targets": [ { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\",container!=\"\"}) by (pod)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\",container!=\"\"}) by (pod)", "format": "table", "instant": true, "intervalFactor": 2, @@ -7985,7 +6072,7 @@ items: "step": 10 }, { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\",container!=\"\"}) by (pod) / sum(kube_pod_container_resource_requests_memory_bytes{namespace=\"$namespace\"}) by (pod)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\",container!=\"\"}) by (pod) / sum(kube_pod_container_resource_requests_memory_bytes{namespace=\"$namespace\"}) by (pod)", "format": "table", "instant": true, "intervalFactor": 2, @@ -8003,7 +6090,7 @@ items: "step": 10 }, { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\",container!=\"\"}) by (pod) / sum(kube_pod_container_resource_limits_memory_bytes{namespace=\"$namespace\"}) by (pod)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\",container!=\"\"}) by (pod) / sum(kube_pod_container_resource_limits_memory_bytes{namespace=\"$namespace\"}) by (pod)", "format": "table", "instant": true, "intervalFactor": 2, @@ -8930,7 +7017,7 @@ items: ], "targets": [ { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", pod=\"$pod\", container!=\"POD\", container!=\"\"}) by (container)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", pod=\"$pod\", container!=\"POD\", container!=\"\"}) by (container)", "format": "table", "instant": true, "intervalFactor": 2, @@ -8948,7 +7035,7 @@ items: "step": 10 }, { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", pod=\"$pod\"}) by (container) / sum(kube_pod_container_resource_requests_memory_bytes{namespace=\"$namespace\", pod=\"$pod\"}) by (container)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", pod=\"$pod\"}) by (container) / sum(kube_pod_container_resource_requests_memory_bytes{namespace=\"$namespace\", pod=\"$pod\"}) by (container)", "format": "table", "instant": true, "intervalFactor": 2, @@ -8966,7 +7053,7 @@ items: "step": 10 }, { - "expr": "sum(container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", pod=\"$pod\", container!=\"\"}) by (container) / sum(kube_pod_container_resource_limits_memory_bytes{namespace=\"$namespace\", pod=\"$pod\"}) by (container)", + "expr": "sum(container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", pod=\"$pod\", container!=\"\"}) by (container) / sum(kube_pod_container_resource_limits_memory_bytes{namespace=\"$namespace\", pod=\"$pod\"}) by (container)", "format": "table", "instant": true, "intervalFactor": 2, @@ -9621,7 +7708,7 @@ items: "steppedLine": false, "targets": [ { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", "format": "time_series", "intervalFactor": 2, "legendFormat": "{{pod}}", @@ -9850,7 +7937,7 @@ items: ], "targets": [ { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", "format": "table", "instant": true, "intervalFactor": 2, @@ -9868,7 +7955,7 @@ items: "step": 10 }, { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n/sum(\n kube_pod_container_resource_requests_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n/sum(\n kube_pod_container_resource_requests_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", "format": "table", "instant": true, "intervalFactor": 2, @@ -9886,7 +7973,7 @@ items: "step": 10 }, { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n/sum(\n kube_pod_container_resource_limits_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n/sum(\n kube_pod_container_resource_limits_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\", workload=\"$workload\", workload_type=\"$type\"}\n) by (pod)\n", "format": "table", "instant": true, "intervalFactor": 2, @@ -10586,7 +8673,7 @@ items: "steppedLine": false, "targets": [ { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", "format": "time_series", "intervalFactor": 2, "legendFormat": "{{workload}} - {{workload_type}}", @@ -10860,7 +8947,7 @@ items: "step": 10 }, { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", "format": "table", "instant": true, "intervalFactor": 2, @@ -10878,7 +8965,7 @@ items: "step": 10 }, { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n/sum(\n kube_pod_container_resource_requests_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n* on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n/sum(\n kube_pod_container_resource_requests_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n* on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", "format": "table", "instant": true, "intervalFactor": 2, @@ -10896,7 +8983,7 @@ items: "step": 10 }, { - "expr": "sum(\n container_memory_usage_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n/sum(\n kube_pod_container_resource_limits_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n* on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", + "expr": "sum(\n container_memory_working_set_bytes{cluster=\"$cluster\", namespace=\"$namespace\", container!=\"\"}\n * on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n/sum(\n kube_pod_container_resource_limits_memory_bytes{cluster=\"$cluster\", namespace=\"$namespace\"}\n* on(namespace,pod)\n group_left(workload, workload_type) mixin_pod_workload{cluster=\"$cluster\", namespace=\"$namespace\"}\n) by (workload, workload_type)\n", "format": "table", "instant": true, "intervalFactor": 2, @@ -15380,1373 +13467,6 @@ items: metadata: name: grafana-dashboard-kubernetes-cluster-dashboard namespace: monitoring -- apiVersion: v1 - data: - nodes.json: |- - { - "__inputs": [ - - ], - "__requires": [ - - ], - "annotations": { - "list": [ - - ] - }, - "editable": false, - "gnetId": null, - "graphTooltip": 0, - "hideControls": false, - "id": null, - "links": [ - - ], - "refresh": "", - "rows": [ - { - "collapse": false, - "collapsed": false, - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 2, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "max(node_load1{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "load 1m", - "refId": "A" - }, - { - "expr": "max(node_load5{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "load 5m", - "refId": "B" - }, - { - "expr": "max(node_load15{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "load 15m", - "refId": "C" - }, - { - "expr": "count(node_cpu_seconds_total{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\", mode=\"user\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "logical cores", - "refId": "D" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "System load", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 3, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "sum by (cpu) (irate(node_cpu_seconds_total{cluster=\"$cluster\", job=\"node-exporter\", mode!=\"idle\", instance=\"$instance\"}[5m]))", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{cpu}}", - "refId": "A" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Usage Per Core", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": false, - "title": "Dashboard Row", - "titleSize": "h6", - "type": "row" - }, - { - "collapse": false, - "collapsed": false, - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 4, - "legend": { - "alignAsTable": "true", - "avg": "true", - "current": "true", - "max": "false", - "min": "false", - "rightSide": "true", - "show": "true", - "total": "false", - "values": "true" - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 9, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "max (sum by (cpu) (irate(node_cpu_seconds_total{cluster=\"$cluster\", job=\"node-exporter\", mode!=\"idle\", instance=\"$instance\"}[2m])) ) * 100\n", - "format": "time_series", - "intervalFactor": 10, - "legendFormat": "{{ cpu }}", - "refId": "A" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "CPU Utilization", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percent", - "label": null, - "logBase": 1, - "max": 100, - "min": 0, - "show": true - }, - { - "format": "percent", - "label": null, - "logBase": 1, - "max": 100, - "min": 0, - "show": true - } - ] - }, - { - "cacheTimeout": null, - "colorBackground": false, - "colorValue": false, - "colors": [ - "rgba(50, 172, 45, 0.97)", - "rgba(237, 129, 40, 0.89)", - "rgba(245, 54, 54, 0.9)" - ], - "datasource": "$datasource", - "format": "percent", - "gauge": { - "maxValue": 100, - "minValue": 0, - "show": true, - "thresholdLabels": false, - "thresholdMarkers": true - }, - "gridPos": { - - }, - "id": 5, - "interval": null, - "links": [ - - ], - "mappingType": 1, - "mappingTypes": [ - { - "name": "value to text", - "value": 1 - }, - { - "name": "range to text", - "value": 2 - } - ], - "maxDataPoints": 100, - "nullPointMode": "connected", - "nullText": null, - "postfix": "", - "postfixFontSize": "50%", - "prefix": "", - "prefixFontSize": "50%", - "rangeMaps": [ - { - "from": "null", - "text": "N/A", - "to": "null" - } - ], - "span": 3, - "sparkline": { - "fillColor": "rgba(31, 118, 189, 0.18)", - "full": false, - "lineColor": "rgb(31, 120, 193)", - "show": false - }, - "tableColumn": "", - "targets": [ - { - "expr": "avg(sum by (cpu) (irate(node_cpu_seconds_total{cluster=\"$cluster\", job=\"node-exporter\", mode!=\"idle\", instance=\"$instance\"}[2m]))) * 100\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "", - "refId": "A" - } - ], - "thresholds": "80, 90", - "title": "CPU Usage", - "tooltip": { - "shared": false - }, - "type": "singlestat", - "valueFontSize": "80%", - "valueMaps": [ - { - "op": "=", - "text": "N/A", - "value": "null" - } - ], - "valueName": "current" - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": false, - "title": "Dashboard Row", - "titleSize": "h6", - "type": "row" - }, - { - "collapse": false, - "collapsed": false, - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 6, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 9, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "max(\n node_memory_MemTotal_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_memory_MemFree_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_memory_Buffers_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_memory_Cached_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n)\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "memory used", - "refId": "A" - }, - { - "expr": "max(node_memory_Buffers_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "memory buffers", - "refId": "B" - }, - { - "expr": "max(node_memory_Cached_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "memory cached", - "refId": "C" - }, - { - "expr": "max(node_memory_MemFree_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "memory free", - "refId": "D" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Memory Usage", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - }, - { - "cacheTimeout": null, - "colorBackground": false, - "colorValue": false, - "colors": [ - "rgba(50, 172, 45, 0.97)", - "rgba(237, 129, 40, 0.89)", - "rgba(245, 54, 54, 0.9)" - ], - "datasource": "$datasource", - "format": "percent", - "gauge": { - "maxValue": 100, - "minValue": 0, - "show": true, - "thresholdLabels": false, - "thresholdMarkers": true - }, - "gridPos": { - - }, - "id": 7, - "interval": null, - "links": [ - - ], - "mappingType": 1, - "mappingTypes": [ - { - "name": "value to text", - "value": 1 - }, - { - "name": "range to text", - "value": 2 - } - ], - "maxDataPoints": 100, - "nullPointMode": "connected", - "nullText": null, - "postfix": "", - "postfixFontSize": "50%", - "prefix": "", - "prefixFontSize": "50%", - "rangeMaps": [ - { - "from": "null", - "text": "N/A", - "to": "null" - } - ], - "span": 3, - "sparkline": { - "fillColor": "rgba(31, 118, 189, 0.18)", - "full": false, - "lineColor": "rgb(31, 120, 193)", - "show": false - }, - "tableColumn": "", - "targets": [ - { - "expr": "max(\n (\n (\n node_memory_MemTotal_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_memory_MemFree_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_memory_Buffers_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_memory_Cached_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n )\n / node_memory_MemTotal_bytes{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n ) * 100)\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "", - "refId": "A" - } - ], - "thresholds": "80, 90", - "title": "Memory Usage", - "tooltip": { - "shared": false - }, - "type": "singlestat", - "valueFontSize": "80%", - "valueMaps": [ - { - "op": "=", - "text": "N/A", - "value": "null" - } - ], - "valueName": "current" - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": false, - "title": "Dashboard Row", - "titleSize": "h6", - "type": "row" - }, - { - "collapse": false, - "collapsed": false, - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 8, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - { - "alias": "read", - "yaxis": 1 - }, - { - "alias": "io time", - "yaxis": 2 - } - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "max(rate(node_disk_read_bytes_total{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}[2m]))", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "read", - "refId": "A" - }, - { - "expr": "max(rate(node_disk_written_bytes_total{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}[2m]))", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "written", - "refId": "B" - }, - { - "expr": "max(rate(node_disk_io_time_seconds_total{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}[2m]))", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "io time", - "refId": "C" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk I/O", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "s", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 9, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "node:node_filesystem_usage:{cluster=\"$cluster\", instance=\"$instance\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{device}} disk used", - "refId": "A" - }, - { - "expr": "node:node_filesystem_usage:{cluster=\"$cluster\", instance=\"$instance\"}", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{device}} disk free", - "refId": "B" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Disk Space Usage", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "percentunit", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": false, - "title": "Dashboard Row", - "titleSize": "h6", - "type": "row" - }, - { - "collapse": false, - "collapsed": false, - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 10, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "rate(node_network_receive_bytes_total{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\", device!~\"lo\"}[5m])", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{device}}", - "refId": "A" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Network Received", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - }, - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 11, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 6, - "stack": true, - "steppedLine": false, - "targets": [ - { - "expr": "rate(node_network_transmit_bytes_total{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\", device!~\"lo\"}[5m])", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "{{device}}", - "refId": "A" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Network Transmitted", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "bytes", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": false, - "title": "Dashboard Row", - "titleSize": "h6", - "type": "row" - }, - { - "collapse": false, - "collapsed": false, - "panels": [ - { - "aliasColors": { - - }, - "bars": false, - "dashLength": 10, - "dashes": false, - "datasource": "$datasource", - "fill": 1, - "gridPos": { - - }, - "id": 12, - "legend": { - "alignAsTable": false, - "avg": false, - "current": false, - "max": false, - "min": false, - "rightSide": false, - "show": true, - "total": false, - "values": false - }, - "lines": true, - "linewidth": 1, - "links": [ - - ], - "nullPointMode": "null", - "percentage": false, - "pointradius": 5, - "points": false, - "renderer": "flot", - "repeat": null, - "seriesOverrides": [ - - ], - "spaceLength": 10, - "span": 9, - "stack": false, - "steppedLine": false, - "targets": [ - { - "expr": "max(\n node_filesystem_files{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_filesystem_files_free{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n)\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "inodes used", - "refId": "A" - }, - { - "expr": "max(node_filesystem_files_free{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"})", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "inodes free", - "refId": "B" - } - ], - "thresholds": [ - - ], - "timeFrom": null, - "timeShift": null, - "title": "Inodes Usage", - "tooltip": { - "shared": false, - "sort": 0, - "value_type": "individual" - }, - "type": "graph", - "xaxis": { - "buckets": null, - "mode": "time", - "name": null, - "show": true, - "values": [ - - ] - }, - "yaxes": [ - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - }, - { - "format": "short", - "label": null, - "logBase": 1, - "max": null, - "min": null, - "show": true - } - ] - }, - { - "cacheTimeout": null, - "colorBackground": false, - "colorValue": false, - "colors": [ - "rgba(50, 172, 45, 0.97)", - "rgba(237, 129, 40, 0.89)", - "rgba(245, 54, 54, 0.9)" - ], - "datasource": "$datasource", - "format": "percent", - "gauge": { - "maxValue": 100, - "minValue": 0, - "show": true, - "thresholdLabels": false, - "thresholdMarkers": true - }, - "gridPos": { - - }, - "id": 13, - "interval": null, - "links": [ - - ], - "mappingType": 1, - "mappingTypes": [ - { - "name": "value to text", - "value": 1 - }, - { - "name": "range to text", - "value": 2 - } - ], - "maxDataPoints": 100, - "nullPointMode": "connected", - "nullText": null, - "postfix": "", - "postfixFontSize": "50%", - "prefix": "", - "prefixFontSize": "50%", - "rangeMaps": [ - { - "from": "null", - "text": "N/A", - "to": "null" - } - ], - "span": 3, - "sparkline": { - "fillColor": "rgba(31, 118, 189, 0.18)", - "full": false, - "lineColor": "rgb(31, 120, 193)", - "show": false - }, - "tableColumn": "", - "targets": [ - { - "expr": "max(\n (\n (\n node_filesystem_files{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n - node_filesystem_files_free{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n )\n / node_filesystem_files{cluster=\"$cluster\", job=\"node-exporter\", instance=\"$instance\"}\n ) * 100)\n", - "format": "time_series", - "intervalFactor": 2, - "legendFormat": "", - "refId": "A" - } - ], - "thresholds": "80, 90", - "title": "Inodes Usage", - "tooltip": { - "shared": false - }, - "type": "singlestat", - "valueFontSize": "80%", - "valueMaps": [ - { - "op": "=", - "text": "N/A", - "value": "null" - } - ], - "valueName": "current" - } - ], - "repeat": null, - "repeatIteration": null, - "repeatRowId": null, - "showTitle": false, - "title": "Dashboard Row", - "titleSize": "h6", - "type": "row" - } - ], - "schemaVersion": 14, - "style": "dark", - "tags": [ - "kubernetes-mixin" - ], - "templating": { - "list": [ - { - "current": { - "text": "Prometheus", - "value": "Prometheus" - }, - "hide": 0, - "label": null, - "name": "datasource", - "options": [ - - ], - "query": "prometheus", - "refresh": 1, - "regex": "", - "type": "datasource" - }, - { - "allValue": null, - "current": { - - }, - "datasource": "$datasource", - "hide": 2, - "includeAll": false, - "label": "cluster", - "multi": false, - "name": "cluster", - "options": [ - - ], - "query": "label_values(kube_pod_info, cluster)", - "refresh": 2, - "regex": "", - "sort": 0, - "tagValuesQuery": "", - "tags": [ - - ], - "tagsQuery": "", - "type": "query", - "useTags": false - }, - { - "allValue": null, - "current": { - - }, - "datasource": "$datasource", - "hide": 0, - "includeAll": false, - "label": null, - "multi": false, - "name": "instance", - "options": [ - - ], - "query": "label_values(node_boot_time_seconds{cluster=\"$cluster\", job=\"node-exporter\"}, instance)", - "refresh": 2, - "regex": "", - "sort": 0, - "tagValuesQuery": "", - "tags": [ - - ], - "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": "", - "title": "Kubernetes / Nodes", - "uid": "fa49a4706d07a042595b664c87fb33ea", - "version": 0 - } - kind: ConfigMap - metadata: - name: grafana-dashboard-nodes - namespace: monitoring - apiVersion: v1 data: persistentvolumesusage.json: |- diff --git a/manifests/grafana-deployment.yaml b/manifests/grafana-deployment.yaml index 225f654..a8c7959 100644 --- a/manifests/grafana-deployment.yaml +++ b/manifests/grafana-deployment.yaml @@ -1,4 +1,4 @@ -apiVersion: apps/v1beta2 +apiVersion: apps/v1 kind: Deployment metadata: labels: @@ -51,12 +51,6 @@ spec: - mountPath: /grafana-dashboard-definitions/0/coredns-dashboard name: grafana-dashboard-coredns-dashboard readOnly: false - - mountPath: /grafana-dashboard-definitions/0/k8s-cluster-rsrc-use - name: grafana-dashboard-k8s-cluster-rsrc-use - readOnly: false - - mountPath: /grafana-dashboard-definitions/0/k8s-node-rsrc-use - name: grafana-dashboard-k8s-node-rsrc-use - readOnly: false - mountPath: /grafana-dashboard-definitions/0/k8s-resources-cluster name: grafana-dashboard-k8s-resources-cluster readOnly: false @@ -78,9 +72,6 @@ spec: - mountPath: /grafana-dashboard-definitions/0/kubernetes-cluster-dashboard name: grafana-dashboard-kubernetes-cluster-dashboard readOnly: false - - mountPath: /grafana-dashboard-definitions/0/nodes - name: grafana-dashboard-nodes - readOnly: false - mountPath: /grafana-dashboard-definitions/0/persistentvolumesusage name: grafana-dashboard-persistentvolumesusage readOnly: false @@ -132,12 +123,6 @@ spec: - configMap: name: grafana-dashboard-coredns-dashboard name: grafana-dashboard-coredns-dashboard - - configMap: - name: grafana-dashboard-k8s-cluster-rsrc-use - name: grafana-dashboard-k8s-cluster-rsrc-use - - configMap: - name: grafana-dashboard-k8s-node-rsrc-use - name: grafana-dashboard-k8s-node-rsrc-use - configMap: name: grafana-dashboard-k8s-resources-cluster name: grafana-dashboard-k8s-resources-cluster @@ -159,9 +144,6 @@ spec: - configMap: name: grafana-dashboard-kubernetes-cluster-dashboard name: grafana-dashboard-kubernetes-cluster-dashboard - - configMap: - name: grafana-dashboard-nodes - name: grafana-dashboard-nodes - configMap: name: grafana-dashboard-persistentvolumesusage name: grafana-dashboard-persistentvolumesusage diff --git a/manifests/prometheus-rules.yaml b/manifests/prometheus-rules.yaml index f35510b..c7f4a03 100644 --- a/manifests/prometheus-rules.yaml +++ b/manifests/prometheus-rules.yaml @@ -153,177 +153,9 @@ spec: node_namespace_pod:kube_pod_info: )) record: node:node_num_cpu:sum - - expr: | - 1 - avg(rate(node_cpu_seconds_total{job="node-exporter",mode="idle"}[1m])) - record: :node_cpu_utilisation:avg1m - - expr: | - 1 - avg by (node) ( - rate(node_cpu_seconds_total{job="node-exporter",mode="idle"}[1m]) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info:) - record: node:node_cpu_utilisation:avg1m - - expr: | - node:node_cpu_utilisation:avg1m - * - node:node_num_cpu:sum - / - scalar(sum(node:node_num_cpu:sum)) - record: node:cluster_cpu_utilisation:ratio - - expr: | - sum(node_load1{job="node-exporter"}) - / - sum(node:node_num_cpu:sum) - record: ':node_cpu_saturation_load1:' - - expr: | - sum by (node) ( - node_load1{job="node-exporter"} - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - / - node:node_num_cpu:sum - record: 'node:node_cpu_saturation_load1:' - - expr: | - 1 - - sum(node_memory_MemFree_bytes{job="node-exporter"} + node_memory_Cached_bytes{job="node-exporter"} + node_memory_Buffers_bytes{job="node-exporter"}) - / - sum(node_memory_MemTotal_bytes{job="node-exporter"}) - record: ':node_memory_utilisation:' - expr: | sum(node_memory_MemFree_bytes{job="node-exporter"} + node_memory_Cached_bytes{job="node-exporter"} + node_memory_Buffers_bytes{job="node-exporter"}) record: :node_memory_MemFreeCachedBuffers_bytes:sum - - expr: | - sum(node_memory_MemTotal_bytes{job="node-exporter"}) - record: :node_memory_MemTotal_bytes:sum - - expr: | - sum by (node) ( - (node_memory_MemFree_bytes{job="node-exporter"} + node_memory_Cached_bytes{job="node-exporter"} + node_memory_Buffers_bytes{job="node-exporter"}) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_memory_bytes_available:sum - - expr: | - sum by (node) ( - node_memory_MemTotal_bytes{job="node-exporter"} - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_memory_bytes_total:sum - - expr: | - (node:node_memory_bytes_total:sum - node:node_memory_bytes_available:sum) - / - node:node_memory_bytes_total:sum - record: node:node_memory_utilisation:ratio - - expr: | - (node:node_memory_bytes_total:sum - node:node_memory_bytes_available:sum) - / - scalar(sum(node:node_memory_bytes_total:sum)) - record: node:cluster_memory_utilisation:ratio - - expr: | - 1e3 * sum( - (rate(node_vmstat_pgpgin{job="node-exporter"}[1m]) - + rate(node_vmstat_pgpgout{job="node-exporter"}[1m])) - ) - record: :node_memory_swap_io_bytes:sum_rate - - expr: | - 1 - - sum by (node) ( - (node_memory_MemFree_bytes{job="node-exporter"} + node_memory_Cached_bytes{job="node-exporter"} + node_memory_Buffers_bytes{job="node-exporter"}) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - / - sum by (node) ( - node_memory_MemTotal_bytes{job="node-exporter"} - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: 'node:node_memory_utilisation:' - - expr: | - 1 - (node:node_memory_bytes_available:sum / node:node_memory_bytes_total:sum) - record: 'node:node_memory_utilisation_2:' - - expr: | - 1e3 * sum by (node) ( - (rate(node_vmstat_pgpgin{job="node-exporter"}[1m]) - + rate(node_vmstat_pgpgout{job="node-exporter"}[1m])) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_memory_swap_io_bytes:sum_rate - - expr: | - avg(irate(node_disk_io_time_seconds_total{job="node-exporter",device=~"nvme.+|rbd.+|sd.+|vd.+|xvd.+|dm-.+"}[1m])) - record: :node_disk_utilisation:avg_irate - - expr: | - avg by (node) ( - irate(node_disk_io_time_seconds_total{job="node-exporter",device=~"nvme.+|rbd.+|sd.+|vd.+|xvd.+|dm-.+"}[1m]) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_disk_utilisation:avg_irate - - expr: | - avg(irate(node_disk_io_time_weighted_seconds_total{job="node-exporter",device=~"nvme.+|rbd.+|sd.+|vd.+|xvd.+|dm-.+"}[1m])) - record: :node_disk_saturation:avg_irate - - expr: | - avg by (node) ( - irate(node_disk_io_time_weighted_seconds_total{job="node-exporter",device=~"nvme.+|rbd.+|sd.+|vd.+|xvd.+|dm-.+"}[1m]) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_disk_saturation:avg_irate - - expr: | - max by (instance, namespace, pod, device) ((node_filesystem_size_bytes{fstype=~"ext[234]|btrfs|xfs|zfs"} - - node_filesystem_avail_bytes{fstype=~"ext[234]|btrfs|xfs|zfs"}) - / node_filesystem_size_bytes{fstype=~"ext[234]|btrfs|xfs|zfs"}) - record: 'node:node_filesystem_usage:' - - expr: | - max by (instance, namespace, pod, device) (node_filesystem_avail_bytes{fstype=~"ext[234]|btrfs|xfs|zfs"} / node_filesystem_size_bytes{fstype=~"ext[234]|btrfs|xfs|zfs"}) - record: 'node:node_filesystem_avail:' - - expr: | - sum(irate(node_network_receive_bytes_total{job="node-exporter",device!~"veth.+"}[1m])) + - sum(irate(node_network_transmit_bytes_total{job="node-exporter",device!~"veth.+"}[1m])) - record: :node_net_utilisation:sum_irate - - expr: | - sum by (node) ( - (irate(node_network_receive_bytes_total{job="node-exporter",device!~"veth.+"}[1m]) + - irate(node_network_transmit_bytes_total{job="node-exporter",device!~"veth.+"}[1m])) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_net_utilisation:sum_irate - - expr: | - sum(irate(node_network_receive_drop_total{job="node-exporter",device!~"veth.+"}[1m])) + - sum(irate(node_network_transmit_drop_total{job="node-exporter",device!~"veth.+"}[1m])) - record: :node_net_saturation:sum_irate - - expr: | - sum by (node) ( - (irate(node_network_receive_drop_total{job="node-exporter",device!~"veth.+"}[1m]) + - irate(node_network_transmit_drop_total{job="node-exporter",device!~"veth.+"}[1m])) - * on (namespace, pod) group_left(node) - node_namespace_pod:kube_pod_info: - ) - record: node:node_net_saturation:sum_irate - - expr: | - max( - max( - kube_pod_info{job="kube-state-metrics", host_ip!=""} - ) by (node, host_ip) - * on (host_ip) group_right (node) - label_replace( - (max(node_filesystem_files{job="node-exporter", mountpoint="/"}) by (instance)), "host_ip", "$1", "instance", "(.*):.*" - ) - ) by (node) - record: 'node:node_inodes_total:' - - expr: | - max( - max( - kube_pod_info{job="kube-state-metrics", host_ip!=""} - ) by (node, host_ip) - * on (host_ip) group_right (node) - label_replace( - (max(node_filesystem_files_free{job="node-exporter", mountpoint="/"}) by (instance)), "host_ip", "$1", "instance", "(.*):.*" - ) - ) by (node) - record: 'node:node_inodes_free:' - name: kube-prometheus-node-recording.rules rules: - expr: sum(rate(node_cpu_seconds_total{mode!="idle",mode!="iowait"}[3m])) BY @@ -446,17 +278,17 @@ spec: runbook_url: https://github.com/kubernetes-monitoring/kubernetes-mixin/tree/master/runbook.md#alert-name-kubepodcrashlooping expr: | rate(kube_pod_container_status_restarts_total{job="kube-state-metrics"}[15m]) * 60 * 5 > 0 - for: 1h + for: 15m labels: severity: critical - alert: KubePodNotReady annotations: message: Pod {{ $labels.namespace }}/{{ $labels.pod }} has been in a non-ready - state for longer than an hour. + state for longer than 15 minutes. runbook_url: https://github.com/kubernetes-monitoring/kubernetes-mixin/tree/master/runbook.md#alert-name-kubepodnotready expr: | sum by (namespace, pod) (kube_pod_status_phase{job="kube-state-metrics", phase=~"Failed|Pending|Unknown"}) > 0 - for: 1h + for: 15m labels: severity: critical - alert: KubeDeploymentGenerationMismatch @@ -475,13 +307,13 @@ spec: - alert: KubeDeploymentReplicasMismatch annotations: message: Deployment {{ $labels.namespace }}/{{ $labels.deployment }} has not - matched the expected number of replicas for longer than an hour. + matched the expected number of replicas for longer than 15 minutes. runbook_url: https://github.com/kubernetes-monitoring/kubernetes-mixin/tree/master/runbook.md#alert-name-kubedeploymentreplicasmismatch expr: | kube_deployment_spec_replicas{job="kube-state-metrics"} != kube_deployment_status_replicas_available{job="kube-state-metrics"} - for: 1h + for: 15m labels: severity: critical - alert: KubeStatefulSetReplicasMismatch @@ -589,7 +421,7 @@ spec: runbook_url: https://github.com/kubernetes-monitoring/kubernetes-mixin/tree/master/runbook.md#alert-name-kubejobfailed expr: | kube_job_status_failed{job="kube-state-metrics"} > 0 - for: 1h + for: 15m labels: severity: warning - name: kubernetes-resources @@ -723,7 +555,7 @@ spec: runbook_url: https://github.com/kubernetes-monitoring/kubernetes-mixin/tree/master/runbook.md#alert-name-kubenodenotready expr: | kube_node_status_condition{job="kube-state-metrics",condition="Ready",status="true"} == 0 - for: 1h + for: 15m labels: severity: warning - alert: KubeVersionMismatch @@ -733,7 +565,7 @@ spec: runbook_url: https://github.com/kubernetes-monitoring/kubernetes-mixin/tree/master/runbook.md#alert-name-kubeversionmismatch expr: | count(count by (gitVersion) (label_replace(kubernetes_build_info{job!~"kube-dns|coredns"},"gitVersion","$1","gitVersion","(v[0-9]*.[0-9]*.[0-9]*).*"))) > 1 - for: 1h + for: 15m labels: severity: warning - alert: KubeClientErrors @@ -949,17 +781,6 @@ spec: for: 4h labels: severity: warning - - alert: PrometheusTSDBWALCorruptions - annotations: - description: Prometheus {{$labels.namespace}}/{{$labels.pod}} has detected - {{$value | humanize}} corruptions of the write-ahead log (WAL) over the - last 3h. - summary: Prometheus is detecting WAL corruptions. - expr: | - increase(tsdb_wal_corruptions_total{job="prometheus-k8s",namespace="monitoring"}[3h]) > 0 - for: 4h - labels: - severity: warning - alert: PrometheusNotIngestingSamples annotations: description: Prometheus {{$labels.namespace}}/{{$labels.pod}} is not ingesting @@ -1054,7 +875,7 @@ spec: message: The configuration of the instances of the Alertmanager cluster `{{$labels.service}}` are out of sync. expr: | - count_values("config_hash", alertmanager_config_hash{job="alertmanager-main",namespace="monitoring"}) BY (service) / ON(service) GROUP_LEFT() label_replace(prometheus_operator_spec_replicas{job="prometheus-operator",namespace="monitoring",controller="alertmanager"}, "service", "alertmanager-$1", "name", "(.*)") != 1 + count_values("config_hash", alertmanager_config_hash{job="alertmanager-main",namespace="monitoring"}) BY (service) / ON(service) GROUP_LEFT() label_replace(max(prometheus_operator_spec_replicas{job="prometheus-operator",namespace="monitoring",controller="alertmanager"}) by (name, job, namespace, controller), "service", "alertmanager-$1", "name", "(.*)") != 1 for: 5m labels: severity: critical diff --git a/vars.jsonnet b/vars.jsonnet index 2e490fd..a3f2118 100644 --- a/vars.jsonnet +++ b/vars.jsonnet @@ -28,15 +28,20 @@ }, ], + k3s: { + enabled: false, + master_ip: '192.168.99.100' + }, + + // Domain suffix for the ingresses + suffixDomain: '192.168.99.100.nip.io', + // Setting these to false, defaults to emptyDirs enablePersistence: { prometheus: false, grafana: false, }, - // Domain suffix for the ingresses - suffixDomain: '192.168.99.100.nip.io', - // Grafana "from" email grafana: { from_address: 'myemail@gmail.com',