[SEBA-201] Moving nem-monitoring chart in the helm-chart repo
Change-Id: I81de49bef81fcec122ddf823d792c8d933ca584e
diff --git a/nem-monitoring/grafana-dashboards/kb8s-app-metrics.json b/nem-monitoring/grafana-dashboards/kb8s-app-metrics.json
new file mode 100644
index 0000000..9b5d28f
--- /dev/null
+++ b/nem-monitoring/grafana-dashboards/kb8s-app-metrics.json
@@ -0,0 +1,1400 @@
+{
+ "__inputs": [
+ {
+ "name": "DS_PROMETHEUS",
+ "label": "prometheus",
+ "description": "",
+ "type": "datasource",
+ "pluginId": "prometheus",
+ "pluginName": "Prometheus"
+ }
+ ],
+ "__requires": [
+ {
+ "type": "grafana",
+ "id": "grafana",
+ "name": "Grafana",
+ "version": "4.1.1"
+ },
+ {
+ "type": "panel",
+ "id": "graph",
+ "name": "Graph",
+ "version": ""
+ },
+ {
+ "type": "datasource",
+ "id": "prometheus",
+ "name": "Prometheus",
+ "version": "1.0.0"
+ }
+ ],
+ "annotations": {
+ "list": []
+ },
+ "editable": true,
+ "gnetId": 1471,
+ "graphTooltip": 1,
+ "hideControls": false,
+ "id": null,
+ "links": [],
+ "refresh": "30s",
+ "rows": [
+ {
+ "collapse": false,
+ "height": "250px",
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "grid": {},
+ "id": 3,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 2,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(http_requests_total{app=\"$container\", handler!=\"prometheus\", kubernetes_namespace=\"$namespace\"}[30s])) by (kubernetes_namespace,app,code)",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "native | {{code}}",
+ "refId": "A",
+ "step": 10
+ },
+ {
+ "expr": "sum(irate(nginx_http_requests_total{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (kubernetes_namespace,app,status)",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "nginx | {{status}}",
+ "refId": "B",
+ "step": 10
+ },
+ {
+ "expr": "sum(irate(haproxy_backend_http_responses_total{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,code)",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "haproxy | {{code}}",
+ "refId": "C",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Request rate",
+ "tooltip": {
+ "msResolution": true,
+ "shared": false,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "ops",
+ "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,
+ "datasource": "Prometheus",
+ "fill": 1,
+ "id": 15,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(haproxy_backend_http_responses_total{app=\"$container\", kubernetes_namespace=\"$namespace\",code=\"5xx\"}[30s])) by (app,kubernetes_namespace) / sum(irate(haproxy_backend_http_responses_total{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace)",
+ "interval": "",
+ "intervalFactor": 2,
+ "legendFormat": "haproxy",
+ "refId": "A",
+ "step": 20
+ },
+ {
+ "expr": "sum(irate(http_requests_total{app=\"$container\", handler!=\"prometheus\", kubernetes_namespace=\"$namespace\", code=~\"5[0-9]+\"}[30s])) by (kubernetes_namespace,app) / sum(irate(http_requests_total{app=\"$container\", handler!=\"prometheus\", kubernetes_namespace=\"$namespace\"}[30s])) by (kubernetes_namespace,app)",
+ "intervalFactor": 2,
+ "legendFormat": "native",
+ "refId": "B",
+ "step": 20
+ },
+ {
+ "expr": "sum(irate(nginx_http_requests_total{app=\"$container\", kubernetes_namespace=\"$namespace\", status=~\"5[0-9]+\"}[30s])) by (kubernetes_namespace,app) / sum(irate(nginx_http_requests_total{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (kubernetes_namespace,app)",
+ "intervalFactor": 2,
+ "legendFormat": "nginx",
+ "refId": "C",
+ "step": 20
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Error rate",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "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": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Request rate",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": true,
+ "height": 224,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "grid": {},
+ "id": 5,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "max",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 2,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 12,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,le))",
+ "intervalFactor": 1,
+ "legendFormat": "native | 0.99",
+ "refId": "A",
+ "step": 1
+ },
+ {
+ "expr": "histogram_quantile(0.90, sum(rate(http_request_duration_seconds_bucket{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,le))",
+ "intervalFactor": 1,
+ "legendFormat": "native | 0.90",
+ "refId": "B",
+ "step": 1
+ },
+ {
+ "expr": "histogram_quantile(0.5, sum(rate(http_request_duration_seconds_bucket{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,le))",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "native | 0.50",
+ "refId": "C",
+ "step": 1
+ },
+ {
+ "expr": "histogram_quantile(0.99, sum(rate(nginx_http_request_duration_seconds_bucket{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,le))",
+ "intervalFactor": 1,
+ "legendFormat": "nginx | 0.99",
+ "refId": "D",
+ "step": 1
+ },
+ {
+ "expr": "histogram_quantile(0.9, sum(rate(nginx_http_request_duration_seconds_bucket{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,le))",
+ "intervalFactor": 1,
+ "legendFormat": "nginx | 0.90",
+ "refId": "E",
+ "step": 1
+ },
+ {
+ "expr": "histogram_quantile(0.5, sum(rate(nginx_http_request_duration_seconds_bucket{app=\"$container\", kubernetes_namespace=\"$namespace\"}[30s])) by (app,kubernetes_namespace,le))",
+ "intervalFactor": 1,
+ "legendFormat": "nginx | 0.50",
+ "refId": "F",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Response time percentiles",
+ "tooltip": {
+ "msResolution": true,
+ "shared": true,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "s",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Response time",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": false,
+ "height": 250,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "id": 7,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 12,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "count(count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (pod_name))",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "pods",
+ "refId": "A",
+ "step": 5
+ },
+ {
+ "expr": "count(count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (kubernetes_io_hostname))",
+ "interval": "",
+ "intervalFactor": 2,
+ "legendFormat": "hosts",
+ "refId": "B",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Number of pods",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Pod count",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": false,
+ "height": 250,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "id": 12,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [
+ {
+ "alias": "elasticsearch-logging-data-20170207a (logging) - system",
+ "color": "#BF1B00"
+ },
+ {
+ "alias": "elasticsearch-logging-data-20170207a (logging) - user",
+ "color": "#508642"
+ }
+ ],
+ "span": 12,
+ "stack": true,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(container_cpu_system_seconds_total{container_name=\"$container\", namespace=\"$namespace\"}[30s])) by (namespace,container_name) / sum(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"} / 1024) by (namespace,container_name)",
+ "intervalFactor": 2,
+ "legendFormat": "system",
+ "refId": "C",
+ "step": 10
+ },
+ {
+ "expr": "sum(irate(container_cpu_user_seconds_total{container_name=\"$container\", namespace=\"$namespace\"}[30s])) by (namespace,container_name) / sum(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"} / 1024) by (namespace,container_name)",
+ "interval": "",
+ "intervalFactor": 2,
+ "legendFormat": "user",
+ "refId": "B",
+ "step": 10
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Cpu usage (relative to request)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "percentunit",
+ "label": "",
+ "logBase": 1,
+ "max": "1",
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Usage relative to request",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": true,
+ "height": 250,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "id": 10,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(container_cpu_usage_seconds_total{container_name=\"$container\", namespace=\"$namespace\"}[30s])) by (namespace,container_name) / sum(container_spec_cpu_quota{container_name=\"$container\", namespace=\"$namespace\"} / container_spec_cpu_period{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name)",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "actual",
+ "metric": "",
+ "refId": "A",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Cpu usage (relative to limit)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "percentunit",
+ "label": "",
+ "logBase": 1,
+ "max": "1",
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "id": 11,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) / sum(container_spec_memory_limit_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name)",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "actual",
+ "refId": "A",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Memory usage (relative to limit)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "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": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Usage relative to limit",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": true,
+ "height": 250,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "fill": 1,
+ "id": 13,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(container_cpu_usage_seconds_total{container_name=\"$container\", namespace=\"$namespace\"}[30s])) by (id,pod_name)",
+ "interval": "",
+ "intervalFactor": 2,
+ "legendFormat": "{{pod_name}}",
+ "refId": "A",
+ "step": 2
+ },
+ {
+ "expr": "sum(container_spec_cpu_quota{container_name=\"$container\", namespace=\"$namespace\"} / container_spec_cpu_period{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) / count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "intervalFactor": 2,
+ "legendFormat": "limit",
+ "refId": "B",
+ "step": 2
+ },
+ {
+ "expr": "sum(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"} / 1024) by (namespace,container_name) / count(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "intervalFactor": 2,
+ "legendFormat": "request",
+ "refId": "C",
+ "step": 2
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Cpu usage (per pod)",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "short",
+ "label": "cores",
+ "logBase": 1,
+ "max": null,
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "fill": 1,
+ "id": 14,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "null",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (id,pod_name)",
+ "interval": "",
+ "intervalFactor": 2,
+ "legendFormat": "{{pod_name}}",
+ "refId": "A",
+ "step": 2
+ },
+ {
+ "expr": "sum(container_spec_memory_limit_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) / count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (container_name,namespace)",
+ "intervalFactor": 2,
+ "legendFormat": "limit",
+ "refId": "B",
+ "step": 2
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Memory usage (per pod)",
+ "tooltip": {
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "bytes",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Usage per pod",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": true,
+ "height": 250,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "id": 8,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(container_cpu_usage_seconds_total{container_name=\"$container\", namespace=\"$namespace\"}[30s])) by (namespace,container_name) / count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "actual",
+ "refId": "A",
+ "step": 1
+ },
+ {
+ "expr": "sum(container_spec_cpu_quota{container_name=\"$container\", namespace=\"$namespace\"} / container_spec_cpu_period{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) / count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "limit",
+ "refId": "B",
+ "step": 1
+ },
+ {
+ "expr": "sum(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"} / 1024) by (namespace,container_name) / count(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "request",
+ "refId": "C",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Cpu usage (avg per pod)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "none",
+ "label": "cores",
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "id": 9,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 1,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) / count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "intervalFactor": 1,
+ "legendFormat": "actual",
+ "metric": "",
+ "refId": "A",
+ "step": 1
+ },
+ {
+ "expr": "sum(container_spec_memory_limit_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) / count(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name) ",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "limit",
+ "refId": "B",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Memory usage (avg per pod)",
+ "tooltip": {
+ "msResolution": false,
+ "shared": true,
+ "sort": 0,
+ "value_type": "individual"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "bytes",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Usage per pod (average)",
+ "titleSize": "h6"
+ },
+ {
+ "collapse": true,
+ "height": 259.4375,
+ "panels": [
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "grid": {},
+ "id": 1,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 2,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(irate(container_cpu_usage_seconds_total{container_name=\"$container\", namespace=\"$namespace\"}[30s])) by (namespace,container_name)",
+ "hide": false,
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "actual",
+ "metric": "",
+ "refId": "A",
+ "step": 1
+ },
+ {
+ "expr": "sum(container_spec_cpu_quota{container_name=\"$container\", namespace=\"$namespace\"} / container_spec_cpu_period{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name)",
+ "intervalFactor": 1,
+ "legendFormat": "limit",
+ "refId": "B",
+ "step": 1
+ },
+ {
+ "expr": "sum(container_spec_cpu_shares{container_name=\"$container\", namespace=\"$namespace\"} / 1024) by (namespace,container_name) ",
+ "intervalFactor": 1,
+ "legendFormat": "request",
+ "refId": "C",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Cpu usage (total)",
+ "tooltip": {
+ "msResolution": true,
+ "shared": false,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "none",
+ "label": "cores",
+ "logBase": 1,
+ "max": null,
+ "min": 0,
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ },
+ {
+ "aliasColors": {},
+ "bars": false,
+ "datasource": "Prometheus",
+ "editable": true,
+ "error": false,
+ "fill": 1,
+ "grid": {},
+ "id": 2,
+ "legend": {
+ "alignAsTable": true,
+ "avg": true,
+ "current": false,
+ "max": true,
+ "min": false,
+ "rightSide": true,
+ "show": true,
+ "sort": "avg",
+ "sortDesc": true,
+ "total": false,
+ "values": true
+ },
+ "lines": true,
+ "linewidth": 2,
+ "links": [],
+ "nullPointMode": "connected",
+ "percentage": false,
+ "pointradius": 5,
+ "points": false,
+ "renderer": "flot",
+ "seriesOverrides": [],
+ "span": 6,
+ "stack": false,
+ "steppedLine": false,
+ "targets": [
+ {
+ "expr": "sum(container_memory_usage_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name)",
+ "interval": "",
+ "intervalFactor": 1,
+ "legendFormat": "actual",
+ "refId": "A",
+ "step": 1
+ },
+ {
+ "expr": "sum(container_spec_memory_limit_bytes{container_name=\"$container\", namespace=\"$namespace\"}) by (namespace,container_name)",
+ "intervalFactor": 1,
+ "legendFormat": "limit",
+ "refId": "B",
+ "step": 1
+ }
+ ],
+ "thresholds": [],
+ "timeFrom": null,
+ "timeShift": null,
+ "title": "Memory usage (total)",
+ "tooltip": {
+ "msResolution": true,
+ "shared": false,
+ "sort": 0,
+ "value_type": "cumulative"
+ },
+ "type": "graph",
+ "xaxis": {
+ "mode": "time",
+ "name": null,
+ "show": true,
+ "values": []
+ },
+ "yaxes": [
+ {
+ "format": "bytes",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": "0",
+ "show": true
+ },
+ {
+ "format": "short",
+ "label": null,
+ "logBase": 1,
+ "max": null,
+ "min": null,
+ "show": true
+ }
+ ]
+ }
+ ],
+ "repeat": null,
+ "repeatIteration": null,
+ "repeatRowId": null,
+ "showTitle": false,
+ "title": "Usage total",
+ "titleSize": "h6"
+ }
+ ],
+ "schemaVersion": 14,
+ "style": "dark",
+ "tags": [],
+ "templating": {
+ "list": [
+ {
+ "allValue": ".+",
+ "current": {},
+ "datasource": "Prometheus",
+ "hide": 0,
+ "includeAll": false,
+ "label": null,
+ "multi": false,
+ "name": "namespace",
+ "options": [],
+ "query": "label_values(container_memory_usage_bytes{namespace=~\".+\",container_name!=\"POD\"},namespace)",
+ "refresh": 1,
+ "regex": "",
+ "sort": 1,
+ "tagValuesQuery": null,
+ "tags": [],
+ "tagsQuery": null,
+ "type": "query",
+ "useTags": false
+ },
+ {
+ "allValue": ".+",
+ "current": {},
+ "datasource": "Prometheus",
+ "hide": 0,
+ "includeAll": false,
+ "label": null,
+ "multi": false,
+ "name": "container",
+ "options": [],
+ "query": "label_values(container_memory_usage_bytes{namespace=~\"$namespace\",container_name!=\"POD\"},container_name)",
+ "refresh": 1,
+ "regex": "",
+ "sort": 1,
+ "tagValuesQuery": null,
+ "tags": [],
+ "tagsQuery": null,
+ "type": "query",
+ "useTags": false
+ }
+ ]
+ },
+ "time": {
+ "from": "now-3h",
+ "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": "browser",
+ "title": "Kubernetes App Metrics",
+ "version": 37,
+ "description": "After selecting your namespace and container you get a wealth of metrics like request rate, error rate, response times, pod count, cpu and memory usage. You can view cpu and memory usage in a variety of ways, compared to the limit, compared to the request, per pod, average per pod, etc."
+ }
\ No newline at end of file