diff --git a/controller/api/destination/watcher/endpoints_watcher.go b/controller/api/destination/watcher/endpoints_watcher.go index 9fa9d3cff..269a679ff 100644 --- a/controller/api/destination/watcher/endpoints_watcher.go +++ b/controller/api/destination/watcher/endpoints_watcher.go @@ -19,6 +19,15 @@ import ( const ( kubeSystem = "kube-system" podIPIndex = "ip" + + // metrics labels + service = "service" + namespace = "namespace" + remoteGatewayNamespace = "remote_gateway_namespace" + remoteGateway = "remote_gateway" + remoteCluster = "remote_cluster" + remoteService = "remote_service" + remoteServiceNamespace = "remote_service_namespace" ) // TODO: prom metrics for all the queues/caches @@ -450,6 +459,30 @@ func (pp *portPublisher) updateEndpoints(endpoints *corev1.Endpoints) { pp.metrics.setExists(true) } +func metricLabels(endpoints *corev1.Endpoints) map[string]string { + labels := map[string]string{service: endpoints.Name, namespace: endpoints.Namespace} + + gateway, hasRemoteGateway := endpoints.Labels[consts.RemoteGatewayNameLabel] + gatewayNs, hasRemoteGatwayNs := endpoints.Labels[consts.RemoteGatewayNsLabel] + remoteClusterName, hasRemoteClusterName := endpoints.Labels[consts.RemoteClusterNameLabel] + serviceFqn, hasServiceFqn := endpoints.Annotations[consts.RemoteServiceFqName] + + if hasRemoteGateway && hasRemoteGatwayNs && hasRemoteClusterName && hasServiceFqn { + // this means we are looking at Endpoints created for the purpose of mirroring + // an out of cluster service. + labels[remoteGatewayNamespace] = gatewayNs + labels[remoteGateway] = gateway + labels[remoteCluster] = remoteClusterName + + fqParts := strings.Split(serviceFqn, ".") + if len(fqParts) >= 2 { + labels[remoteService] = fqParts[0] + labels[remoteServiceNamespace] = fqParts[1] + } + } + return labels +} + func (pp *portPublisher) endpointsToAddresses(endpoints *corev1.Endpoints) AddressSet { addresses := make(map[ID]Address) for _, subset := range endpoints.Subsets { @@ -504,7 +537,7 @@ func (pp *portPublisher) endpointsToAddresses(endpoints *corev1.Endpoints) Addre } return AddressSet{ Addresses: addresses, - Labels: map[string]string{"service": endpoints.Name, "namespace": endpoints.Namespace}, + Labels: metricLabels(endpoints), } } diff --git a/grafana/dashboards/multicluster.json b/grafana/dashboards/multicluster.json new file mode 100644 index 000000000..2599b9774 --- /dev/null +++ b/grafana/dashboards/multicluster.json @@ -0,0 +1,968 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "gnetId": null, + "graphTooltip": 1, + "id": null, + "iteration": 1531434867463, + "links": [], + "panels": [ + { + "content": "
\n  \n Cluster: $cluster, Gateway: $gateway\n
", + "gridPos": { + "h": 2, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 20, + "links": [], + "mode": "html", + "options": {}, + "title": "", + "transparent": true, + "type": "text" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#d44a3a", + "rgba(237, 129, 40, 0.89)", + "#299c46" + ], + "datasource": "prometheus", + "decimals": null, + "format": "percentunit", + "gauge": { + "maxValue": 1, + "minValue": 0, + "show": true, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 4, + "w": 8, + "x": 0, + "y": 2 + }, + "id": 5, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "options": {}, + "postfix": "", + "postfixFontSize": "50%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": true, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "sum(irate(response_total{classification=\"success\", dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) / sum(irate(response_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s]))", + "format": "time_series", + "instant": false, + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "0.9,.99", + "title": "SUCCESS RATE", + "transparent": true, + "type": "singlestat", + "valueFontSize": "80%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#299c46", + "rgba(237, 129, 40, 0.89)", + "#d44a3a" + ], + "datasource": "prometheus", + "decimals": null, + "format": "none", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 4, + "w": 8, + "x": 8, + "y": 2 + }, + "id": 4, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "options": {}, + "postfix": " RPS", + "postfixFontSize": "100%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": true, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "sum(irate(request_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s]))", + "format": "time_series", + "instant": false, + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "", + "title": "REQUEST RATE", + "transparent": true, + "type": "singlestat", + "valueFontSize": "100%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "cacheTimeout": null, + "colorBackground": false, + "colorValue": false, + "colors": [ + "#299c46", + "rgba(237, 129, 40, 0.89)", + "#d44a3a" + ], + "datasource": "prometheus", + "decimals": null, + "format": "none", + "gauge": { + "maxValue": 100, + "minValue": 0, + "show": false, + "thresholdLabels": false, + "thresholdMarkers": true + }, + "gridPos": { + "h": 4, + "w": 8, + "x": 16, + "y": 2 + }, + "id": 81, + "interval": null, + "links": [], + "mappingType": 1, + "mappingTypes": [ + { + "name": "value to text", + "value": 1 + }, + { + "name": "range to text", + "value": 2 + } + ], + "maxDataPoints": 100, + "nullPointMode": "connected", + "nullText": null, + "options": {}, + "postfix": " ms", + "postfixFontSize": "100%", + "prefix": "", + "prefixFontSize": "50%", + "rangeMaps": [ + { + "from": "null", + "text": "N/A", + "to": "null" + } + ], + "sparkline": { + "fillColor": "rgba(31, 118, 189, 0.18)", + "full": true, + "lineColor": "rgb(31, 120, 193)", + "show": true + }, + "tableColumn": "", + "targets": [ + { + "expr": "histogram_quantile(0.95, sum(irate(response_latency_ms_bucket{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (le))", + "format": "time_series", + "instant": false, + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "thresholds": "", + "title": "P95 LATENCY", + "transparent": true, + "type": "singlestat", + "valueFontSize": "100%", + "valueMaps": [ + { + "op": "=", + "text": "N/A", + "value": "null" + } + ], + "valueName": "current" + }, + { + "content": "
\n TOP-LINE TRAFFIC\n
", + "gridPos": { + "h": 2, + "w": 24, + "x": 0, + "y": 6 + }, + "id": 17, + "links": [], + "mode": "html", + "options": {}, + "title": "", + "transparent": true, + "type": "text" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "prometheus", + "fill": 1, + "gridPos": { + "h": 7, + "w": 8, + "x": 0, + "y": 8 + }, + "id": 67, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null", + "options": {}, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(irate(response_total{classification=\"success\", dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (dst_remote_gateway) / sum(irate(response_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (dst_remote_gateway)", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "gateway/{{dst_remote_gateway}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "SUCCESS RATE", + "tooltip": { + "shared": true, + "sort": 2, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": null, + "format": "percentunit", + "label": "", + "logBase": 1, + "max": "1", + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "prometheus", + "fill": 0, + "gridPos": { + "h": 7, + "w": 8, + "x": 8, + "y": 8 + }, + "id": 2, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null", + "options": {}, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(irate(request_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\", tls=\"true\"}[30s])) by (dst_remote_gateway)", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "🔒gateway/{{dst_remote_gateway}}", + "refId": "A" + }, + { + "expr": "sum(irate(request_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", tls!=\"true\"}[30s])) by (dst_remote_gateway)", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "gatewy/{{dst_remote_gateway}}", + "refId": "B" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "REQUEST RATE", + "tooltip": { + "shared": true, + "sort": 2, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": null, + "format": "rps", + "label": "", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "prometheus", + "fill": 1, + "gridPos": { + "h": 7, + "w": 8, + "x": 16, + "y": 8 + }, + "id": 68, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null", + "options": {}, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "histogram_quantile(0.5, sum(irate(response_latency_ms_bucket{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (le, dst_remote_gateway))", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "p50 gateway/{{dst_remote_gateway}}", + "refId": "A" + }, + { + "expr": "histogram_quantile(0.95, sum(irate(response_latency_ms_bucket{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (le, dst_remote_gateway))", + "format": "time_series", + "hide": false, + "intervalFactor": 1, + "legendFormat": "p95 gateway/{{dst_remote_gateway}}", + "refId": "B" + }, + { + "expr": "histogram_quantile(0.99, sum(irate(response_latency_ms_bucket{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (le, dst_remote_gateway))", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "p99 gateway/{{dst_remote_gateway}}", + "refId": "C" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "LATENCY", + "tooltip": { + "shared": true, + "sort": 2, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": null, + "format": "ms", + "label": "", + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "content": "
\n TRAFFIC BY REMOTE SERVICE\n
", + "gridPos": { + "h": 2, + "w": 24, + "x": 0, + "y": 15 + }, + "id": 32, + "links": [], + "mode": "html", + "options": {}, + "title": "", + "transparent": true, + "type": "text" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "prometheus", + "fill": 1, + "gridPos": { + "h": 7, + "w": 8, + "x": 0, + "y": 17 + }, + "id": 77, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null", + "options": {}, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(irate(response_total{classification=\"success\", dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (dst_remote_service) / sum(irate(response_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (dst_remote_service)", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "remote-svc/{{dst_remote_service}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "SUCCESS RATE", + "tooltip": { + "shared": true, + "sort": 2, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "decimals": null, + "format": "percentunit", + "label": "", + "logBase": 1, + "max": "1", + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "prometheus", + "fill": 0, + "gridPos": { + "h": 7, + "w": 8, + "x": 8, + "y": 17 + }, + "id": 78, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null", + "options": {}, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum(irate(request_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", tls=\"true\"}[30s])) by (dst_remote_service)", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "🔒remote-svc/{{dst_remote_service}}", + "refId": "A" + }, + { + "expr": "sum(irate(request_total{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", tls!=\"true\"}[30s])) by (dst_remote_service)", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "remote-svc/{{dst_remote_service}}", + "refId": "B" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "REQUEST RATE", + "tooltip": { + "shared": true, + "sort": 2, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "rps", + "label": "", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "prometheus", + "fill": 1, + "gridPos": { + "h": 7, + "w": 8, + "x": 16, + "y": 17 + }, + "id": 79, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 2, + "links": [], + "nullPointMode": "null", + "options": {}, + "percentage": false, + "pointradius": 5, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "histogram_quantile(0.95, sum(rate(response_latency_ms_bucket{dst_remote_gateway=\"$gateway\", dst_remote_gateway!=\"\", dst_remote_cluster=\"$cluster\", dst_remote_cluster!=\"\", direction=\"outbound\"}[30s])) by (le, dst_remote_service))", + "format": "time_series", + "intervalFactor": 1, + "legendFormat": "P95 remote-svc/{{dst_remote_service}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "P95 LATENCY", + "tooltip": { + "shared": true, + "sort": 2, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "refresh": "1m", + "schemaVersion": 18, + "style": "dark", + "tags": [ + "linkerd" + ], + "templating": { + "list": [ + { + "allValue": ".*", + "current": { + "text": "All", + "value": "$__all" + }, + "datasource": "prometheus", + "definition": "", + "hide": 0, + "includeAll": false, + "label": "Cluster", + "multi": false, + "name": "cluster", + "options": [], + "query": "label_values(request_total, dst_remote_cluster)", + "refresh": 2, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + }, + { + "allValue": ".*", + "current": { + "text": "All", + "value": "$__all" + }, + "datasource": "prometheus", + "definition": "", + "hide": 0, + "includeAll": false, + "label": "Gateway", + "multi": false, + "name": "gateway", + "options": [], + "query": "label_values(request_total{dst_remote_cluster=\"$cluster\"}, dst_remote_gateway)", + "refresh": 2, + "regex": "", + "skipUrlSync": false, + "sort": 1, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + } + ] + }, + "time": { + "from": "now-5m", + "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": "Linkerd Multicluster", + "uid": "linkerd-multicluster", + "version": 1 +}