Merge pull request #4227 from RolandMa1986/feat-gateway-matrics

feat: ingress metrics query apis

pkg/simple/client/monitoring/prometheus/promql.go

@@ -165,6 +165,23 @@ var promQLTemplates = map[string]string{
 	"namespace_ingresses_extensions_count": `sum by (namespace) (kube_ingress_labels{namespace!=""} * on (namespace) group_left(workspace) kube_namespace_labels{$1})`,
 	"namespace_s2ibuilder_count":           `sum by (namespace) (s2i_s2ibuilder_created{namespace!=""} * on (namespace) group_left(workspace) kube_namespace_labels{$1})`,
 
+	// ingress
+	"ingress_request_count":                 `round(sum(increase(nginx_ingress_controller_requests{$1,$2}[$3])))`,
+	"ingress_request_4xx_count":             `round(sum(increase(nginx_ingress_controller_requests{$1,$2,status="[4].*"}[$3])))`,
+	"ingress_request_5xx_count":             `round(sum(increase(nginx_ingress_controller_requests{$1,$2,status="[5].*"}[$3])))`,
+	"ingress_active_connections":            `sum(avg_over_time(nginx_ingress_controller_nginx_process_connections{$2,state="active"}[$3]))`,
+	"ingress_success_rate":                  `sum(rate(nginx_ingress_controller_requests{$1,$2,status!~"[4-5].*"}[$3])) / sum(rate(nginx_ingress_controller_requests{$1,$2}[$3]))`,
+	"ingress_request_duration_average":      `sum_over_time(nginx_ingress_controller_request_duration_seconds_sum{$1,$2}[$3])/sum_over_time(nginx_ingress_controller_request_duration_seconds_count{$1,$2}[$3])`,
+	"ingress_request_duration_50percentage": `histogram_quantile(0.50, sum by (le) (rate(nginx_ingress_controller_request_duration_seconds_bucket{$1,$2}[$3])))`,
+	"ingress_request_duration_95percentage": `histogram_quantile(0.90, sum by (le) (rate(nginx_ingress_controller_request_duration_seconds_bucket{$1,$2}[$3])))`,
+	"ingress_request_duration_99percentage": `histogram_quantile(0.99, sum by (le) (rate(nginx_ingress_controller_request_duration_seconds_bucket{$1,$2}[$3])))`,
+	"ingress_request_volume":                `round(sum(irate(nginx_ingress_controller_requests{$1,$2}[$3])), 0.001)`,
+	"ingress_request_volume_by_ingress":     `round(sum(irate(nginx_ingress_controller_requests{$1,$2}[$3])) by (ingress), 0.001)`,
+	"ingress_request_network_sent":          `sum(irate(nginx_ingress_controller_response_size_sum{$1,$2}[$3]))`,
+	"ingress_request_network_received":      `sum(irate(nginx_ingress_controller_request_size_sum{$1,$2}[$3]))`,
+	"ingress_request_memory_bytes":          `avg(nginx_ingress_controller_nginx_process_resident_memory_bytes{$2})`,
+	"ingress_request_cpu_usage":             `avg(rate(nginx_ingress_controller_nginx_process_cpu_seconds_total{$2}[5m]))`,
+
 	// workload
 	"workload_cpu_usage":             `round(namespace:workload_cpu_usage:sum{$1}, 0.001)`,
 	"workload_memory_usage":          `namespace:workload_memory_usage:sum{$1}`,
@@ -259,6 +276,8 @@ func makeExpr(metric string, opts monitoring.QueryOptions) string {
 		return makeContainerMetricExpr(tmpl, opts)
 	case monitoring.LevelPVC:
 		return makePVCMetricExpr(tmpl, opts)
+	case monitoring.LevelIngress:
+		return makeIngressMetricExpr(tmpl, opts)
 	case monitoring.LevelComponent:
 		return tmpl
 	default:
@@ -449,3 +468,37 @@ func makePVCMetricExpr(tmpl string, o monitoring.QueryOptions) string {
 	}
 	return strings.Replace(tmpl, "$1", pvcSelector, -1)
 }
+
+func makeIngressMetricExpr(tmpl string, o monitoring.QueryOptions) string {
+	var ingressSelector string
+	var jobSelector string
+	duration := "5m"
+
+	// parse Range Vector Selectors metric{key=value}[duration]
+	if o.Duration != nil {
+		duration = o.Duration.String()
+	}
+
+	// For monitoring ingress in the specific namespace
+	// GET /namespaces/{namespace}/ingress/{ingress} or
+	// GET /namespaces/{namespace}/ingress
+	if o.NamespaceName != "" {
+		if o.Ingress != "" {
+			ingressSelector = fmt.Sprintf(`exported_namespace="%s", ingress="%s"`, o.NamespaceName, o.Ingress)
+		} else {
+			ingressSelector = fmt.Sprintf(`exported_namespace="%s", ingress=~"%s"`, o.NamespaceName, o.ResourceFilter)
+		}
+	}
+	// job is a reqiuried filter
+	// GET /namespaces/{namespace}/ingress?job=xxx&pod=xxx
+	if o.Job != "" {
+		jobSelector = fmt.Sprintf(`job="%s"`, o.Job)
+		if o.PodName != "" {
+			jobSelector = fmt.Sprintf(`%s,controller_pod="%s"`, jobSelector, o.PodName)
+		}
+	}
+
+	tmpl = strings.Replace(tmpl, "$1", ingressSelector, -1)
+	tmpl = strings.Replace(tmpl, "$2", jobSelector, -1)
+	return strings.Replace(tmpl, "$3", duration, -1)
+}

pkg/simple/client/monitoring/prometheus/promql_test.go

@@ -186,6 +186,16 @@ func TestMakeExpr(t *testing.T) {
 				ResourceFilter:   "db-123",
 			},
 		},
+		{
+			name: "ingress_request_count",
+			opts: monitoring.QueryOptions{
+				Level:         monitoring.LevelIngress,
+				NamespaceName: "default",
+				Ingress:       "ingress-1",
+				Job:           "job-1",
+				PodName:       "pod-1",
+			},
+		},
 		{
 			name: "etcd_server_list",
 			opts: monitoring.QueryOptions{

pkg/simple/client/monitoring/prometheus/testdata/promqls.go

@@ -37,5 +37,6 @@ var PromQLs = map[string]string{
 	"pvc_inodes_available":                  `max by (namespace, persistentvolumeclaim) (kubelet_volume_stats_inodes_free) * on (namespace, persistentvolumeclaim) group_left (storageclass) kube_persistentvolumeclaim_info{namespace="default", persistentvolumeclaim="db-123"}`,
 	"pvc_inodes_used":                       `max by (namespace, persistentvolumeclaim) (kubelet_volume_stats_inodes_used) * on (namespace, persistentvolumeclaim) group_left (storageclass) kube_persistentvolumeclaim_info{namespace="default", persistentvolumeclaim=~"db-123"}`,
 	"pvc_inodes_total":                      `max by (namespace, persistentvolumeclaim) (kubelet_volume_stats_inodes) * on (namespace, persistentvolumeclaim) group_left (storageclass) kube_persistentvolumeclaim_info{storageclass="default", persistentvolumeclaim=~"db-123"}`,
+	"ingress_request_count":                 `round(sum(increase(nginx_ingress_controller_requests{exported_namespace="default", ingress="ingress-1",job="job-1",controller_pod="pod-1"}[5m])))`,
 	"etcd_server_list":                      `label_replace(up{job="etcd"}, "node_ip", "$1", "instance", "(.*):.*")`,
 }

pkg/simple/client/monitoring/query_options.go

@@ -37,6 +37,7 @@ const (
 	LevelContainer
 	LevelPVC
 	LevelComponent
+	LevelIngress
 )
 
 var MeteringLevelMap = map[string]int{
@@ -81,6 +82,9 @@ type QueryOptions struct {
 	PVCFilter                 string
 	ApplicationName           string
 	ServiceName               string
+	Ingress                   string
+	Job                       string
+	Duration                  *time.Duration
 	MeterOptions              *Meteroptions
 }
 
@@ -287,6 +291,25 @@ func (po PVCOption) Apply(o *QueryOptions) {
 	o.PVCFilter = po.PersistentVolumeClaimName
 }
 
+type IngressOption struct {
+	ResourceFilter string
+	NamespaceName  string
+	Ingress        string
+	Job            string
+	Pod            string
+	Duration       *time.Duration
+}
+
+func (no IngressOption) Apply(o *QueryOptions) {
+	o.Level = LevelIngress
+	o.ResourceFilter = no.ResourceFilter
+	o.NamespaceName = no.NamespaceName
+	o.Ingress = no.Ingress
+	o.Job = no.Job
+	o.PodName = no.Pod
+	o.Duration = no.Duration
+}
+
 type ComponentOption struct{}
 
 func (_ ComponentOption) Apply(o *QueryOptions) {