Fix open connections metric

Co-authored-by: Romain <rtribotte@users.noreply.github.com>
traefik · Mar 20, 2023 · 7c2af10 · 7c2af10
1 parent 598a257
commit 7c2af10
Show file tree

Hide file tree

Showing 17 changed files with 126 additions and 236 deletions.
diff --git a/cmd/traefik/traefik.go b/cmd/traefik/traefik.go
@@ -193,9 +193,14 @@ func setupServer(staticConfiguration *static.Configuration) (*server.Server, err
 
 	tsProviders := initTailscaleProviders(staticConfiguration, &providerAggregator)
 
+	// Metrics
+
+	metricRegistries := registerMetricClients(staticConfiguration.Metrics)
+	metricsRegistry := metrics.NewMultiRegistry(metricRegistries)
+
 	// Entrypoints
 
-	serverEntryPointsTCP, err := server.NewTCPEntryPoints(staticConfiguration.EntryPoints, staticConfiguration.HostResolver)
+	serverEntryPointsTCP, err := server.NewTCPEntryPoints(staticConfiguration.EntryPoints, staticConfiguration.HostResolver, metricsRegistry)
 	if err != nil {
 		return nil, err
 	}
@@ -243,11 +248,6 @@ func setupServer(staticConfiguration *static.Configuration) (*server.Server, err
 		}
 	}
 
-	// Metrics
-
-	metricRegistries := registerMetricClients(staticConfiguration.Metrics)
-	metricsRegistry := metrics.NewMultiRegistry(metricRegistries)
-
 	// Service manager factory
 
 	var spiffeX509Source *workloadapi.X509Source

diff --git a/docs/content/observability/metrics/overview.md b/docs/content/observability/metrics/overview.md
@@ -16,59 +16,74 @@ Traefik Proxy hosts an official Grafana dashboard for both [on-premises](https:/
 
 ## Global Metrics
 
-| Metric                                      | Type    | Description                                             |
-|---------------------------------------------|---------|---------------------------------------------------------|
-| Config reload total                         | Count   | The total count of configuration reloads.               |
-| Config reload last success                  | Gauge   | The timestamp of the last configuration reload success. |
-| TLS certificates not after                  | Gauge   | The expiration date of certificates.                    |
+| Metric                     | Type  | [Labels](#labels)        | Description                                                        |
+|----------------------------|-------|--------------------------|--------------------------------------------------------------------|
+| Config reload total        | Count |                          | The total count of configuration reloads.                          |
+| Config reload last success | Gauge |                          | The timestamp of the last configuration reload success.            |
+| Open connections           | Gauge | `entrypoint`, `protocol` | The current count of open connections, by entrypoint and protocol. |
+| TLS certificates not after | Gauge |                          | The expiration date of certificates.                               |
 
 ```prom tab="Prometheus"
 traefik_config_reloads_total
 traefik_config_last_reload_success
+traefik_open_connections
 traefik_tls_certs_not_after
 ```
 
 ```dd tab="Datadog"
 config.reload.total
 config.reload.lastSuccessTimestamp
+open.connections
 tls.certs.notAfterTimestamp
 ```
 
 ```influxdb tab="InfluxDB2"
 traefik.config.reload.total
 traefik.config.reload.lastSuccessTimestamp
+traefik.open.connections
 traefik.tls.certs.notAfterTimestamp
 ```
 
 ```statsd tab="StatsD"
 # Default prefix: "traefik"
 {prefix}.config.reload.total
 {prefix}.config.reload.lastSuccessTimestamp
+{prefix}.open.connections
 {prefix}.tls.certs.notAfterTimestamp
 ```
 
 ```opentelemetry tab="OpenTelemetry"
 traefik_config_reloads_total
 traefik_config_last_reload_success
+traefik_open_connections
 traefik_tls_certs_not_after
 ```
 
-## EntryPoint Metrics
+### Labels
+
+Here is a comprehensive list of labels that are provided by the global metrics:
+
+| Label         | Description                            | example              |
+|---------------|----------------------------------------|----------------------|
+| `entrypoint`  | Entrypoint that handled the connection | "example_entrypoint" |
+| `protocol`    | Connection protocol                    | "TCP"                |
+
+## HTTP Metrics
+
+### EntryPoint Metrics
 
 | Metric                | Type      | [Labels](#labels)                          | Description                                                         |
 |-----------------------|-----------|--------------------------------------------|---------------------------------------------------------------------|
 | Requests total        | Count     | `code`, `method`, `protocol`, `entrypoint` | The total count of HTTP requests received by an entrypoint.         |
 | Requests TLS total    | Count     | `tls_version`, `tls_cipher`, `entrypoint`  | The total count of HTTPS requests received by an entrypoint.        |
 | Request duration      | Histogram | `code`, `method`, `protocol`, `entrypoint` | Request processing duration histogram on an entrypoint.             |
-| Open connections      | Count     | `method`, `protocol`, `entrypoint`         | The current count of open connections on an entrypoint.             |
 | Requests bytes total  | Count     | `code`, `method`, `protocol`, `entrypoint` | The total size of HTTP requests in bytes handled by an entrypoint.  |
 | Responses bytes total | Count     | `code`, `method`, `protocol`, `entrypoint` | The total size of HTTP responses in bytes handled by an entrypoint. |
 
 ```prom tab="Prometheus"
 traefik_entrypoint_requests_total
 traefik_entrypoint_requests_tls_total
 traefik_entrypoint_request_duration_seconds
-traefik_entrypoint_open_connections
 traefik_entrypoint_requests_bytes_total
 traefik_entrypoint_responses_bytes_total
 ```
@@ -77,7 +92,6 @@ traefik_entrypoint_responses_bytes_total
 entrypoint.request.total
 entrypoint.request.tls.total
 entrypoint.request.duration
-entrypoint.connections.open
 entrypoint.requests.bytes.total
 entrypoint.responses.bytes.total
 ```
@@ -86,7 +100,6 @@ entrypoint.responses.bytes.total
 traefik.entrypoint.requests.total
 traefik.entrypoint.requests.tls.total
 traefik.entrypoint.request.duration
-traefik.entrypoint.connections.open
 traefik.entrypoint.requests.bytes.total
 traefik.entrypoint.responses.bytes.total
 ```
@@ -96,7 +109,6 @@ traefik.entrypoint.responses.bytes.total
 {prefix}.entrypoint.request.total
 {prefix}.entrypoint.request.tls.total
 {prefix}.entrypoint.request.duration
-{prefix}.entrypoint.connections.open
 {prefix}.entrypoint.requests.bytes.total
 {prefix}.entrypoint.responses.bytes.total
 ```
@@ -110,22 +122,20 @@ traefik_entrypoint_requests_bytes_total
 traefik_entrypoint_responses_bytes_total
 ```
 
-## Router Metrics
+### Router Metrics
 
 | Metric                | Type      | [Labels](#labels)                                 | Description                                                    |
 |-----------------------|-----------|---------------------------------------------------|----------------------------------------------------------------|
 | Requests total        | Count     | `code`, `method`, `protocol`, `router`, `service` | The total count of HTTP requests handled by a router.          |
 | Requests TLS total    | Count     | `tls_version`, `tls_cipher`, `router`, `service`  | The total count of HTTPS requests handled by a router.         |
 | Request duration      | Histogram | `code`, `method`, `protocol`, `router`, `service` | Request processing duration histogram on a router.             |
-| Open connections      | Count     | `method`, `protocol`, `router`, `service`         | The current count of open connections on a router.             |
 | Requests bytes total  | Count     | `code`, `method`, `protocol`, `router`, `service` | The total size of HTTP requests in bytes handled by a router.  |
 | Responses bytes total | Count     | `code`, `method`, `protocol`, `router`, `service` | The total size of HTTP responses in bytes handled by a router. |
 
 ```prom tab="Prometheus"
 traefik_router_requests_total
 traefik_router_requests_tls_total
 traefik_router_request_duration_seconds
-traefik_router_open_connections
 traefik_router_requests_bytes_total
 traefik_router_responses_bytes_total
 ```
@@ -134,7 +144,6 @@ traefik_router_responses_bytes_total
 router.request.total
 router.request.tls.total
 router.request.duration
-router.connections.open
 router.requests.bytes.total
 router.responses.bytes.total
 ```
@@ -143,7 +152,6 @@ router.responses.bytes.total
 traefik.router.requests.total
 traefik.router.requests.tls.total
 traefik.router.request.duration
-traefik.router.connections.open
 traefik.router.requests.bytes.total
 traefik.router.responses.bytes.total
 ```
@@ -153,7 +161,6 @@ traefik.router.responses.bytes.total
 {prefix}.router.request.total
 {prefix}.router.request.tls.total
 {prefix}.router.request.duration
-{prefix}.router.connections.open
 {prefix}.router.requests.bytes.total
 {prefix}.router.responses.bytes.total
 ```
@@ -167,14 +174,13 @@ traefik_router_requests_bytes_total
 traefik_router_responses_bytes_total
 ```
 
-## Service Metrics
+### Service Metrics
 
 | Metric                | Type      | Labels                                  | Description                                                 |
 |-----------------------|-----------|-----------------------------------------|-------------------------------------------------------------|
 | Requests total        | Count     | `code`, `method`, `protocol`, `service` | The total count of HTTP requests processed on a service.    |
 | Requests TLS total    | Count     | `tls_version`, `tls_cipher`, `service`  | The total count of HTTPS requests processed on a service.   |
 | Request duration      | Histogram | `code`, `method`, `protocol`, `service` | Request processing duration histogram on a service.         |
-| Open connections      | Count     | `method`, `protocol`, `service`         | The current count of open connections on a service.         |
 | Retries total         | Count     | `service`                               | The count of requests retries on a service.                 |
 | Server UP             | Gauge     | `service`, `url`                        | Current service's server status, 0 for a down or 1 for up.  |
 | Requests bytes total  | Count     | `code`, `method`, `protocol`, `service` | The total size of requests in bytes received by a service.  |
@@ -184,7 +190,6 @@ traefik_router_responses_bytes_total
 traefik_service_requests_total
 traefik_service_requests_tls_total
 traefik_service_request_duration_seconds
-traefik_service_open_connections
 traefik_service_retries_total
 traefik_service_server_up
 traefik_service_requests_bytes_total
@@ -195,7 +200,6 @@ traefik_service_responses_bytes_total
 service.request.total
 router.service.tls.total
 service.request.duration
-service.connections.open
 service.retries.total
 service.server.up
 service.requests.bytes.total
@@ -206,7 +210,6 @@ service.responses.bytes.total
 traefik.service.requests.total
 traefik.service.requests.tls.total
 traefik.service.request.duration
-traefik.service.connections.open
 traefik.service.retries.total
 traefik.service.server.up
 traefik.service.requests.bytes.total
@@ -218,7 +221,6 @@ traefik.service.responses.bytes.total
 {prefix}.service.request.total
 {prefix}.service.request.tls.total
 {prefix}.service.request.duration
-{prefix}.service.connections.open
 {prefix}.service.retries.total
 {prefix}.service.server.up
 {prefix}.service.requests.bytes.total
@@ -236,7 +238,7 @@ traefik_service_requests_bytes_total
 traefik_service_responses_bytes_total
 ```
 
-## Labels
+### Labels
 
 Here is a comprehensive list of labels that are provided by the metrics:
 

diff --git a/pkg/metrics/datadog.go b/pkg/metrics/datadog.go
@@ -18,29 +18,28 @@ var (
 
 // Metric names consistent with https://github.com/DataDog/integrations-extras/pull/64
 const (
-	ddConfigReloadsName             = "config.reload.total"
-	ddLastConfigReloadSuccessName   = "config.reload.lastSuccessTimestamp"
+	ddConfigReloadsName           = "config.reload.total"
+	ddLastConfigReloadSuccessName = "config.reload.lastSuccessTimestamp"
+	ddOpenConnsName               = "open.connections"
+
 	ddTLSCertsNotAfterTimestampName = "tls.certs.notAfterTimestamp"
 
 	ddEntryPointReqsName        = "entrypoint.request.total"
 	ddEntryPointReqsTLSName     = "entrypoint.request.tls.total"
 	ddEntryPointReqDurationName = "entrypoint.request.duration"
-	ddEntryPointOpenConnsName   = "entrypoint.connections.open"
 	ddEntryPointReqsBytesName   = "entrypoint.requests.bytes.total"
 	ddEntryPointRespsBytesName  = "entrypoint.responses.bytes.total"
 
 	ddRouterReqsName         = "router.request.total"
 	ddRouterReqsTLSName      = "router.request.tls.total"
 	ddRouterReqsDurationName = "router.request.duration"
-	ddRouterOpenConnsName    = "router.connections.open"
 	ddRouterReqsBytesName    = "router.requests.bytes.total"
 	ddRouterRespsBytesName   = "router.responses.bytes.total"
 
 	ddServiceReqsName         = "service.request.total"
 	ddServiceReqsTLSName      = "service.request.tls.total"
 	ddServiceReqsDurationName = "service.request.duration"
 	ddServiceRetriesName      = "service.retries.total"
-	ddServiceOpenConnsName    = "service.connections.open"
 	ddServiceServerUpName     = "service.server.up"
 	ddServiceReqsBytesName    = "service.requests.bytes.total"
 	ddServiceRespsBytesName   = "service.responses.bytes.total"
@@ -63,6 +62,7 @@ func RegisterDatadog(ctx context.Context, config *types.Datadog) Registry {
 	registry := &standardRegistry{
 		configReloadsCounter:           datadogClient.NewCounter(ddConfigReloadsName, 1.0),
 		lastConfigReloadSuccessGauge:   datadogClient.NewGauge(ddLastConfigReloadSuccessName),
+		openConnectionsGauge:           datadogClient.NewGauge(ddOpenConnsName),
 		tlsCertsNotAfterTimestampGauge: datadogClient.NewGauge(ddTLSCertsNotAfterTimestampName),
 	}
 
@@ -71,7 +71,6 @@ func RegisterDatadog(ctx context.Context, config *types.Datadog) Registry {
 		registry.entryPointReqsCounter = datadogClient.NewCounter(ddEntryPointReqsName, 1.0)
 		registry.entryPointReqsTLSCounter = datadogClient.NewCounter(ddEntryPointReqsTLSName, 1.0)
 		registry.entryPointReqDurationHistogram, _ = NewHistogramWithScale(datadogClient.NewHistogram(ddEntryPointReqDurationName, 1.0), time.Second)
-		registry.entryPointOpenConnsGauge = datadogClient.NewGauge(ddEntryPointOpenConnsName)
 		registry.entryPointReqsBytesCounter = datadogClient.NewCounter(ddEntryPointReqsBytesName, 1.0)
 		registry.entryPointRespsBytesCounter = datadogClient.NewCounter(ddEntryPointRespsBytesName, 1.0)
 	}
@@ -81,7 +80,6 @@ func RegisterDatadog(ctx context.Context, config *types.Datadog) Registry {
 		registry.routerReqsCounter = datadogClient.NewCounter(ddRouterReqsName, 1.0)
 		registry.routerReqsTLSCounter = datadogClient.NewCounter(ddRouterReqsTLSName, 1.0)
 		registry.routerReqDurationHistogram, _ = NewHistogramWithScale(datadogClient.NewHistogram(ddRouterReqsDurationName, 1.0), time.Second)
-		registry.routerOpenConnsGauge = datadogClient.NewGauge(ddRouterOpenConnsName)
 		registry.routerReqsBytesCounter = datadogClient.NewCounter(ddRouterReqsBytesName, 1.0)
 		registry.routerRespsBytesCounter = datadogClient.NewCounter(ddRouterRespsBytesName, 1.0)
 	}
@@ -92,7 +90,6 @@ func RegisterDatadog(ctx context.Context, config *types.Datadog) Registry {
 		registry.serviceReqsTLSCounter = datadogClient.NewCounter(ddServiceReqsTLSName, 1.0)
 		registry.serviceReqDurationHistogram, _ = NewHistogramWithScale(datadogClient.NewHistogram(ddServiceReqsDurationName, 1.0), time.Second)
 		registry.serviceRetriesCounter = datadogClient.NewCounter(ddServiceRetriesName, 1.0)
-		registry.serviceOpenConnsGauge = datadogClient.NewGauge(ddServiceOpenConnsName)
 		registry.serviceServerUpGauge = datadogClient.NewGauge(ddServiceServerUpName)
 		registry.serviceReqsBytesCounter = datadogClient.NewCounter(ddServiceReqsBytesName, 1.0)
 		registry.serviceRespsBytesCounter = datadogClient.NewCounter(ddServiceRespsBytesName, 1.0)