feat(tracing)!: Improve control plane tracing configuration (#14557)

* chore(docs): Update tracing documentation

This clarifies the tracing documentation that the trace collector must be meshed, along with the specifics of how the service account name for the collector should be set.

Signed-off-by: Scott Fleener <[email protected]>

* chore(docs): Include `serviceAccountNamespace` for tracing collector

This separates out the service account namespace from the name to make it clearer how to set the correct service account for the trace collector.

Signed-off-by: Scott Fleener <[email protected]>

* chore(docs): Make docs clearer about what needs to be set for the collector mesh identity

Signed-off-by: Scott Fleener <[email protected]>

* feat!(tracing): Improve control plane tracing configuration

The current control plane tracing relies on the linkerd-jaeger extension, and does not work when using the native tracing configuration.

This removes the previous configs, and adds new control plane tracing config that mirrors the existing proxy tracing configs.

The previous configuration was meant entirely for internal testing purposes, and shouldn't be subject to any breaking change guarantees.

Signed-off-by: Scott Fleener <[email protected]>

* fix(test): Update tracing test values

Signed-off-by: Scott Fleener <[email protected]>

* feat(charts): Switch ServiceAccountName to separate Name and Namespace fields

Signed-off-by: Scott Fleener <[email protected]>

* fix: Change deprecated casing for lint

Signed-off-by: Scott Fleener <[email protected]>

* chore: Rename fields and semantics of tracing identity

Signed-off-by: Scott Fleener <[email protected]>

* chore: Fix bad auto-rename

Signed-off-by: Scott Fleener <[email protected]>

* chore: Change to breaking API change

Signed-off-by: Scott Fleener <[email protected]>

* chore: Regenerate goldens

Signed-off-by: Scott Fleener <[email protected]>

* fix: Use correct trace enable value

Signed-off-by: Scott Fleener <[email protected]>

* fix(docs): Typos in docs

Signed-off-by: Scott Fleener <[email protected]>

---------

Signed-off-by: Scott Fleener <[email protected]>

Author: sfleen

BUILD.md

@@ -198,25 +198,21 @@ bin/linkerd viz -n emojivoto tap deploy voting
 #### Deploying Control Plane components with Tracing
 
 Control Plane components have the `trace-collector` flag used to enable
-[Distributed Tracing](https://opentracing.io/docs/overview/what-is-tracing/) for
-development purposes. It can be enabled globally i.e Control plane components
-and their proxies by using the `--set controlPlaneTracing=true` installation
-flag.
+[Distributed Tracing](https://opentracing.io/docs/overview/what-is-tracing/)
+for development purposes. It can be enabled globally i.e Control plane
+components and their proxies by using the
+`--set controller.tracing.enable=true` installation flag.
 
-This will configure all the components to send the traces at
-`collector.{{.Values.controlPlaneTracingNamespace}}.svc.{{.Values.ClusterDomain}}:4317`
+This will configure all the components to send the traces to the collector you
+have configured for your cluster.
 
 ```bash
 
 # install Linkerd with tracing
-linkerd install --set controlPlaneTracing=true | kubectl apply -f -
-
-# install the Jaeger extension
-linkerd jaeger install | kubectl apply -f -
-
-# restart the control plane components so that the jaeger-injector enables
-# tracing in their proxies
-kubectl -n linkerd rollout restart deploy
+linkerd install \
+  --set controller.tracing.enable=true \
+  --set controller.tracing.collector.endpoint=<your trace collector endpoint> \
+  | kubectl apply -f -
 ```
 
 ### Publishing images

charts/linkerd-control-plane/values.yaml

@@ -19,10 +19,6 @@ revisionHistoryLimit: 10
 controllerLogLevel: info
 # -- Log format for the control plane components
 controllerLogFormat: plain
-# -- enables control plane tracing
-controlPlaneTracing: false
-# -- namespace to send control plane traces to
-controlPlaneTracingNamespace: linkerd-jaeger
 # -- control plane version. See Proxy section for proxy version
 linkerdVersion: linkerdVersionValue
 # -- default kubernetes deployment strategy
@@ -51,6 +47,13 @@ controller:
   podDisruptionBudget:
     # -- Maximum number of pods that can be unavailable during disruption
     maxUnavailable: 1
+  # Configures tracing in the controllers and how traces are exported
+  tracing:
+    # -- Enables trace collection and export in the proxy
+    enabled: false
+    collector:
+      # -- The collector endpoint to send traces to. Required if tracing is enabled
+      endpoint: ""
 # -- enabling this omits the NET_ADMIN capability in the PSP
 # and the proxy-init container when injecting the proxy;
 # requires the linkerd-cni plugin to already be installed

charts/partials/templates/_trace.tpl

@@ -1,5 +1,8 @@
 {{ define "partials.linkerd.trace" -}}
-{{ if .Values.controlPlaneTracing -}}
-- -trace-collector=collector.{{.Values.controlPlaneTracingNamespace}}.svc.{{.Values.clusterDomain}}:4317
+{{ if .Values.controller.tracing.enabled -}}
+{{- if empty .Values.controller.tracing.collector.endpoint }}
+{{- fail "controller.tracing.collector.endpoint must be set if control plane tracing is enabled" }}
+{{- end }}
+- -trace-collector={{.Values.controller.tracing.collector.endpoint}}
 {{ end -}}
 {{- end }}

cli/cmd/install_test.go

@@ -73,6 +73,7 @@ func TestRender(t *testing.T) {
 		CNIEnabled:              false,
 		IdentityTrustDomain:     defaultValues.IdentityTrustDomain,
 		IdentityTrustAnchorsPEM: defaultValues.IdentityTrustAnchorsPEM,
+		Controller:              defaultValues.Controller,
 		DestinationController:   defaultValues.DestinationController,
 		PodAnnotations:          map[string]string{},
 		PodLabels:               map[string]string{},
@@ -124,12 +125,12 @@ func TestRender(t *testing.T) {
 			Metrics: &charts.ProxyMetrics{
 				HostnameLabels: false,
 			},
-			Tracing: &charts.ProxyTracing{
+			Tracing: &charts.Tracing{
 				Enabled:          false,
 				TraceServiceName: "linkerd-proxy",
-				Collector: &charts.ProxyTracingCollector{
+				Collector: &charts.TracingCollector{
 					Endpoint: "",
-					MeshIdentity: &charts.ProxyTracingCollectorIdentity{
+					MeshIdentity: &charts.TracingCollectorIdentity{
 						ServiceAccountName: "",
 						Namespace:          "",
 					},
@@ -231,7 +232,12 @@ func TestRender(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Unexpected error: %v\n", err)
 	}
-	withControlPlaneTracingValues.ControlPlaneTracing = true
+	withControlPlaneTracingValues.Controller.Tracing = &charts.Tracing{
+		Enabled: true,
+		Collector: &charts.TracingCollector{
+			Endpoint: "tracing.foo:4317",
+		},
+	}
 	addFakeTLSSecrets(withControlPlaneTracingValues)
 
 	customRegistryOverride := "my.custom.registry/linkerd-io"

cli/cmd/options.go

@@ -114,18 +114,6 @@ func makeInstallUpgradeFlags(defaults *l5dcharts.Values) ([]flag.Flag, *pflag.Fl
 				return nil
 			}),
 
-		flag.NewBoolFlag(installUpgradeFlags, "control-plane-tracing", defaults.ControlPlaneTracing,
-			"Enables Control Plane Tracing with the defaults", func(values *l5dcharts.Values, value bool) error {
-				values.ControlPlaneTracing = value
-				return nil
-			}),
-
-		flag.NewStringFlag(installUpgradeFlags, "control-plane-tracing-namespace", defaults.ControlPlaneTracingNamespace,
-			"Send control plane traces to Linkerd-Jaeger extension in this namespace", func(values *l5dcharts.Values, value string) error {
-				values.ControlPlaneTracingNamespace = value
-				return nil
-			}),
-
 		flag.NewStringFlag(installUpgradeFlags, "identity-issuer-certificate-file", "",
 			"A path to a PEM-encoded file containing the Linkerd Identity issuer certificate (generated by default)",
 			func(values *l5dcharts.Values, value string) error {
@@ -181,8 +169,6 @@ func makeInstallUpgradeFlags(defaults *l5dcharts.Values) ([]flag.Flag, *pflag.Fl
 	if release {
 		installUpgradeFlags.MarkHidden("control-plane-version")
 	}
-	installUpgradeFlags.MarkHidden("control-plane-tracing")
-	installUpgradeFlags.MarkHidden("control-plane-tracing-namespace")
 
 	return flags, installUpgradeFlags, nil
 }