feat: add data aggregation level (microsoft#556)

# Description

This PR introduced the config map option for data aggregation. More
details can be found in the docs.
## Related Issue
microsoft#138

## Checklist

- [x] I have read the [contributing
documentation](https://retina.sh/docs/contributing).
- [x] I signed and signed-off the commits (`git commit -S -s ...`). See
[this
documentation](https://docs.github.com/en/authentication/managing-commit-signature-verification/about-commit-signature-verification)
on signing commits.
- [x] I have correctly attributed the author(s) of the code.
- [x] I have tested the changes locally.
- [x] I have followed the project's style guidelines.
- [x] I have updated the documentation, if necessary.
- [x] I have added tests, if applicable.

## Screenshots (if applicable) or Testing Completed
- Added unit tests for parsing data aggregation level from config file
- Deploy on cluster with different data aggregation level, verified that
on `high`, `packetparser` doesn't attach bpf program to eth0
## Additional Notes

Add any additional notes or context about the pull request here.

---

Please refer to the [CONTRIBUTING.md](../CONTRIBUTING.md) file for more
information on how to contribute to this project.

---------

Signed-off-by: Quang Nguyen <nguyenquang@microsoft.com>

cmd/hubble/daemon_main_linux.go

@@ -254,6 +254,7 @@ func setupZapLogger(retinaConfig *config.Config, k8sCfg *rest.Config) *log.ZapLo
 		zap.String("version", buildinfo.Version),
 		zap.String("apiserver", k8sCfg.Host),
 		zap.Strings("plugins", retinaConfig.EnabledPlugin),
+		zap.String("data aggregation level", retinaConfig.DataAggregationLevel.String()),
 	}
 
 	_, err := log.SetupZapLogger(logOpts, persistentFields...)

cmd/legacy/daemon.go

@@ -111,6 +111,7 @@ func (d *Daemon) Start() error {
 		zap.String("version", buildinfo.Version),
 		zap.String("apiserver", cfg.Host),
 		zap.String("plugins", strings.Join(daemonConfig.EnabledPlugin, `,`)),
+		zap.String("data aggregation level", daemonConfig.DataAggregationLevel.String()),
 	)
 	if err != nil {
 		panic(err)
@@ -120,6 +121,8 @@ func (d *Daemon) Start() error {
 
 	metrics.InitializeMetrics()
 
+	mainLogger.Info(zap.String("data aggregation level", daemonConfig.DataAggregationLevel.String()))
+
 	var tel telemetry.Telemetry
 	if daemonConfig.EnableTelemetry && buildinfo.ApplicationInsightsID != "" {
 		mainLogger.Info("telemetry enabled", zap.String("applicationInsightsID", buildinfo.ApplicationInsightsID))

controller/Dockerfile

@@ -3,19 +3,19 @@ ARG OS_VERSION
 # pinned base images
 
 # mcr.microsoft.com/oss/go/microsoft/golang:1.22.3-1-cbl-mariner2.0
-FROM --platform=$BUILDPLATFORM mcr.microsoft.com/oss/go/microsoft/golang@sha256:8253def0216b87b2994b7ad689aeec7440f6eb67f981e438071d8d67e36ff69f as golang
+FROM --platform=$BUILDPLATFORM mcr.microsoft.com/oss/go/microsoft/golang@sha256:8253def0216b87b2994b7ad689aeec7440f6eb67f981e438071d8d67e36ff69f AS golang
 
 # mcr.microsoft.com/cbl-mariner/base/core:2.0
-FROM --platform=$TARGETPLATFORM mcr.microsoft.com/cbl-mariner/base/core@sha256:77651116f2e83cf50fddd8a0316945499f8ce6521ff8e94e67539180d1e5975a as mariner-core
+FROM --platform=$TARGETPLATFORM mcr.microsoft.com/cbl-mariner/base/core@sha256:77651116f2e83cf50fddd8a0316945499f8ce6521ff8e94e67539180d1e5975a AS mariner-core
 
 # mcr.microsoft.com/cbl-mariner/distroless/minimal:2.0
-FROM --platform=$TARGETPLATFORM mcr.microsoft.com/cbl-mariner/distroless/minimal@sha256:63a0a70ceaa1320bc6eb98b81106667d43e46b674731ea8d28e4de1b87e0747f as mariner-distroless
+FROM --platform=$TARGETPLATFORM mcr.microsoft.com/cbl-mariner/distroless/minimal@sha256:63a0a70ceaa1320bc6eb98b81106667d43e46b674731ea8d28e4de1b87e0747f AS mariner-distroless
 
 # mcr.microsoft.com/windows/servercore:ltsc2019 
-FROM --platform=$TARGETPLATFORM mcr.microsoft.com/windows/servercore@sha256:6fdf140282a2f809dae9b13fe441635867f0a27c33a438771673b8da8f3348a4 as ltsc2019
+FROM --platform=$TARGETPLATFORM mcr.microsoft.com/windows/servercore@sha256:6fdf140282a2f809dae9b13fe441635867f0a27c33a438771673b8da8f3348a4 AS ltsc2019
 
 # mcr.microsoft.com/windows/servercore:ltsc2022
-FROM --platform=$TARGETPLATFORM mcr.microsoft.com/windows/servercore@sha256:45952938708fbde6ec0b5b94de68bcdec3f8c838be018536b1e9e5bd95e6b943 as ltsc2022
+FROM --platform=$TARGETPLATFORM mcr.microsoft.com/windows/servercore@sha256:45952938708fbde6ec0b5b94de68bcdec3f8c838be018536b1e9e5bd95e6b943 AS ltsc2022
 
 
 # build stages
@@ -103,7 +103,7 @@ RUN echo "Hubble version: $HUBBLE_VERSION" && \
     rm hubble-linux-${HUBBLE_ARCH}.tar.gz && rm hubble-linux-${HUBBLE_ARCH}.tar.gz.sha256sum
 
 # init final image
-FROM mariner-distroless as init
+FROM mariner-distroless AS init
 COPY --from=init-bin /go/bin/retina/initretina /retina/initretina
 COPY --from=tools /lib/ /lib
 COPY --from=tools /usr/lib/ /usr/lib
@@ -113,7 +113,7 @@ ENTRYPOINT ["./retina/initretina"]
 # agent final image
 # mcr.microsoft.com/cbl-mariner/distroless/minimal:2.0
 # mcr.microsoft.com/cbl-mariner/distroless/minimal@sha256:63a0a70ceaa1320bc6eb98b81106667d43e46b674731ea8d28e4de1b87e0747f
-FROM mariner-distroless as agent
+FROM mariner-distroless AS agent
 COPY --from=tools /lib/ /lib
 COPY --from=tools /usr/lib/ /usr/lib
 COPY --from=tools /tmp/bin/ /bin
@@ -128,7 +128,7 @@ ENTRYPOINT ["./retina/controller"]
 
 
 # agent final image for windows 
-FROM ${OS_VERSION} as agent-win
+FROM ${OS_VERSION} AS agent-win
 COPY --from=controller-bin /go/src/github.com/microsoft/retina/windows/kubeconfigtemplate.yaml kubeconfigtemplate.yaml
 COPY --from=controller-bin /go/src/github.com/microsoft/retina/windows/setkubeconfigpath.ps1 setkubeconfigpath.ps1
 COPY --from=controller-bin /go/bin/retina/controller controller.exe

controller/Dockerfile.windows-2019

@@ -1,4 +1,4 @@
-FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.21 as builder
+FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.21 AS builder
 # Build args
 ARG VERSION
 ARG APP_INSIGHTS_ID

controller/Dockerfile.windows-2022

@@ -1,4 +1,6 @@
-FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.21 as builder
+FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.21 AS builder
+
+# Build args
 ARG VERSION
 ARG APP_INSIGHTS_ID
 

controller/Dockerfile.windows-cgo

@@ -1,4 +1,4 @@
-FROM  --platform=windows/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22.2-windowsservercore-ltsc2022 as cgo
+FROM  --platform=windows/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22.2-windowsservercore-ltsc2022 AS cgo
 
 SHELL ["powershell", "-Command", "$ErrorActionPreference = 'Stop'; $ProgressPreference = 'SilentlyContinue';"]
 

controller/Dockerfile.windows-native

@@ -3,7 +3,7 @@
 # buildx targets, and this one requires legacy build.
 # Maybe one day: https://github.com/moby/buildkit/issues/616
 ARG BUILDER_IMAGE
-FROM  --platform=windows/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22-windowsservercore-ltsc2022 as builder
+FROM  --platform=windows/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22-windowsservercore-ltsc2022 AS builder
 WORKDIR C:\\retina
 COPY go.mod .
 COPY go.sum .
@@ -22,7 +22,7 @@ RUN go build -v -o captureworkload.exe -ldflags="-X github.com/microsoft/retina/
 FROM --platform=windows/amd64 ${BUILDER_IMAGE} as pktmon-builder
 WORKDIR C:\\retina
 
-FROM --platform=windows/amd64 mcr.microsoft.com/windows/nanoserver:ltsc2022 as final
+FROM --platform=windows/amd64 mcr.microsoft.com/windows/nanoserver:ltsc2022 AS final
 ADD https://github.com/microsoft/etl2pcapng/releases/download/v1.10.0/etl2pcapng.exe /etl2pcapng.exe
 SHELL ["powershell", "-Command", "$ErrorActionPreference = 'Stop'; $ProgressPreference = 'Continue';"]
 COPY --from=builder C:\\retina\\controller.exe controller.exe

deploy/hubble/manifests/controller/helm/retina/templates/agent/configmap.yaml

@@ -110,6 +110,7 @@ data:
     remoteContext: {{ .Values.remoteContext }}
     enableAnnotations: {{ .Values.enableAnnotations }}
     bypassLookupIPOfInterest: {{ .Values.bypassLookupIPOfInterest }}
+    dataAggregationLevel: {{ .Values.dataAggregationLevel }}
 {{- end}}
 ---
 {{- if .Values.os.windows}}

deploy/hubble/manifests/controller/helm/retina/values.yaml

@@ -60,6 +60,7 @@ enablePodLevel: true
 remoteContext: false
 enableAnnotations: false
 bypassLookupIPOfInterest: true
+dataAggregationLevel: "high"
 
 imagePullSecrets: []
 nameOverride: "retina"
@@ -178,7 +179,7 @@ hubble:
   # Possible values are:
   #   1, 3, 7, 15, 31, 63, 127, 255, 511, 1023,
   #   2047, 4095, 8191, 16383, 32767, 65535
-  # eventBufferCapacity: "4095"
+  eventBufferCapacity: "4095"
 
   # -- Hubble metrics configuration.
   # See https://docs.cilium.io/en/stable/observability/metrics/#hubble-metrics

deploy/legacy/manifests/controller/helm/retina/templates/configmap.yaml

@@ -20,6 +20,8 @@ data:
     enablePodLevel: {{ .Values.enablePodLevel }}
     remoteContext: {{ .Values.remoteContext }}
     enableAnnotations: {{ .Values.enableAnnotations }}
+    bypassLookupIPOfInterest: {{ .Values.bypassLookupIPOfInterest }}
+    dataAggregationLevel: {{ .Values.dataAggregationLevel }}
 {{- end}}
 ---
 {{- if .Values.os.windows}}

deploy/legacy/manifests/controller/helm/retina/values.yaml

@@ -41,6 +41,7 @@ enablePodLevel: false
 remoteContext: false
 enableAnnotations: false
 bypassLookupIPOfInterest: false
+dataAggregationLevel: "low"
 
 imagePullSecrets: []
 nameOverride: "retina"

docs/concepts/dataAggregation.md

@@ -0,0 +1,7 @@
+# Data Aggregation
+
+Under Retina's hood, data are communicate between plugins and the control plane via [`Flow` objects](https://github.com/cilium/cilium/tree/main/api/v1/flow). Retina's data aggregation settings are designed to manage the amount of data that can be potentially generate by the agent .i.e the number of `flows` being generated. At a higher aggregation level, fewer `flows` objects are produced, which ensures resource efficiency in large clusters. Conversely, a lower level of aggregation results in more `flow` objects being generated, offering more detailed information regarding packets being observed at different points in the Linux kernel.The operational behaviors of Retina at each aggregation level are detailed in the table below:
+| Level  | Description|
+|---    |---   |
+| `low` | `packetparser` will attach a bpf program to the node's default interface in the node namespace, which will help capture metrics for `TO_NETWORK` and `FROM_NETWORK` packets. This will give users a more granular view of packet flows and offers more reliable apiserver latency metrics. |
+| `high` | `packetparser` will not attach a bpf program to the node's default interface in the node namespace. As a result, packet observation at this location will be disabled, leading to a reduction in metrics being generated. This configuration is recommended when scalability is the primary concern. However, it is important to note that, due to the absence of packet observation at the default interface, the apiserver latency metrics may not be as reliable. |

docs/installation/config.md

@@ -13,6 +13,7 @@ Defaults are specified for each component in *deploy/legacy/manifests/controller
 * `enabledPlugin_linux`: Array of enabled plugins for linux.
 * `enabledPlugin_win`: Array of enabled plugins for windows.
 * `metricsInterval`: the interval for which metrics will be gathered.
+* `dataAggregationLevel`: This config defines the level of data aggregation for Retina. See [Data Aggregation](../concepts/dataAggregation.md) for more details.
 
 ## Operator Config
 

operator/Dockerfile

@@ -21,7 +21,7 @@ RUN --mount=type=cache,target="/root/.cache/go-build" \
 	-a -o retina-operator operator/main.go
 
 ##################### controller #######################
-FROM --platform=$BUILDPLATFORM mcr.microsoft.com/mirror/gcr/distroless/cc-debian11:latest@sha256:b53fbf5f81f4a120a489fedff2092e6fcbeacf7863fce3e45d99cc58dc230ccc as controller
+FROM --platform=$BUILDPLATFORM mcr.microsoft.com/mirror/gcr/distroless/cc-debian11:latest@sha256:b53fbf5f81f4a120a489fedff2092e6fcbeacf7863fce3e45d99cc58dc230ccc AS controller
 
 WORKDIR /
 COPY --from=builder /workspace/retina-operator .

operator/Dockerfile.windows-2019

@@ -1,4 +1,4 @@
-FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22 as builder
+FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22 AS builder
 
 # Build args
 ARG VERSION

operator/Dockerfile.windows-2022

@@ -1,4 +1,4 @@
-FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22 as builder
+FROM --platform=linux/amd64 mcr.microsoft.com/oss/go/microsoft/golang:1.22 AS builder
 
 # Build args
 ARG VERSION

pkg/config/config.go

@@ -4,18 +4,54 @@ package config
 
 import (
 	"fmt"
+	"reflect"
+	"strings"
 	"time"
 
+	"github.com/mitchellh/mapstructure"
 	"github.com/spf13/viper"
 )
 
+// Level defines the level of monitor aggregation.
+type Level int
+
+const (
+	Low Level = iota
+	High
+)
+
+func (l *Level) UnmarshalText(text []byte) error {
+	s := strings.ToLower(string(text))
+	switch s {
+	case "low":
+		*l = Low
+	case "high":
+		*l = High
+	default:
+		// Default to Low if the text is not recognized.
+		*l = Low
+	}
+	return nil
+}
+
+func (l *Level) String() string {
+	switch *l {
+	case Low:
+		return "low"
+	case High:
+		return "high"
+	default:
+		return ""
+	}
+}
+
 type Server struct {
 	Host string `yaml:"host"`
 	Port int    `yaml:"port"`
 }
 
 type Config struct {
-	ApiServer                Server        `yaml:"apiServer"`
+	APIServer                Server        `yaml:"apiServer"`
 	LogLevel                 string        `yaml:"logLevel"`
 	EnabledPlugin            []string      `yaml:"enabledPlugin"`
 	MetricsInterval          time.Duration `yaml:"metricsInterval"`
@@ -25,6 +61,7 @@ type Config struct {
 	RemoteContext            bool          `yaml:"remoteContext"`
 	EnableAnnotations        bool          `yaml:"enableAnnotations"`
 	BypassLookupIPOfInterest bool          `yaml:"bypassLookupIPOfInterest"`
+	DataAggregationLevel     Level         `yaml:"dataAggregationLevel"`
 }
 
 func GetConfig(cfgFilename string) (*Config, error) {
@@ -46,7 +83,14 @@ func GetConfig(cfgFilename string) (*Config, error) {
 		return nil, fmt.Errorf("fatal error config file: %s", err)
 	}
 	var config Config
-	err = viper.Unmarshal(&config)
+	decoderConfigOption := func(dc *mapstructure.DecoderConfig) {
+		dc.DecodeHook = mapstructure.ComposeDecodeHookFunc(
+			mapstructure.StringToTimeDurationHookFunc(), // default hook.
+			mapstructure.StringToSliceHookFunc(","),     // default hook.
+			decodeLevelHook,
+		)
+	}
+	err = viper.Unmarshal(&config, decoderConfigOption)
 	if err != nil {
 		return nil, fmt.Errorf("fatal error config file: %s", err)
 	}
@@ -55,3 +99,20 @@ func GetConfig(cfgFilename string) (*Config, error) {
 
 	return &config, nil
 }
+
+func decodeLevelHook(field, target reflect.Type, data interface{}) (interface{}, error) {
+	// Check if the field we are decoding is a string.
+	if field.Kind() != reflect.String {
+		return data, nil
+	}
+	// Check if the type we are decoding to is a Level.
+	if target != reflect.TypeOf(Level(0)) {
+		return data, nil
+	}
+	var level Level
+	err := level.UnmarshalText([]byte(data.(string)))
+	if err != nil {
+		return nil, err
+	}
+	return level, nil
+}

pkg/config/config_test.go

@@ -4,24 +4,48 @@
 package config
 
 import (
+	"reflect"
 	"testing"
 	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestGetConfig(t *testing.T) {
 	c, err := GetConfig("./testwith/config.yaml")
 	if err != nil {
 		t.Fatalf("Expected no error, instead got %+v", err)
 	}
-	if c.ApiServer.Host != "0.0.0.0" ||
-		c.ApiServer.Port != 10093 ||
+	if c.APIServer.Host != "0.0.0.0" ||
+		c.APIServer.Port != 10093 ||
 		c.LogLevel != "info" ||
 		c.MetricsInterval != 10*time.Second ||
 		len(c.EnabledPlugin) != 3 ||
 		c.EnablePodLevel ||
 		!c.EnableRetinaEndpoint ||
 		c.RemoteContext ||
-		c.EnableAnnotations {
+		c.EnableAnnotations ||
+		c.DataAggregationLevel != Low {
 		t.Fatalf("Expeted config should be same as ./testwith/config.yaml; instead got %+v", c)
 	}
 }
+
+func TestDecodeLevelHook(t *testing.T) {
+	tests := []struct {
+		input    interface{}
+		expected interface{}
+	}{
+		{"low", Low},
+		{"high", High},
+		{"invalid", Low}, // Unimplemented or invalid input should default to Low
+		{123, 123},       // Non-string input should be returned as is
+	}
+
+	for _, test := range tests {
+		result, err := decodeLevelHook(reflect.TypeOf(test.input), reflect.TypeOf(Level(0)), test.input)
+		require.NoError(t, err)
+		assert.Equal(t, test.expected, result)
+
+	}
+}

pkg/config/hubble_config_linux.go

@@ -50,6 +50,7 @@ var (
 		EnablePodLevel:           true,
 		LogLevel:                 "info",
 		BypassLookupIPOfInterest: true,
+		DataAggregationLevel:     High,
 	}
 
 	Cell = cell.Module(

pkg/config/testwith/config.yaml

@@ -8,3 +8,4 @@ enabledPlugin: ["dropreason", "packetforward", "linuxutil"]
 metricsInterval: 10
 # used to export telemetry to AppInsights
 telemetryEnabled: true
+dataAggregationLevel: "low"

pkg/managers/controllermanager/controllermanager.go

@@ -62,8 +62,8 @@ func NewControllerManager(conf *kcfg.Config, kubeclient kubernetes.Interface, te
 
 	// create HTTP server for API server
 	httpServer := sm.NewHTTPServer(
-		conf.ApiServer.Host,
-		conf.ApiServer.Port,
+		conf.APIServer.Host,
+		conf.APIServer.Port,
 	)
 
 	return &Controller{

pkg/plugin/dropreason/dropreason_linux_test.go

@@ -30,6 +30,7 @@ var (
 		MetricsInterval:          1 * time.Second,
 		EnablePodLevel:           true,
 		BypassLookupIPOfInterest: true,
+		DataAggregationLevel:     kcfg.Low,
 	}
 	cfgPodLevelDisabled = &kcfg.Config{
 		MetricsInterval: 1 * time.Second,