Add profiling tools (#1087)

* Add gin-contrib/pprof dependency

* Run pprof server on agent with --profiler flag

* Add --profiler flag to cli

* Fix error message

* Print cpu usage percentage

* measure cpu of current pid instead of globaly on the system

* Add scripts to plot performance

* Plot packetsCount in analysis

* Concat to DataFrame

* Plot in turbo colorscheme

* Make COLORMAP const

* Fix rss units

* Reduce code repetition by adding function for plotting

* Allow grouping based on filenames

* Temporary: Marked with comments where to disable code for experiments

* Add newline at end of file

* Add tap.cpuprofile flag. Change memprofile flag to tap.memprofile

* create tapper modes for debugging using env vars

* Fix rss plot units (MB instead of bytes)

* Remove comment

* Add info to plot script

* Remove tap.cpumemprofile. Rename tap.memprofile to memprofile

* Remove unused import

* Remove whitespaces

Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

* Remove whitespaces

Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

* Remove whitespaces

Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

* Remove whitespaces

Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

* Remove whitespaces

Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

* Remove whitespaces

Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

* Rename debug env vars

* Create package for debug env vars, read each env var once

* Run go mod tidy

* Increment MatchedPairs before emitting

* Only count cores once

* Count virtual and physical cores

* Add dbgctl replace in cli

* Fix lint: Check return values

* Add tap/dbgctl to test-lint make rule

* Replace tap/dbgctl in all modules

* #run_acceptance_tests

* Copy dbgctl module to docker image

* Debug/profile tapper benchmark (#1093)

* add mizu debug env to avoid all extensions

* add readme + run_tapper_benchmark.sh

* temporary change branch name

* fix readme

* fix MIZU_BENCHMARK_CLIENTS_COUNT env

* change tap target to tcp stream

* track live tcp streams

* pr fixes

* rename tapperPacketsCount to ignored_packets_count

* change mizu tapper to mizu debugg

Co-authored-by: David Levanon <dvdlevanon@gmail.com>
Co-authored-by: M. Mert Yıldıran <mehmet@up9.com>

cli/cmd/tap.go

@@ -123,4 +123,5 @@ func init() {
 	tapCmd.Flags().String(configStructs.ContractFile, defaultTapConfig.ContractFile, "OAS/Swagger file to validate to monitor the contracts")
 	tapCmd.Flags().Bool(configStructs.ServiceMeshName, defaultTapConfig.ServiceMesh, "Record decrypted traffic if the cluster is configured with a service mesh and with mtls")
 	tapCmd.Flags().Bool(configStructs.TlsName, defaultTapConfig.Tls, "Record tls traffic")
+	tapCmd.Flags().Bool(configStructs.ProfilerName, defaultTapConfig.Profiler, "Run pprof server")
 }

cli/cmd/tapRunner.go

@@ -124,7 +124,7 @@ func RunMizuTap() {
 	}
 
 	logger.Log.Infof("Waiting for Mizu Agent to start...")
-	if state.mizuServiceAccountExists, err = resources.CreateTapMizuResources(ctx, kubernetesProvider, serializedValidationRules, serializedContract, serializedMizuConfig, config.Config.IsNsRestrictedMode(), config.Config.MizuResourcesNamespace, config.Config.AgentImage, getSyncEntriesConfig(), config.Config.Tap.MaxEntriesDBSizeBytes(), config.Config.Tap.ApiServerResources, config.Config.ImagePullPolicy(), config.Config.LogLevel()); err != nil {
+	if state.mizuServiceAccountExists, err = resources.CreateTapMizuResources(ctx, kubernetesProvider, serializedValidationRules, serializedContract, serializedMizuConfig, config.Config.IsNsRestrictedMode(), config.Config.MizuResourcesNamespace, config.Config.AgentImage, getSyncEntriesConfig(), config.Config.Tap.MaxEntriesDBSizeBytes(), config.Config.Tap.ApiServerResources, config.Config.ImagePullPolicy(), config.Config.LogLevel(), config.Config.Tap.Profiler); err != nil {
 		var statusError *k8serrors.StatusError
 		if errors.As(err, &statusError) && (statusError.ErrStatus.Reason == metav1.StatusReasonAlreadyExists) {
 			logger.Log.Info("Mizu is already running in this namespace, change the `mizu-resources-namespace` configuration or run `mizu clean` to remove the currently running Mizu instance")

cli/config/configStructs/tapConfig.go

@@ -30,6 +30,7 @@ const (
 	ContractFile                  = "contract"
 	ServiceMeshName               = "service-mesh"
 	TlsName                       = "tls"
+	ProfilerName                  = "profiler"
 )
 
 type TapConfig struct {
@@ -54,6 +55,7 @@ type TapConfig struct {
 	TapperResources        shared.Resources `yaml:"tapper-resources"`
 	ServiceMesh            bool             `yaml:"service-mesh" default:"false"`
 	Tls                    bool             `yaml:"tls" default:"false"`
+	Profiler               bool             `yaml:"profiler" default:"false"`
 }
 
 func (config *TapConfig) PodRegex() *regexp.Regexp {

cli/go.mod

@@ -74,6 +74,7 @@ require (
 	github.com/pmezard/go-difflib v1.0.0 // indirect
 	github.com/russross/blackfriday v1.6.0 // indirect
 	github.com/stretchr/testify v1.7.0 // indirect
+	github.com/up9inc/mizu/tap/dbgctl v0.0.0 // indirect
 	github.com/xlab/treeprint v1.1.0 // indirect
 	go.starlark.net v0.0.0-20220203230714-bb14e151c28f // indirect
 	golang.org/x/crypto v0.0.0-20220208050332-20e1d8d225ab // indirect
@@ -104,3 +105,5 @@ replace github.com/up9inc/mizu/logger v0.0.0 => ../logger
 replace github.com/up9inc/mizu/shared v0.0.0 => ../shared
 
 replace github.com/up9inc/mizu/tap/api v0.0.0 => ../tap/api
+
+replace github.com/up9inc/mizu/tap/dbgctl v0.0.0 => ../tap/dbgctl

cli/resources/createResources.go

@@ -14,7 +14,7 @@ import (
 	core "k8s.io/api/core/v1"
 )
 
-func CreateTapMizuResources(ctx context.Context, kubernetesProvider *kubernetes.Provider, serializedValidationRules string, serializedContract string, serializedMizuConfig string, isNsRestrictedMode bool, mizuResourcesNamespace string, agentImage string, syncEntriesConfig *shared.SyncEntriesConfig, maxEntriesDBSizeBytes int64, apiServerResources shared.Resources, imagePullPolicy core.PullPolicy, logLevel logging.Level) (bool, error) {
+func CreateTapMizuResources(ctx context.Context, kubernetesProvider *kubernetes.Provider, serializedValidationRules string, serializedContract string, serializedMizuConfig string, isNsRestrictedMode bool, mizuResourcesNamespace string, agentImage string, syncEntriesConfig *shared.SyncEntriesConfig, maxEntriesDBSizeBytes int64, apiServerResources shared.Resources, imagePullPolicy core.PullPolicy, logLevel logging.Level, profiler bool) (bool, error) {
 	if !isNsRestrictedMode {
 		if err := createMizuNamespace(ctx, kubernetesProvider, mizuResourcesNamespace); err != nil {
 			return false, err
@@ -50,6 +50,7 @@ func CreateTapMizuResources(ctx context.Context, kubernetesProvider *kubernetes.
 		Resources:             apiServerResources,
 		ImagePullPolicy:       imagePullPolicy,
 		LogLevel:              logLevel,
+		Profiler:              profiler,
 	}
 
 	if err := createMizuApiServerPod(ctx, kubernetesProvider, opts); err != nil {