Resolved merge conflicts

Author: tanuja-sunda73

cmd/csi_driver/main.go

@@ -28,6 +28,7 @@ import (
 	"time"
 
 	"cloud.google.com/go/profiler"
+	"github.com/KimMachineGun/automemlimit/memlimit"
 	"github.com/googlecloudplatform/gcs-fuse-csi-driver/pkg/cloud_provider/auth"
 	"github.com/googlecloudplatform/gcs-fuse-csi-driver/pkg/cloud_provider/clientset"
 	"github.com/googlecloudplatform/gcs-fuse-csi-driver/pkg/cloud_provider/metadata"
@@ -70,6 +71,8 @@ var (
 	enableSidecarBucketAccessCheck = flag.Bool("enable-sidecar-bucket-access-check", false, "Enable bucket access check on sidecar, this does not disable bucket access check in node driver.")
 	enableCloudProfilerForDriver   = flag.Bool("enable-cloud-profiler-for-driver", false, "Enable cloud profiler to collect analysis data.")
 	assumeGoodSidecarVersion       = flag.Bool("assume-good-sidecar-version", false, "Assume the sidecar version is compatible with all features in the running version of the driver.")
+	enableAutoGoMemLimit           = flag.Bool("enable-auto-gomemlimit", false, "Automatically set GOMEMLIMIT to a percentage of the container's cgroup memory limit.")
+	autoGoMemLimitRatio            = flag.Float64("auto-gomemlimit-ratio", util.GoMemLimitCgroupPercentage, "The ratio of the container's cgroup memory limit to set as GOMEMLIMIT when enable-auto-gomemlimit is enabled.")
 
 	// GCSFuse kernel params feature.
 	enableGCSFuseKernelParams = flag.Bool("enable-gcsfuse-kernel-params", false, "Enable gcsfuse kernel params feature.")
@@ -101,6 +104,20 @@ func main() {
 	klog.InitFlags(nil)
 	flag.Parse()
 
+	if *enableAutoGoMemLimit {
+		// Sets GOMEMLIMIT to a percentage of the cgroup limit.
+		// If cgroup limits are unavailable, it falls back to doing nothing
+		// (no limit).
+		if _, err := memlimit.SetGoMemLimitWithOpts(
+			memlimit.WithRatio(*autoGoMemLimitRatio),
+			memlimit.WithProvider(memlimit.FromCgroup),
+		); err != nil {
+			// This can happen if the gcs-fuse-csi-driver container memory
+			// limit is not set.
+			klog.Warningf("Failed to automatically set GOMEMLIMIT: %v", err)
+		}
+	}
+
 	// All CSI sidecars use http-endpoint for metrics and health checks.
 	// Example: https://gke-internal.googlesource.com/third_party/kubernetes-csi/livenessprobe/+/refs/heads/master/cmd/livenessprobe/main.go#113
 	// At some point, we should replace "metrics-endpoint" with "http-endpoint".
@@ -221,6 +238,10 @@ func main() {
 			EnableGcsfuseProfilesInternal: *enableGcsfuseProfilesInternal,
 		},
 		EnableGCSFuseKernelParams: *enableGCSFuseKernelParams,
+		GoMemLimitOptions: &driver.GoMemLimitOptions{
+			EnableAutoGoMemLimit: *enableAutoGoMemLimit,
+			AutoGoMemLimitRatio:  *autoGoMemLimitRatio,
+		},
 	}
 
 	var mounter mount.Interface

cmd/sidecar_mounter/gcsfuse_binary

@@ -1 +1 @@
-gs://gke-release-staging/gcsfuse/v3.9.0-gke.0/gcsfuse_bin
+gs://gke-release-staging/gcsfuse/v3.9.1-gke.0/gcsfuse_bin

cmd/webhook/main.go

@@ -22,7 +22,9 @@ import (
 	"net/http"
 	"time"
 
+	"github.com/KimMachineGun/automemlimit/memlimit"
 	"github.com/go-logr/logr"
+	"github.com/googlecloudplatform/gcs-fuse-csi-driver/pkg/util"
 	wh "github.com/googlecloudplatform/gcs-fuse-csi-driver/pkg/webhook"
 	"k8s.io/apimachinery/pkg/runtime"
 	"k8s.io/apimachinery/pkg/util/version"
@@ -56,6 +58,8 @@ var (
 	metadataSidecarImage                    = flag.String("metadata-sidecar-image", "", "The metadata prefetch sidecar container image.")
 	injectSAVol                             = flag.Bool("should-inject-sa-vol", false, "Inject projected service account volume when true")
 	enableGcsfuseProfiles                   = flag.Bool("enable-gcsfuse-profiles", false, "Enable gcsfuse profiles when true")
+	enableAutoGoMemLimit                    = flag.Bool("enable-auto-gomemlimit", false, "Automatically set GOMEMLIMIT to a percentage of the container's cgroup memory limit.")
+	autoGoMemLimitRatio                     = flag.Float64("auto-gomemlimit-ratio", util.GoMemLimitCgroupPercentage, "The ratio of the container's cgroup memory limit to set as GOMEMLIMIT when enable-auto-gomemlimit is enabled.")
 	metadataMemoryRequest                   = flag.String("metadata-sidecar-memory-request", "10Mi", "Flag to use default value for gcsfuse memory prefetch sidecar container memory request.")
 	metadataMemoryLimit                     = flag.String("metadata-sidecar-memory-limit", "10Mi", "Flag to use default value for gcsfuse memory prefetch sidecar container memory limit.")
 	metadataPrefetchCPURequest              = flag.String("metadata-sidecar-cpu-request", "10m", "The default cpu request for gcsfuse memory prefetch sidecar container cpu request.")
@@ -78,6 +82,20 @@ func main() {
 	// This line prevents controller-runtime from complaining about log.SetLogger never being called
 	log.SetLogger(logr.New(log.NullLogSink{}))
 
+	if *enableAutoGoMemLimit {
+		// Sets GOMEMLIMIT to a percentage of the cgroup limit.
+		// If cgroup limits are unavailable, it falls back to doing nothing
+		// (no limit).
+		if _, err := memlimit.SetGoMemLimitWithOpts(
+			memlimit.WithRatio(*autoGoMemLimitRatio),
+			memlimit.WithProvider(memlimit.FromCgroup),
+		); err != nil {
+			// This can happen if the gcs-fuse-csi-driver-webhook container
+			// memory limit is not set.
+			klog.Warningf("Failed to automatically set GOMEMLIMIT: %v", err)
+		}
+	}
+
 	klog.Infof("Running Google Cloud Storage FUSE CSI driver admission webhook version %v, sidecar container image %v", webhookVersion, *sidecarImage)
 
 	// Load webhook config

deploy/base/controller/controller.yaml

@@ -45,6 +45,8 @@ spec:
             - "--cluster-location=$(CLUSTER_LOCATION)"
             - "--project-number=$(PROJECT_NUMBER)"
             - "--http-endpoint=:29633"
+            - "--enable-auto-gomemlimit=true"
+            - "--auto-gomemlimit-ratio=0.95"
           ports:
             - containerPort: 29633
               name: http-endpoint

deploy/base/node/node.yaml

@@ -62,6 +62,8 @@ spec:
             - --enable-gcsfuse-profiles-internal=true
             - --stream-metrics-export=true
             - --enable-cloud-profiler-for-driver=true
+            - --enable-auto-gomemlimit=true
+            - --auto-gomemlimit-ratio=0.95
           ports:
             - containerPort: 9920
               name: metrics

deploy/base/webhook/deployment.yaml

@@ -61,6 +61,8 @@ spec:
             - --health-probe-bind-address=:22031
             - --should-inject-sa-vol=true
             - --enable-gcsfuse-profiles=true
+            - --enable-auto-gomemlimit=true
+            - --auto-gomemlimit-ratio=0.95
           env:
             - name: SIDECAR_IMAGE_PULL_POLICY
               value: "IfNotPresent"

go.mod

@@ -8,6 +8,7 @@ require (
 	cloud.google.com/go/monitoring v1.24.2
 	cloud.google.com/go/profiler v0.4.3
 	cloud.google.com/go/storage v1.56.3
+	github.com/KimMachineGun/automemlimit v0.7.5
 	github.com/container-storage-interface/spec v1.11.0
 	github.com/distribution/reference v0.6.0
 	github.com/go-logr/logr v1.4.3
@@ -89,6 +90,7 @@ require (
 	github.com/onsi/ginkgo/v2 v2.27.0 // indirect
 	github.com/onsi/gomega v1.38.2 // indirect
 	github.com/opencontainers/go-digest v1.0.0 // indirect
+	github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
 	github.com/planetscale/vtprotobuf v0.6.1-0.20240319094008-0393e58bdf10 // indirect
 	github.com/prometheus/procfs v0.15.1 // indirect

go.sum

@@ -30,6 +30,8 @@ github.com/GoogleCloudPlatform/opentelemetry-operations-go/internal/cloudmock v0
 github.com/GoogleCloudPlatform/opentelemetry-operations-go/internal/cloudmock v0.53.0/go.mod h1:jUZ5LYlw40WMd07qxcQJD5M40aUxrfwqQX1g7zxYnrQ=
 github.com/GoogleCloudPlatform/opentelemetry-operations-go/internal/resourcemapping v0.53.0 h1:Ron4zCA/yk6U7WOBXhTJcDpsUBG9npumK6xw2auFltQ=
 github.com/GoogleCloudPlatform/opentelemetry-operations-go/internal/resourcemapping v0.53.0/go.mod h1:cSgYe11MCNYunTnRXrKiR/tHc0eoKjICUuWpNZoVCOo=
+github.com/KimMachineGun/automemlimit v0.7.5 h1:RkbaC0MwhjL1ZuBKunGDjE/ggwAX43DwZrJqVwyveTk=
+github.com/KimMachineGun/automemlimit v0.7.5/go.mod h1:QZxpHaGOQoYvFhv/r4u3U0JTC2ZcOwbSr11UZF46UBM=
 github.com/Masterminds/semver/v3 v3.4.0 h1:Zog+i5UMtVoCU8oKka5P7i9q9HgrJeGzI9SA1Xbatp0=
 github.com/Masterminds/semver/v3 v3.4.0/go.mod h1:4V+yj/TJE1HU9XfppCwVMZq3I84lprf4nC11bSS5beM=
 github.com/antlr4-go/antlr/v4 v4.13.0 h1:lxCg3LAv+EUK6t1i0y1V6/SLeUi0eKEKdhQAlS8TVTI=
@@ -157,6 +159,8 @@ github.com/onsi/gomega v1.38.2 h1:eZCjf2xjZAqe+LeWvKb5weQ+NcPwX84kqJ0cZNxok2A=
 github.com/onsi/gomega v1.38.2/go.mod h1:W2MJcYxRGV63b418Ai34Ud0hEdTVXq9NW9+Sx6uXf3k=
 github.com/opencontainers/go-digest v1.0.0 h1:apOUWs51W5PlhuyGyz9FCeeBIOUDA/6nW8Oi/yOhh5U=
 github.com/opencontainers/go-digest v1.0.0/go.mod h1:0JzlMkj0TRzQZfJkVvzbP0HBR3IKzErnv2BNG4W4MAM=
+github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58 h1:onHthvaw9LFnH4t2DcNVpwGmV9E1BkGknEliJkfwQj0=
+github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58/go.mod h1:DXv8WO4yhMYhSNPKjeNKa5WY9YCIEBRbNzFFPJbWO6Y=
 github.com/pkg/errors v0.9.1 h1:FEBLx1zS214owpjy7qsBeixbURkuhQAwrK5UwLGTwt4=
 github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
 github.com/planetscale/vtprotobuf v0.6.1-0.20240319094008-0393e58bdf10 h1:GFCKgmp0tecUJ0sJuv4pzYCqS9+RGSn52M3FUwPs+uo=

pkg/csi_driver/gcs_fuse_driver.go

@@ -45,9 +45,15 @@ type FeatureGCSFuseProfiles struct {
 	EnableGcsfuseProfilesInternal bool
 }
 
+type GoMemLimitOptions struct {
+	EnableAutoGoMemLimit bool
+	AutoGoMemLimitRatio  float64
+}
+
 type GCSDriverFeatureOptions struct {
 	EnableGCSFuseKernelParams bool
 	FeatureGCSFuseProfiles    *FeatureGCSFuseProfiles
+	GoMemLimitOptions         *GoMemLimitOptions
 }
 
 type GCSDriverConfig struct {

pkg/csi_driver/node.go

@@ -217,6 +217,42 @@ func (s *nodeServer) NodePublishVolume(ctx context.Context, req *csi.NodePublish
 		})
 	}
 
+	// Check if the user explicitly provided GOMEMLIMIT flags in their
+	// volume's mountOptions.
+	userProvidedAutoMemLimit := false
+	userProvidedRatio := false
+	for _, opt := range args.fuseMountOptions {
+		if opt == util.EnableAutoGoMemLimitConst || strings.HasPrefix(opt, util.EnableAutoGoMemLimitConst+"=") {
+			userProvidedAutoMemLimit = true
+		}
+		if strings.HasPrefix(opt, util.AutoGoMemLimitRatioConst+"=") {
+			userProvidedRatio = true
+		}
+	}
+
+	// Inject driver defaults if the sidecar supports the GOMEMLIMIT feature.
+	// We evaluate the ratio independently so users who manually opt-in via
+	// mountOptions without specifying a ratio still receive the driver's
+	// default.
+	if s.driver.isSidecarVersionSupportedForGivenFeature(gcsFuseSidecarImage, SidecarAutoGoMemLimitMinVersion) {
+		var extraOpts []string
+		enableAutoGoMemLimit := s.driver.config.FeatureOptions.GoMemLimitOptions != nil && s.driver.config.FeatureOptions.GoMemLimitOptions.EnableAutoGoMemLimit
+		if enableAutoGoMemLimit && !userProvidedAutoMemLimit {
+			extraOpts = append(extraOpts, util.EnableAutoGoMemLimitConst+"=true")
+		}
+		//  We check this to prevent injecting the ratio and adding unnecessary
+		// noise to the mount options when the feature is completely disabled.
+		sidecarFeatureWillBeEnabled := enableAutoGoMemLimit || userProvidedAutoMemLimit
+		if !userProvidedRatio && sidecarFeatureWillBeEnabled && s.driver.config.FeatureOptions.GoMemLimitOptions != nil {
+			autoGoMemLimitRatio := s.driver.config.FeatureOptions.GoMemLimitOptions.AutoGoMemLimitRatio
+			extraOpts = append(extraOpts, util.AutoGoMemLimitRatioConst+"="+strconv.FormatFloat(autoGoMemLimitRatio, 'f', -1, 64))
+		}
+
+		if len(extraOpts) > 0 {
+			args.fuseMountOptions = joinMountOptions(args.fuseMountOptions, extraOpts)
+		}
+	}
+
 	node, err := s.k8sClients.GetNode(s.driver.config.NodeID)
 	if err != nil {
 		return nil, status.Errorf(codes.NotFound, "failed to get node: %v", err)