STOR-2582: Rebase gcp-pd-csi-driver to v1.21.6 for OCP 4.21

.ci-operator.yaml

@@ -0,0 +1,4 @@
+build_root_image:
+  name: release
+  namespace: openshift
+  tag: rhel-9-release-golang-1.24-openshift-4.21

.snyk

@@ -0,0 +1,8 @@
+# References:
+# https://docs.snyk.io/scan-applications/snyk-code/using-snyk-code-from-the-cli/excluding-directories-and-files-from-the-snyk-code-cli-test
+# https://docs.snyk.io/snyk-cli/commands/ignore
+exclude:
+  global:
+    - vendor/**
+    - release-tools/**
+    - test/**

Dockerfile

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-FROM --platform=$BUILDPLATFORM golang:1.24.3 AS builder
+FROM --platform=$BUILDPLATFORM golang:1.24.6 AS builder
 
 ARG STAGINGVERSION
 ARG TARGETPLATFORM

Dockerfile.Windows

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 ARG BASE_IMAGE
-FROM --platform=$BUILDPLATFORM golang:1.24.3 AS builder
+FROM --platform=$BUILDPLATFORM golang:1.24.6 AS builder
 
 ARG TARGETPLATFORM
 ARG STAGINGVERSION

Dockerfile.debug

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-FROM golang:1.24.3 as builder
+FROM golang:1.24.6 as builder
 WORKDIR /go/src/sigs.k8s.io/gcp-compute-persistent-disk-csi-driver
 ADD . .
 

Dockerfile.openshift

@@ -0,0 +1,18 @@
+FROM registry.ci.openshift.org/ocp/builder:rhel-9-golang-1.24-openshift-4.21 AS builder
+WORKDIR /go/src/github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver
+COPY . .
+# Skip Windows driver build
+RUN make gce-pd-driver
+
+FROM registry.ci.openshift.org/ocp/4.21:base-rhel9
+
+# Get all deps
+# nvme-cli and /usr/bin/xxd are deps of google_nvme_id
+RUN yum update -y && \
+    yum install --setopt=tsflags=nodocs -y e2fsprogs xfsprogs util-linux nvme-cli /usr/lib/udev/scsi_id /usr/bin/xxd && \
+    yum clean all && rm -rf /var/cache/yum/* && \
+    mkdir -p /lib/udev_containerized && cp /usr/lib/udev/scsi_id /lib/udev_containerized/scsi_id # The driver assumes this path
+
+COPY --from=builder /go/src/github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/bin/gce-pd-csi-driver  /usr/bin/
+COPY --from=builder /go/src/github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/deploy/kubernetes/udev/google_nvme_id /lib/udev_containerized/google_nvme_id
+ENTRYPOINT ["/usr/bin/gce-pd-csi-driver"]

Makefile

@@ -38,7 +38,8 @@ endif
 all: gce-pd-driver gce-pd-driver-windows
 gce-pd-driver: require-GCE_PD_CSI_STAGING_VERSION
 	mkdir -p bin
-	CGO_ENABLED=0 go build -mod=vendor -gcflags=$(GCFLAGS) -ldflags "-extldflags=static -X main.version=$(STAGINGVERSION)" -o bin/${DRIVERBINARY} ./cmd/gce-pd-csi-driver/
+	# OpenShift carry: remove -extldflags=static to compile with FIPS OpenSSL
+	CGO_ENABLED=0 go build -mod=vendor -gcflags=$(GCFLAGS) -ldflags "-X main.version=$(STAGINGVERSION)" -o bin/${DRIVERBINARY} ./cmd/gce-pd-csi-driver/
 
 gce-pd-driver-windows: require-GCE_PD_CSI_STAGING_VERSION
 ifeq (${GOARCH}, amd64)

OWNERS

@@ -1,24 +1,3 @@
-reviewers:
-  - amacaskill
-  - leiyiz
-  - mattcary
-  - msau42
-  - saikat-royc
-  - sunnylovestiramisu
-  - pwschuurman
-  - tyuchn
-  - tonyzhc
-  - juliankatz
 approvers:
-  - amacaskill
-  - leiyiz
-  - mattcary
-  - msau42
-  - saikat-royc
-  - sunnylovestiramisu
-  - pwschuurman
-  - tyuchn
-emeritus_reviewers:
-  - davidz627
-  - jingxu97
-  - saad-ali
+- openshift-storage-maintainers
+component: "Storage / Kubernetes External Components"

README.md

@@ -12,7 +12,7 @@ lifecycle of Google Compute Engine Persistent Disks.
 ## Project Status
 
 Status: GA
-Latest stable image: `registry.k8s.io/cloud-provider-gcp/gcp-compute-persistent-disk-csi-driver:v1.17.2`
+Latest stable image: `registry.k8s.io/cloud-provider-gcp/gcp-compute-persistent-disk-csi-driver:v1.20.0`
 
 ### Test Status
 

cmd/gce-pd-csi-driver/main.go

@@ -34,6 +34,7 @@ import (
 	gce "sigs.k8s.io/gcp-compute-persistent-disk-csi-driver/pkg/gce-cloud-provider/compute"
 	metadataservice "sigs.k8s.io/gcp-compute-persistent-disk-csi-driver/pkg/gce-cloud-provider/metadata"
 	driver "sigs.k8s.io/gcp-compute-persistent-disk-csi-driver/pkg/gce-pd-csi-driver"
+	"sigs.k8s.io/gcp-compute-persistent-disk-csi-driver/pkg/linkcache"
 	"sigs.k8s.io/gcp-compute-persistent-disk-csi-driver/pkg/metrics"
 	mountmanager "sigs.k8s.io/gcp-compute-persistent-disk-csi-driver/pkg/mount-manager"
 )
@@ -57,10 +58,10 @@ var (
 	attachDiskBackoffJitter   = flag.Float64("attach-disk-backoff-jitter", 0.0, "Jitter for attachDisk backoff")
 	attachDiskBackoffSteps    = flag.Int("attach-disk-backoff-steps", 24, "Steps for attachDisk backoff")
 	attachDiskBackoffCap      = flag.Duration("attach-disk-backoff-cap", 0, "Cap for attachDisk backoff")
-	waitForOpBackoffDuration  = flag.Duration("wait-op-backoff-duration", 3*time.Second, "Duration for wait for operation backoff")
+	waitForOpBackoffDuration  = flag.Duration("wait-op-backoff-duration", 2*time.Minute, "Duration for wait for operation backoff")
 	waitForOpBackoffFactor    = flag.Float64("wait-op-backoff-factor", 0.0, "Factor for wait for operation backoff")
 	waitForOpBackoffJitter    = flag.Float64("wait-op-backoff-jitter", 0.0, "Jitter for wait for operation backoff")
-	waitForOpBackoffSteps     = flag.Int("wait-op-backoff-steps", 100, "Steps for wait for operation backoff")
+	waitForOpBackoffSteps     = flag.Int("wait-op-backoff-steps", 3, "Steps for wait for operation backoff")
 	waitForOpBackoffCap       = flag.Duration("wait-op-backoff-cap", 0, "Cap for wait for operation backoff")
 
 	enableDeviceInUseCheck = flag.Bool("enable-device-in-use-check-on-node-unstage", true, "If set to true, block NodeUnstageVolume requests until the specified device is not in use")
@@ -97,6 +98,10 @@ var (
 
 	diskTopology = flag.Bool("disk-topology", false, "If set to true, the driver will add a disk-type.gke.io/[disk-type] topology label when the StorageClass has the use-allowed-disk-topology parameter set to true. That topology label is included in the Topologies returned in CreateVolumeResponse. This flag is disabled by default.")
 
+	diskCacheSyncPeriod = flag.Duration("disk-cache-sync-period", 10*time.Minute, "Period for the disk cache to check the /dev/disk/by-id/ directory and evaluate the symlinks")
+
+	enableDiskSizeValidation = flag.Bool("enable-disk-size-validation", false, "If set to true, the driver will validate that the requested disk size is matches the physical disk size. This flag is disabled by default.")
+
 	version string
 )
 
@@ -248,7 +253,8 @@ func handle() {
 		maxBackoffDuration := time.Duration(*errorBackoffMaxDurationMs) * time.Millisecond
 		// TODO(2042): Move more of the constructor args into this struct
 		args := &driver.GCEControllerServerArgs{
-			EnableDiskTopology: *diskTopology,
+			EnableDiskTopology:       *diskTopology,
+			EnableDiskSizeValidation: *enableDiskSizeValidation,
 		}
 
 		controllerServer = driver.NewControllerServer(gceDriver, cloudProvider, initialBackoffDuration, maxBackoffDuration, fallbackRequisiteZones, *enableStoragePoolsFlag, *enableDataCacheFlag, multiZoneVolumeHandleConfig, listVolumesConfig, provisionableDisksConfig, *enableHdHAFlag, args)
@@ -270,11 +276,18 @@ func handle() {
 		if err != nil {
 			klog.Fatalf("Failed to set up metadata service: %v", err.Error())
 		}
-		isDataCacheEnabledNodePool, err := isDataCacheEnabledNodePool(ctx, *nodeName)
+		isDataCacheEnabledNodePool, err := driver.IsDataCacheEnabledNodePool(ctx, *nodeName, *enableDataCacheFlag)
 		if err != nil {
 			klog.Fatalf("Failed to get node info from API server: %v", err.Error())
 		}
 
+		deviceCache, err := linkcache.NewDeviceCacheForNode(ctx, *diskCacheSyncPeriod, *nodeName, driverName, deviceUtils)
+		if err != nil {
+			klog.Warningf("Failed to create device cache: %v", err.Error())
+		} else {
+			go deviceCache.Run(ctx)
+		}
+
 		// TODO(2042): Move more of the constructor args into this struct
 		nsArgs := &driver.NodeServerArgs{
 			EnableDeviceInUseCheck:   *enableDeviceInUseCheck,
@@ -283,6 +296,7 @@ func handle() {
 			DataCacheEnabledNodePool: isDataCacheEnabledNodePool,
 			SysfsPath:                "/sys",
 			MetricsManager:           metricsManager,
+			DeviceCache:              deviceCache,
 		}
 		nodeServer = driver.NewNodeServer(gceDriver, mounter, deviceUtils, meta, statter, nsArgs)
 
@@ -297,9 +311,10 @@ func handle() {
 				if err := setupDataCache(ctx, *nodeName, nodeServer.MetadataService.GetName()); err != nil {
 					klog.Errorf("Data Cache setup failed: %v", err)
 				}
-				go driver.StartWatcher(*nodeName)
+				go driver.StartWatcher(ctx, *nodeName)
 			}
 		}
+
 	}
 
 	err = gceDriver.SetupGCEDriver(driverName, version, extraVolumeLabels, extraTags, identityServer, controllerServer, nodeServer)
@@ -381,17 +396,6 @@ func urlFlag(target **url.URL, name string, usage string) {
 	})
 }
 
-func isDataCacheEnabledNodePool(ctx context.Context, nodeName string) (bool, error) {
-	if !*enableDataCacheFlag {
-		return false, nil
-	}
-	if len(nodeName) > 0 && nodeName != common.TestNode { // disregard logic below when E2E testing.
-		dataCacheLSSDCount, err := driver.GetDataCacheCountFromNodeLabel(ctx, nodeName)
-		return dataCacheLSSDCount != 0, err
-	}
-	return true, nil
-}
-
 func fetchLssdsForRaiding(lssdCount int) ([]string, error) {
 	allLssds, err := driver.FetchAllLssds()
 	if err != nil {

deploy/kubernetes/base/controller/controller.yaml

@@ -145,6 +145,7 @@ spec:
             - "--supports-dynamic-iops-provisioning=hyperdisk-balanced,hyperdisk-extreme"
             - "--supports-dynamic-throughput-provisioning=hyperdisk-balanced,hyperdisk-throughput,hyperdisk-ml"
             - --enable-data-cache
+            - --run-node-service=false
             - --enable-multitenancy
           command:
             - /gce-pd-csi-driver

deploy/kubernetes/images/prow-stable-sidecar-rc-master/image.yaml

@@ -48,6 +48,6 @@ metadata:
 imageTag:
   name: registry.k8s.io/cloud-provider-gcp/gcp-compute-persistent-disk-csi-driver
   newName: gcr.io/k8s-staging-cloud-provider-gcp/gcp-compute-persistent-disk-csi-driver
-  newTag: "v1.17.12"
+  newTag: "v1.20.0"
 ---
 

deploy/kubernetes/images/stable-master/image.yaml

@@ -44,9 +44,8 @@ metadata:
   name: imagetag-gcepd-driver
 imageTag:
   name: gke.gcr.io/gcp-compute-persistent-disk-csi-driver
-  # Don't change stable image without changing pdImagePlaceholder in
-  # test/k8s-integration/main.go
-  newName: us-central1-docker.pkg.dev/enginakdemir-gke-dev/csi-dev/gcp-compute-persistent-disk-csi-driver
-  newTag: "latest"
+  # pdImagePlaceholder in test/k8s-integration/main.go is updated automatically with the newTag
+  newName: registry.k8s.io/cloud-provider-gcp/gcp-compute-persistent-disk-csi-driver
+  newTag: "v1.17.4"
 ---
 

go.mod

@@ -25,23 +25,23 @@ require (
 	go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc v1.35.0
 	go.opentelemetry.io/otel/sdk v1.35.0
 	golang.org/x/oauth2 v0.30.0
-	golang.org/x/sys v0.33.0
-	golang.org/x/time v0.11.0
+	golang.org/x/sys v0.35.0
+	golang.org/x/time v0.12.0
 	google.golang.org/api v0.187.0
 	google.golang.org/genproto v0.0.0-20240624140628-dc46fd24d27d
 	google.golang.org/grpc v1.71.0
 	google.golang.org/protobuf v1.36.6
 	gopkg.in/gcfg.v1 v1.2.3
-	k8s.io/api v0.33.1
-	k8s.io/apimachinery v0.33.1
+	k8s.io/api v0.33.3
+	k8s.io/apimachinery v0.33.3
 	k8s.io/client-go v11.0.1-0.20190805182717-6502b5e7b1b5+incompatible
-	k8s.io/cloud-provider v0.33.1
-	k8s.io/component-base v0.33.1
+	k8s.io/cloud-provider v0.33.3
+	k8s.io/component-base v0.33.3
 	k8s.io/klog/v2 v2.130.1
-	k8s.io/mount-utils v0.33.1
+	k8s.io/mount-utils v0.33.3
 	k8s.io/utils v0.0.0-20241104100929-3ea5e8cea738
 	sigs.k8s.io/boskos v0.0.0-20220711194915-6cb8a6fb2dd1
-	sigs.k8s.io/yaml v1.4.0
+	sigs.k8s.io/yaml v1.6.0
 )
 
 require (
@@ -97,13 +97,14 @@ require (
 	go.opentelemetry.io/otel/metric v1.35.0 // indirect
 	go.opentelemetry.io/otel/trace v1.35.0 // indirect
 	go.opentelemetry.io/proto/otlp v1.5.0 // indirect
+	go.yaml.in/yaml/v2 v2.4.2 // indirect
 	go4.org v0.0.0-20201209231011-d4a079459e60 // indirect
-	golang.org/x/crypto v0.38.0 // indirect
-	golang.org/x/net v0.40.0 // indirect
-	golang.org/x/sync v0.14.0 // indirect
-	golang.org/x/term v0.32.0 // indirect
-	golang.org/x/text v0.25.0 // indirect
-	golang.org/x/tools v0.33.0 // indirect
+	golang.org/x/crypto v0.41.0 // indirect
+	golang.org/x/net v0.42.0 // indirect
+	golang.org/x/sync v0.16.0 // indirect
+	golang.org/x/term v0.34.0 // indirect
+	golang.org/x/text v0.28.0 // indirect
+	golang.org/x/tools v0.35.0 // indirect
 	google.golang.org/genproto/googleapis/api v0.0.0-20250218202821-56aae31c358a // indirect
 	google.golang.org/genproto/googleapis/rpc v0.0.0-20250218202821-56aae31c358a // indirect
 	gopkg.in/evanphx/json-patch.v4 v4.12.0 // indirect