fix: gcs tier going offline due to customer HTTPclient (#19973)

specifying customer HTTP client makes the gcs SDK
ignore the passed credentials, instead let the GCS
SDK manage the transport.

this PR fixes #19922 a regression from #19565

.github/workflows/helm-lint.yml

@@ -22,7 +22,7 @@ jobs:
         uses: actions/checkout@v4
 
       - name: Install Helm
-        uses: azure/setup-helm@v3
+        uses: azure/setup-helm@v4
 
       - name: Run helm lint
         run: |

.github/workflows/mint.yml

@@ -55,9 +55,10 @@ jobs:
         run: |
           ${GITHUB_WORKSPACE}/.github/workflows/run-mint.sh "erasure" "minio" "minio123" "${{ steps.vars.outputs.sha_short }}"
 
-      - name: resiliency
+      # FIXME: renable this back when we have a valid way to add deadlines for PUT()s (internode CreateFile)
-        run: |
+      # - name: resiliency
-          ${GITHUB_WORKSPACE}/.github/workflows/run-mint.sh "resiliency" "minio" "minio123" "${{ steps.vars.outputs.sha_short }}"
+      #   run: |
+      #     ${GITHUB_WORKSPACE}/.github/workflows/run-mint.sh "resiliency" "minio" "minio123" "${{ steps.vars.outputs.sha_short }}"
 
       - name: The job must cleanup
         if: ${{ always() }}

.github/workflows/vulncheck.yml

@@ -21,7 +21,7 @@ jobs:
       - name: Set up Go
         uses: actions/setup-go@v5
         with:
-          go-version: 1.22.3
+          go-version: 1.22.4
           check-latest: true
       - name: Get official govulncheck
         run: go install golang.org/x/vuln/cmd/govulncheck@latest

Makefile

@@ -47,7 +47,7 @@ lint-fix: getdeps ## runs golangci-lint suite of linters with automatic fixes
 check: test
 test: verifiers build ## builds minio, runs linters, tests
 	@echo "Running unit tests"
-	@MINIO_API_REQUESTS_MAX=10000 CGO_ENABLED=0 go test -v -tags kqueue ./...
+	@MINIO_API_REQUESTS_MAX=10000 CGO_ENABLED=0 go test -v -tags kqueue,dev ./...
 
 test-root-disable: install-race
 	@echo "Running minio root lockdown tests"
@@ -84,13 +84,13 @@ test-race: verifiers build ## builds minio, runs linters, tests (race)
 	@echo "Running unit tests under -race"
 	@(env bash $(PWD)/buildscripts/race.sh)
 
-test-iam: build ## verify IAM (external IDP, etcd backends)
+test-iam: install-race ## verify IAM (external IDP, etcd backends)
 	@echo "Running tests for IAM (external IDP, etcd backends)"
-	@MINIO_API_REQUESTS_MAX=10000 CGO_ENABLED=0 go test -tags kqueue -v -run TestIAM* ./cmd
+	@MINIO_API_REQUESTS_MAX=10000 CGO_ENABLED=0 go test -tags kqueue,dev -v -run TestIAM* ./cmd
 	@echo "Running tests for IAM (external IDP, etcd backends) with -race"
-	@MINIO_API_REQUESTS_MAX=10000 GORACE=history_size=7 CGO_ENABLED=1 go test -race -tags kqueue -v -run TestIAM* ./cmd
+	@MINIO_API_REQUESTS_MAX=10000 GORACE=history_size=7 CGO_ENABLED=1 go test -race -tags kqueue,dev -v -run TestIAM* ./cmd
 
-test-iam-ldap-upgrade-import: build ## verify IAM (external LDAP IDP)
+test-iam-ldap-upgrade-import: install-race ## verify IAM (external LDAP IDP)
 	@echo "Running upgrade tests for IAM (LDAP backend)"
 	@env bash $(PWD)/buildscripts/minio-iam-ldap-upgrade-import-test.sh
 
@@ -106,7 +106,10 @@ test-replication-3site:
 test-delete-replication:
 	@(env bash $(PWD)/docs/bucket/replication/delete-replication.sh)
 
-test-replication: install-race test-replication-2site test-replication-3site test-delete-replication test-sio-error ## verify multi site replication
+test-delete-marker-proxying:
+	@(env bash $(PWD)/docs/bucket/replication/test_del_marker_proxying.sh)
+
+test-replication: install-race test-replication-2site test-replication-3site test-delete-replication test-sio-error test-delete-marker-proxying ## verify multi site replication
 	@echo "Running tests for replicating three sites"
 
 test-site-replication-ldap: install-race ## verify automatic site replication
@@ -162,9 +165,9 @@ hotfix-vars:
 	$(eval VERSION := $(shell git describe --tags --abbrev=0).hotfix.$(shell git rev-parse --short HEAD))
 
 hotfix: hotfix-vars clean install ## builds minio binary with hotfix tags
-	@wget -q -c https://github.com/minio/pkger/releases/download/v2.2.9/pkger_2.2.9_linux_amd64.deb
+	@wget -q -c https://github.com/minio/pkger/releases/download/v2.3.1/pkger_2.3.1_linux_amd64.deb
 	@wget -q -c https://raw.githubusercontent.com/minio/minio-service/v1.0.1/linux-systemd/distributed/minio.service
-	@sudo apt install ./pkger_2.2.9_linux_amd64.deb --yes
+	@sudo apt install ./pkger_2.3.1_linux_amd64.deb --yes
 	@mkdir -p minio-release/$(GOOS)-$(GOARCH)/archive
 	@cp -af ./minio minio-release/$(GOOS)-$(GOARCH)/minio
 	@cp -af ./minio minio-release/$(GOOS)-$(GOARCH)/minio.$(VERSION)
@@ -193,7 +196,7 @@ docker: build ## builds minio docker container
 
 install-race: checks build-debugging ## builds minio to $(PWD)
 	@echo "Building minio binary with -race to './minio'"
-	@GORACE=history_size=7 CGO_ENABLED=1 go build -tags kqueue -race -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@GORACE=history_size=7 CGO_ENABLED=1 go build -tags kqueue,dev -race -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
 	@echo "Installing minio binary with -race to '$(GOPATH)/bin/minio'"
 	@mkdir -p $(GOPATH)/bin && cp -af $(PWD)/minio $(GOPATH)/bin/minio
 

buildscripts/minio-upgrade.sh

@@ -4,10 +4,22 @@ trap 'cleanup $LINENO' ERR
 
 # shellcheck disable=SC2120
 cleanup() {
-	MINIO_VERSION=dev docker-compose \
+	MINIO_VERSION=dev /tmp/gopath/bin/docker-compose \
 		-f "buildscripts/upgrade-tests/compose.yml" \
-		rm -s -f
+		down || true
+
+	MINIO_VERSION=dev /tmp/gopath/bin/docker-compose \
+		-f "buildscripts/upgrade-tests/compose.yml" \
+		rm || true
+
+	for volume in $(docker volume ls -q | grep upgrade); do
+		docker volume rm ${volume} || true
+	done
+
 	docker volume prune -f
+	docker system prune -f || true
+	docker volume prune -f || true
+	docker volume rm $(docker volume ls -q -f dangling=true) || true
 }
 
 verify_checksum_after_heal() {
@@ -60,6 +72,8 @@ __init__() {
 	go install github.com/docker/compose/v2/cmd@latest
 	mv -v /tmp/gopath/bin/cmd /tmp/gopath/bin/docker-compose
 
+	cleanup
+
 	TAG=minio/minio:dev make docker
 
 	MINIO_VERSION=RELEASE.2019-12-19T22-52-26Z docker-compose \
@@ -77,11 +91,11 @@ __init__() {
 
 	curl -s http://127.0.0.1:9000/minio-test/to-read/hosts | sha256sum
 
-	MINIO_VERSION=dev docker-compose -f "buildscripts/upgrade-tests/compose.yml" stop
+	MINIO_VERSION=dev /tmp/gopath/bin/docker-compose -f "buildscripts/upgrade-tests/compose.yml" stop
 }
 
 main() {
-	MINIO_VERSION=dev docker-compose -f "buildscripts/upgrade-tests/compose.yml" up -d --build
+	MINIO_VERSION=dev /tmp/gopath/bin/docker-compose -f "buildscripts/upgrade-tests/compose.yml" up -d --build
 
 	add_alias
 

buildscripts/upgrade-tests/compose.yml

@@ -1,5 +1,3 @@
-version: '3.7'
-
 # Settings and configurations that are common for all containers
 x-minio-common: &minio-common
   image: minio/minio:${MINIO_VERSION}

buildscripts/verify-healing-empty-erasure-set.sh

@@ -38,40 +38,47 @@ function start_minio_3_node() {
 	disown $pid3
 
 	export MC_HOST_myminio="http://minio:minio123@127.0.0.1:$((start_port + 1))"
-	/tmp/mc ready myminio
+
+	timeout 15m /tmp/mc ready myminio || fail
+
+	# Wait for all drives to be online and formatted
+	while [ $(/tmp/mc admin info --json myminio | jq '.info.servers[].drives[].state | select(. != "ok")' | wc -l) -gt 0 ]; do sleep 1; done
+	# Wait for all drives to be healed
+	while [ $(/tmp/mc admin info --json myminio | jq '.info.servers[].drives[].healing | select(. != null) | select(. == true)' | wc -l) -gt 0 ]; do sleep 1; done
+
+	# Wait for Status: in MinIO output
+	while true; do
+		rv=$(check_online)
+		if [ "$rv" != "1" ]; then
+			# success
+			break
+		fi
+
+		# Check if we should retry
+		retry=$((retry + 1))
+		if [ $retry -le 20 ]; then
+			sleep 5
+			continue
+		fi
+
+		# Failure
+		fail
+	done
 
 	if ! ps -p $pid1 1>&2 >/dev/null; then
-		echo "server1 log:"
+		echo "minio-server-1 is not running." && fail
-		cat "${WORK_DIR}/dist-minio-server1.log"
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	if ! ps -p $pid2 1>&2 >/dev/null; then
-		echo "server2 log:"
+		echo "minio-server-2 is not running." && fail
-		cat "${WORK_DIR}/dist-minio-server2.log"
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	if ! ps -p $pid3 1>&2 >/dev/null; then
-		echo "server3 log:"
+		echo "minio-server-3 is not running." && fail
-		cat "${WORK_DIR}/dist-minio-server3.log"
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	if ! pkill minio; then
-		for i in $(seq 1 3); do
+		fail
-			echo "server$i log:"
-			cat "${WORK_DIR}/dist-minio-server$i.log"
-		done
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	sleep 1
@@ -83,6 +90,16 @@ function start_minio_3_node() {
 	fi
 }
 
+function fail() {
+	for i in $(seq 1 3); do
+		echo "server$i log:"
+		cat "${WORK_DIR}/dist-minio-server$i.log"
+	done
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+}
+
 function check_online() {
 	if ! grep -q 'Status:' ${WORK_DIR}/dist-minio-*.log; then
 		echo "1"
@@ -90,7 +107,7 @@ function check_online() {
 }
 
 function purge() {
-	rm -rf "$1"
+	echo rm -rf "$1"
 }
 
 function __init__() {
@@ -117,18 +134,6 @@ function perform_test() {
 
 	set -x
 	start_minio_3_node $2
-
-	rv=$(check_online)
-	if [ "$rv" == "1" ]; then
-		for i in $(seq 1 3); do
-			echo "server$i log:"
-			cat "${WORK_DIR}/dist-minio-server$i.log"
-		done
-		pkill -9 minio
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
-	fi
 }
 
 function main() {

buildscripts/verify-healing.sh

@@ -47,43 +47,25 @@ function start_minio_3_node() {
 	disown $pid3
 
 	export MC_HOST_myminio="http://minio:minio123@127.0.0.1:$((start_port + 1))"
-	/tmp/mc ready myminio
+	timeout 15m /tmp/mc ready myminio || fail
 
 	[ ${first_time} -eq 0 ] && upload_objects
 	[ ${first_time} -ne 0 ] && sleep 120
 
 	if ! ps -p $pid1 1>&2 >/dev/null; then
-		echo "server1 log:"
+		echo "minio server 1 is not running" && fail
-		cat "${WORK_DIR}/dist-minio-server1.log"
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	if ! ps -p $pid2 1>&2 >/dev/null; then
-		echo "server2 log:"
+		echo "minio server 2 is not running" && fail
-		cat "${WORK_DIR}/dist-minio-server2.log"
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	if ! ps -p $pid3 1>&2 >/dev/null; then
-		echo "server3 log:"
+		echo "minio server 3 is not running" && fail
-		cat "${WORK_DIR}/dist-minio-server3.log"
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	if ! pkill minio; then
-		for i in $(seq 1 3); do
+		fail
-			echo "server$i log:"
-			cat "${WORK_DIR}/dist-minio-server$i.log"
-		done
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 
 	sleep 1
@@ -118,6 +100,17 @@ function purge() {
 	rm -rf "$1"
 }
 
+function fail() {
+	for i in $(seq 1 3); do
+		echo "server$i log:"
+		cat "${WORK_DIR}/dist-minio-server$i.log"
+	done
+	pkill -9 minio
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+}
+
 function __init__() {
 	echo "Initializing environment"
 	mkdir -p "$WORK_DIR"
@@ -155,14 +148,7 @@ function perform_test() {
 	check_heal ${1}
 	rv=$?
 	if [ "$rv" == "1" ]; then
-		for i in $(seq 1 3); do
+		fail
-			echo "server$i log:"
-			cat "${WORK_DIR}/dist-minio-server$i.log"
-		done
-		pkill -9 minio
-		echo "FAILED"
-		purge "$WORK_DIR"
-		exit 1
 	fi
 }
 

cmd/admin-handlers-idp-ldap.go

@@ -20,6 +20,7 @@ package cmd
 import (
 	"encoding/json"
 	"errors"
+	"fmt"
 	"io"
 	"net/http"
 	"strings"
@@ -284,6 +285,18 @@ func (a adminAPIHandlers) AddServiceAccountLDAP(w http.ResponseWriter, r *http.R
 		opts.claims[ldapUser] = targetUser // DN
 		opts.claims[ldapActualUser] = lookupResult.ActualDN
 
+		// Check if this user or their groups have a policy applied.
+		ldapPolicies, err := globalIAMSys.PolicyDBGet(targetUser, targetGroups...)
+		if err != nil {
+			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+			return
+		}
+		if len(ldapPolicies) == 0 {
+			err = fmt.Errorf("No policy set for user `%s` or any of their groups: `%s`", opts.claims[ldapActualUser], strings.Join(targetGroups, "`,`"))
+			writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminNoSuchUser, err), r.URL)
+			return
+		}
+
 		// Add LDAP attributes that were looked up into the claims.
 		for attribKey, attribValue := range lookupResult.Attributes {
 			opts.claims[ldapAttribPrefix+attribKey] = attribValue

cmd/admin-handlers-users.go

@@ -29,6 +29,7 @@ import (
 	"sort"
 	"strconv"
 	"time"
+	"unicode/utf8"
 
 	"github.com/klauspost/compress/zip"
 	"github.com/minio/madmin-go/v3"
@@ -474,6 +475,11 @@ func (a adminAPIHandlers) AddUser(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	if !utf8.ValidString(accessKey) {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAddUserValidUTF), r.URL)
+		return
+	}
+
 	checkDenyOnly := false
 	if accessKey == cred.AccessKey {
 		// Check that there is no explicit deny - otherwise it's allowed
@@ -824,7 +830,11 @@ func (a adminAPIHandlers) UpdateServiceAccount(w http.ResponseWriter, r *http.Re
 	}
 
 	condValues := getConditionValues(r, "", cred)
-	addExpirationToCondValues(updateReq.NewExpiration, condValues)
+	err = addExpirationToCondValues(updateReq.NewExpiration, condValues)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
 
 	// Permission checks:
 	//
@@ -2453,11 +2463,16 @@ func (a adminAPIHandlers) ImportIAM(w http.ResponseWriter, r *http.Request) {
 	}
 }
 
-func addExpirationToCondValues(exp *time.Time, condValues map[string][]string) {
+func addExpirationToCondValues(exp *time.Time, condValues map[string][]string) error {
-	if exp == nil {
+	if exp == nil || exp.IsZero() || exp.Equal(timeSentinel) {
-		return
+		return nil
 	}
-	condValues["DurationSeconds"] = []string{strconv.FormatInt(int64(exp.Sub(time.Now()).Seconds()), 10)}
+	dur := exp.Sub(time.Now())
+	if dur <= 0 {
+		return errors.New("unsupported expiration time")
+	}
+	condValues["DurationSeconds"] = []string{strconv.FormatInt(int64(dur.Seconds()), 10)}
+	return nil
 }
 
 func commonAddServiceAccount(r *http.Request) (context.Context, auth.Credentials, newServiceAccountOpts, madmin.AddServiceAccountReq, string, APIError) {
@@ -2522,7 +2537,10 @@ func commonAddServiceAccount(r *http.Request) (context.Context, auth.Credentials
 	}
 
 	condValues := getConditionValues(r, "", cred)
-	addExpirationToCondValues(createReq.Expiration, condValues)
+	err = addExpirationToCondValues(createReq.Expiration, condValues)
+	if err != nil {
+		return ctx, auth.Credentials{}, newServiceAccountOpts{}, madmin.AddServiceAccountReq{}, "", toAdminAPIErr(ctx, err)
+	}
 
 	// Check if action is allowed if creating access key for another user
 	// Check if action is explicitly denied if for self

cmd/admin-handlers.go

@@ -237,6 +237,7 @@ func (a adminAPIHandlers) ServerUpdateV2Handler(w http.ResponseWriter, r *http.R
 
 	if globalIsDistErasure {
 		// Notify all other MinIO peers signal service.
+		startTime := time.Now().Add(restartUpdateDelay)
 		ng := WithNPeers(len(globalNotificationSys.peerClients))
 		for idx, client := range globalNotificationSys.peerClients {
 			_, ok := failedClients[idx]
@@ -247,7 +248,7 @@ func (a adminAPIHandlers) ServerUpdateV2Handler(w http.ResponseWriter, r *http.R
 			ng.Go(ctx, func() error {
 				prs, ok := peerResults[client.String()]
 				if ok && prs.CurrentVersion != prs.UpdatedVersion && prs.UpdatedVersion != "" {
-					return client.SignalService(serviceRestart, "", dryRun)
+					return client.SignalService(serviceRestart, "", dryRun, &startTime)
 				}
 				return nil
 			}, idx, *client.host)
@@ -542,9 +543,12 @@ func (a adminAPIHandlers) ServiceV2Handler(w http.ResponseWriter, r *http.Reques
 	}
 
 	var objectAPI ObjectLayer
+	var execAt *time.Time
 	switch serviceSig {
 	case serviceRestart:
 		objectAPI, _ = validateAdminReq(ctx, w, r, policy.ServiceRestartAdminAction)
+		t := time.Now().Add(restartUpdateDelay)
+		execAt = &t
 	case serviceStop:
 		objectAPI, _ = validateAdminReq(ctx, w, r, policy.ServiceStopAdminAction)
 	case serviceFreeze, serviceUnFreeze:
@@ -571,7 +575,7 @@ func (a adminAPIHandlers) ServiceV2Handler(w http.ResponseWriter, r *http.Reques
 	}
 
 	if globalIsDistErasure {
-		for _, nerr := range globalNotificationSys.SignalServiceV2(serviceSig, dryRun) {
+		for _, nerr := range globalNotificationSys.SignalServiceV2(serviceSig, dryRun, execAt) {
 			if nerr.Err != nil && process {
 				waitingDrives := map[string]madmin.DiskMetrics{}
 				jerr := json.Unmarshal([]byte(nerr.Err.Error()), &waitingDrives)
@@ -2417,7 +2421,7 @@ func getServerInfo(ctx context.Context, pools, metrics bool, r *http.Request) ma
 		Mode:          string(mode),
 		Domain:        domain,
 		Region:        globalSite.Region(),
-		SQSARN:        globalEventNotifier.GetARNList(false),
+		SQSARN:        globalEventNotifier.GetARNList(),
 		DeploymentID:  globalDeploymentID(),
 		Buckets:       buckets,
 		Objects:       objects,

cmd/api-errors.go

@@ -287,6 +287,7 @@ const (
 	ErrAdminNoSuchGroup
 	ErrAdminGroupNotEmpty
 	ErrAdminGroupDisabled
+	ErrAdminInvalidGroupName
 	ErrAdminNoSuchJob
 	ErrAdminNoSuchPolicy
 	ErrAdminPolicyChangeAlreadyApplied
@@ -425,6 +426,7 @@ const (
 	ErrAdminProfilerNotEnabled
 	ErrInvalidDecompressedSize
 	ErrAddUserInvalidArgument
+	ErrAddUserValidUTF
 	ErrAdminResourceInvalidArgument
 	ErrAdminAccountNotEligible
 	ErrAccountNotEligible
@@ -443,6 +445,8 @@ const (
 	ErrAdminNoAccessKey
 	ErrAdminNoSecretKey
 
+	ErrIAMNotInitialized
+
 	apiErrCodeEnd // This is used only for the testing code
 )
 
@@ -965,7 +969,7 @@ var errorCodes = errorCodeMap{
 	ErrReplicationRemoteConnectionError: {
 		Code:           "XMinioAdminReplicationRemoteConnectionError",
 		Description:    "Remote service connection error",
-		HTTPStatusCode: http.StatusNotFound,
+		HTTPStatusCode: http.StatusServiceUnavailable,
 	},
 	ErrReplicationBandwidthLimitError: {
 		Code:           "XMinioAdminReplicationBandwidthLimitError",
@@ -1303,6 +1307,11 @@ var errorCodes = errorCodeMap{
 		Description:    "Server not initialized yet, please try again.",
 		HTTPStatusCode: http.StatusServiceUnavailable,
 	},
+	ErrIAMNotInitialized: {
+		Code:           "XMinioIAMNotInitialized",
+		Description:    "IAM sub-system not initialized yet, please try again.",
+		HTTPStatusCode: http.StatusServiceUnavailable,
+	},
 	ErrBucketMetadataNotInitialized: {
 		Code:           "XMinioBucketMetadataNotInitialized",
 		Description:    "Bucket metadata not initialized yet, please try again.",
@@ -2101,6 +2110,16 @@ var errorCodes = errorCodeMap{
 		Description:    "Expected LDAP short username but was given full DN.",
 		HTTPStatusCode: http.StatusBadRequest,
 	},
+	ErrAdminInvalidGroupName: {
+		Code:           "XMinioInvalidGroupName",
+		Description:    "The group name is invalid.",
+		HTTPStatusCode: http.StatusBadRequest,
+	},
+	ErrAddUserValidUTF: {
+		Code:           "XMinioInvalidUTF",
+		Description:    "Invalid UTF-8 character detected.",
+		HTTPStatusCode: http.StatusBadRequest,
+	},
 }
 
 // toAPIErrorCode - Converts embedded errors. Convenience
@@ -2140,6 +2159,8 @@ func toAPIErrorCode(ctx context.Context, err error) (apiErr APIErrorCode) {
 		apiErr = ErrAdminNoSuchGroup
 	case errGroupNotEmpty:
 		apiErr = ErrAdminGroupNotEmpty
+	case errGroupNameContainsReservedChars:
+		apiErr = ErrAdminInvalidGroupName
 	case errNoSuchJob:
 		apiErr = ErrAdminNoSuchJob
 	case errNoPolicyToAttachOrDetach:
@@ -2154,6 +2175,8 @@ func toAPIErrorCode(ctx context.Context, err error) (apiErr APIErrorCode) {
 		apiErr = ErrEntityTooSmall
 	case errAuthentication:
 		apiErr = ErrAccessDenied
+	case auth.ErrContainsReservedChars:
+		apiErr = ErrAdminInvalidAccessKey
 	case auth.ErrInvalidAccessKeyLength:
 		apiErr = ErrAdminInvalidAccessKey
 	case auth.ErrInvalidSecretKeyLength:

cmd/api-headers.go

@@ -136,7 +136,7 @@ func setObjectHeaders(ctx context.Context, w http.ResponseWriter, objInfo Object
 	// Set tag count if object has tags
 	if len(objInfo.UserTags) > 0 {
 		tags, _ := tags.ParseObjectTags(objInfo.UserTags)
-		if tags.Count() > 0 {
+		if tags != nil && tags.Count() > 0 {
 			w.Header()[xhttp.AmzTagCount] = []string{strconv.Itoa(tags.Count())}
 			if opts.Tagging {
 				// This is MinIO only extension to return back tags along with the count.

cmd/api-response.go

@@ -789,8 +789,8 @@ func generateInitiateMultipartUploadResponse(bucket, key, uploadID string) Initi
 }
 
 // generates CompleteMultipartUploadResponse for given bucket, key, location and ETag.
-func generateCompleteMultpartUploadResponse(bucket, key, location string, oi ObjectInfo) CompleteMultipartUploadResponse {
+func generateCompleteMultipartUploadResponse(bucket, key, location string, oi ObjectInfo, h http.Header) CompleteMultipartUploadResponse {
-	cs := oi.decryptChecksums(0)
+	cs := oi.decryptChecksums(0, h)
 	c := CompleteMultipartUploadResponse{
 		Location: location,
 		Bucket:   bucket,

cmd/api-router.go

@@ -436,7 +436,7 @@ func registerAPIRouter(router *mux.Router) {
 			Queries("notification", "")
 		// ListenNotification
 		router.Methods(http.MethodGet).
-			HandlerFunc(s3APIMiddleware(api.ListenNotificationHandler, noThrottleS3HFlag)).
+			HandlerFunc(s3APIMiddleware(api.ListenNotificationHandler, noThrottleS3HFlag, traceHdrsS3HFlag)).
 			Queries("events", "{events:.*}")
 		// ResetBucketReplicationStatus - MinIO extension API
 		router.Methods(http.MethodGet).
@@ -615,7 +615,7 @@ func registerAPIRouter(router *mux.Router) {
 
 	// ListenNotification
 	apiRouter.Methods(http.MethodGet).Path(SlashSeparator).
-		HandlerFunc(s3APIMiddleware(api.ListenNotificationHandler, noThrottleS3HFlag)).
+		HandlerFunc(s3APIMiddleware(api.ListenNotificationHandler, noThrottleS3HFlag, traceHdrsS3HFlag)).
 		Queries("events", "{events:.*}")
 
 	// ListBuckets

cmd/background-newdisks-heal-ops.go

@@ -141,14 +141,14 @@ func initHealingTracker(disk StorageAPI, healID string) *healingTracker {
 	return h
 }
 
-func (h healingTracker) getLastUpdate() time.Time {
+func (h *healingTracker) getLastUpdate() time.Time {
 	h.mu.RLock()
 	defer h.mu.RUnlock()
 
 	return h.LastUpdate
 }
 
-func (h healingTracker) getBucket() string {
+func (h *healingTracker) getBucket() string {
 	h.mu.RLock()
 	defer h.mu.RUnlock()
 
@@ -162,7 +162,7 @@ func (h *healingTracker) setBucket(bucket string) {
 	h.Bucket = bucket
 }
 
-func (h healingTracker) getObject() string {
+func (h *healingTracker) getObject() string {
 	h.mu.RLock()
 	defer h.mu.RUnlock()
 
@@ -453,10 +453,6 @@ func healFreshDisk(ctx context.Context, z *erasureServerPools, endpoint Endpoint
 
 	healingLogEvent(ctx, "Healing of drive '%s' is finished (healed: %d, skipped: %d, failed: %d).", disk, tracker.ItemsHealed, tracker.ItemsSkipped, tracker.ItemsFailed)
 
-	if len(tracker.QueuedBuckets) > 0 {
-		return fmt.Errorf("not all buckets were healed: %v", tracker.QueuedBuckets)
-	}
-
 	if serverDebugLog {
 		tracker.printTo(os.Stdout)
 		fmt.Printf("\n")

cmd/batch-expire.go

@@ -432,7 +432,7 @@ func batchObjsForDelete(ctx context.Context, r *BatchJobExpire, ri *batchJobInfo
 					})
 					if err != nil {
 						stopFn(exp, err)
-						batchLogIf(ctx, fmt.Errorf("Failed to expire %s/%s versionID=%s due to %v (attempts=%d)", toExpire[i].Bucket, toExpire[i].Name, toExpire[i].VersionID, err, attempts))
+						batchLogIf(ctx, fmt.Errorf("Failed to expire %s/%s due to %v (attempts=%d)", exp.Bucket, exp.Name, err, attempts))
 					} else {
 						stopFn(exp, err)
 						success = true
@@ -464,13 +464,13 @@ func batchObjsForDelete(ctx context.Context, r *BatchJobExpire, ri *batchJobInfo
 				copy(toDelCopy, toDel)
 				var failed int
 				errs := r.Expire(ctx, api, vc, toDel)
-				// reslice toDel in preparation for next retry
+				// reslice toDel in preparation for next retry attempt
-				// attempt
 				toDel = toDel[:0]
 				for i, err := range errs {
 					if err != nil {
 						stopFn(toDelCopy[i], err)
-						batchLogIf(ctx, fmt.Errorf("Failed to expire %s/%s versionID=%s due to %v (attempts=%d)", ri.Bucket, toDelCopy[i].ObjectName, toDelCopy[i].VersionID, err, attempts))
+						batchLogIf(ctx, fmt.Errorf("Failed to expire %s/%s versionID=%s due to %v (attempts=%d)", ri.Bucket, toDelCopy[i].ObjectName, toDelCopy[i].VersionID,
+							err, attempts))
 						failed++
 						if oi, ok := oiCache.Get(toDelCopy[i]); ok {
 							ri.trackCurrentBucketObject(r.Bucket, *oi, false, attempts)

cmd/batch-handlers.go

@@ -1476,7 +1476,7 @@ func (j *BatchJobRequest) load(ctx context.Context, api ObjectLayer, name string
 
 func batchReplicationOpts(ctx context.Context, sc string, objInfo ObjectInfo) (putOpts miniogo.PutObjectOptions, err error) {
 	// TODO: support custom storage class for remote replication
-	putOpts, err = putReplicationOpts(ctx, "", objInfo)
+	putOpts, err = putReplicationOpts(ctx, "", objInfo, 0)
 	if err != nil {
 		return putOpts, err
 	}
@@ -1633,7 +1633,7 @@ func (a adminAPIHandlers) StartBatchJob(w http.ResponseWriter, r *http.Request)
 		return
 	}
 
-	job.ID = fmt.Sprintf("%s:%d", shortuuid.New(), GetProxyEndpointLocalIndex(globalProxyEndpoints))
+	job.ID = fmt.Sprintf("%s%s%d", shortuuid.New(), getKeySeparator(), GetProxyEndpointLocalIndex(globalProxyEndpoints))
 	job.User = user
 	job.Started = time.Now()
 

cmd/bootstrap-peer-server.go

@@ -19,9 +19,13 @@ package cmd
 
 import (
 	"context"
+	"crypto/md5"
+	"encoding/hex"
 	"errors"
 	"fmt"
+	"io"
 	"math/rand"
+	"os"
 	"reflect"
 	"strings"
 	"sync"
@@ -43,10 +47,15 @@ type ServerSystemConfig struct {
 	NEndpoints int
 	CmdLines   []string
 	MinioEnv   map[string]string
+	Checksum   string
 }
 
 // Diff - returns error on first difference found in two configs.
 func (s1 *ServerSystemConfig) Diff(s2 *ServerSystemConfig) error {
+	if s1.Checksum != s2.Checksum {
+		return fmt.Errorf("Expected MinIO binary checksum: %s, seen: %s", s1.Checksum, s2.Checksum)
+	}
+
 	ns1 := s1.NEndpoints
 	ns2 := s2.NEndpoints
 	if ns1 != ns2 {
@@ -82,7 +91,7 @@ func (s1 *ServerSystemConfig) Diff(s2 *ServerSystemConfig) error {
 			extra = append(extra, k)
 		}
 	}
-	msg := "Expected same MINIO_ environment variables and values across all servers: "
+	msg := "Expected MINIO_* environment name and values across all servers to be same: "
 	if len(missing) > 0 {
 		msg += fmt.Sprintf(`Missing environment values: %v. `, missing)
 	}
@@ -120,7 +129,7 @@ func getServerSystemCfg() *ServerSystemConfig {
 		}
 		envValues[envK] = logger.HashString(env.Get(envK, ""))
 	}
-	scfg := &ServerSystemConfig{NEndpoints: globalEndpoints.NEndpoints(), MinioEnv: envValues}
+	scfg := &ServerSystemConfig{NEndpoints: globalEndpoints.NEndpoints(), MinioEnv: envValues, Checksum: binaryChecksum}
 	var cmdLines []string
 	for _, ep := range globalEndpoints {
 		cmdLines = append(cmdLines, ep.CmdLine)
@@ -167,6 +176,18 @@ func (client *bootstrapRESTClient) String() string {
 	return client.gridConn.String()
 }
 
+var binaryChecksum = getBinaryChecksum()
+
+func getBinaryChecksum() string {
+	mw := md5.New()
+	b, err := os.Open(os.Args[0])
+	if err == nil {
+		defer b.Close()
+		io.Copy(mw, b)
+	}
+	return hex.EncodeToString(mw.Sum(nil))
+}
+
 func verifyServerSystemConfig(ctx context.Context, endpointServerPools EndpointServerPools, gm *grid.Manager) error {
 	srcCfg := getServerSystemCfg()
 	clnts := newBootstrapRESTClients(endpointServerPools, gm)
@@ -196,7 +217,7 @@ func verifyServerSystemConfig(ctx context.Context, endpointServerPools EndpointS
 				err := clnt.Verify(ctx, srcCfg)
 				mu.Lock()
 				if err != nil {
-					bootstrapTraceMsg(fmt.Sprintf("clnt.Verify: %v, endpoint: %s", err, clnt))
+					bootstrapTraceMsg(fmt.Sprintf("bootstrapVerify: %v, endpoint: %s", err, clnt))
 					if !isNetworkError(err) {
 						bootLogOnceIf(context.Background(), fmt.Errorf("%s has incorrect configuration: %w", clnt, err), "incorrect_"+clnt.String())
 						incorrectConfigs = append(incorrectConfigs, fmt.Errorf("%s has incorrect configuration: %w", clnt, err))

cmd/bootstrap-peer-server_gen.go

@@ -79,6 +79,12 @@ func (z *ServerSystemConfig) DecodeMsg(dc *msgp.Reader) (err error) {
 				}
 				z.MinioEnv[za0002] = za0003
 			}
+		case "Checksum":
+			z.Checksum, err = dc.ReadString()
+			if err != nil {
+				err = msgp.WrapError(err, "Checksum")
+				return
+			}
 		default:
 			err = dc.Skip()
 			if err != nil {
@@ -92,9 +98,9 @@ func (z *ServerSystemConfig) DecodeMsg(dc *msgp.Reader) (err error) {
 
 // EncodeMsg implements msgp.Encodable
 func (z *ServerSystemConfig) EncodeMsg(en *msgp.Writer) (err error) {
-	// map header, size 3
+	// map header, size 4
 	// write "NEndpoints"
-	err = en.Append(0x83, 0xaa, 0x4e, 0x45, 0x6e, 0x64, 0x70, 0x6f, 0x69, 0x6e, 0x74, 0x73)
+	err = en.Append(0x84, 0xaa, 0x4e, 0x45, 0x6e, 0x64, 0x70, 0x6f, 0x69, 0x6e, 0x74, 0x73)
 	if err != nil {
 		return
 	}
@@ -142,15 +148,25 @@ func (z *ServerSystemConfig) EncodeMsg(en *msgp.Writer) (err error) {
 			return
 		}
 	}
+	// write "Checksum"
+	err = en.Append(0xa8, 0x43, 0x68, 0x65, 0x63, 0x6b, 0x73, 0x75, 0x6d)
+	if err != nil {
+		return
+	}
+	err = en.WriteString(z.Checksum)
+	if err != nil {
+		err = msgp.WrapError(err, "Checksum")
+		return
+	}
 	return
 }
 
 // MarshalMsg implements msgp.Marshaler
 func (z *ServerSystemConfig) MarshalMsg(b []byte) (o []byte, err error) {
 	o = msgp.Require(b, z.Msgsize())
-	// map header, size 3
+	// map header, size 4
 	// string "NEndpoints"
-	o = append(o, 0x83, 0xaa, 0x4e, 0x45, 0x6e, 0x64, 0x70, 0x6f, 0x69, 0x6e, 0x74, 0x73)
+	o = append(o, 0x84, 0xaa, 0x4e, 0x45, 0x6e, 0x64, 0x70, 0x6f, 0x69, 0x6e, 0x74, 0x73)
 	o = msgp.AppendInt(o, z.NEndpoints)
 	// string "CmdLines"
 	o = append(o, 0xa8, 0x43, 0x6d, 0x64, 0x4c, 0x69, 0x6e, 0x65, 0x73)
@@ -165,6 +181,9 @@ func (z *ServerSystemConfig) MarshalMsg(b []byte) (o []byte, err error) {
 		o = msgp.AppendString(o, za0002)
 		o = msgp.AppendString(o, za0003)
 	}
+	// string "Checksum"
+	o = append(o, 0xa8, 0x43, 0x68, 0x65, 0x63, 0x6b, 0x73, 0x75, 0x6d)
+	o = msgp.AppendString(o, z.Checksum)
 	return
 }
 
@@ -241,6 +260,12 @@ func (z *ServerSystemConfig) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				}
 				z.MinioEnv[za0002] = za0003
 			}
+		case "Checksum":
+			z.Checksum, bts, err = msgp.ReadStringBytes(bts)
+			if err != nil {
+				err = msgp.WrapError(err, "Checksum")
+				return
+			}
 		default:
 			bts, err = msgp.Skip(bts)
 			if err != nil {
@@ -266,5 +291,6 @@ func (z *ServerSystemConfig) Msgsize() (s int) {
 			s += msgp.StringPrefixSize + len(za0002) + msgp.StringPrefixSize + len(za0003)
 		}
 	}
+	s += 9 + msgp.StringPrefixSize + len(z.Checksum)
 	return
 }

cmd/bucket-handlers.go

@@ -51,7 +51,6 @@ import (
 	sse "github.com/minio/minio/internal/bucket/encryption"
 	objectlock "github.com/minio/minio/internal/bucket/object/lock"
 	"github.com/minio/minio/internal/bucket/replication"
-	"github.com/minio/minio/internal/config/cache"
 	"github.com/minio/minio/internal/config/dns"
 	"github.com/minio/minio/internal/crypto"
 	"github.com/minio/minio/internal/event"
@@ -72,6 +71,8 @@ const (
 
 	xMinIOErrCodeHeader = "x-minio-error-code"
 	xMinIOErrDescHeader = "x-minio-error-desc"
+
+	postPolicyBucketTagging = "tagging"
 )
 
 // Check if there are buckets on server without corresponding entry in etcd backend and
@@ -90,7 +91,7 @@ const (
 // -- If IP of the entry doesn't match, this means entry is
 //
 //	for another instance. Log an error to console.
-func initFederatorBackend(buckets []BucketInfo, objLayer ObjectLayer) {
+func initFederatorBackend(buckets []string, objLayer ObjectLayer) {
 	if len(buckets) == 0 {
 		return
 	}
@@ -111,10 +112,10 @@ func initFederatorBackend(buckets []BucketInfo, objLayer ObjectLayer) {
 	domainMissing := err == dns.ErrDomainMissing
 	if dnsBuckets != nil {
 		for _, bucket := range buckets {
-			bucketsSet.Add(bucket.Name)
+			bucketsSet.Add(bucket)
-			r, ok := dnsBuckets[bucket.Name]
+			r, ok := dnsBuckets[bucket]
 			if !ok {
-				bucketsToBeUpdated.Add(bucket.Name)
+				bucketsToBeUpdated.Add(bucket)
 				continue
 			}
 			if !globalDomainIPs.Intersection(set.CreateStringSet(getHostsSlice(r)...)).IsEmpty() {
@@ -133,7 +134,7 @@ func initFederatorBackend(buckets []BucketInfo, objLayer ObjectLayer) {
 				// but if we do see a difference with local domain IPs with
 				// hostSlice from etcd then we should update with newer
 				// domainIPs, we proceed to do that here.
-				bucketsToBeUpdated.Add(bucket.Name)
+				bucketsToBeUpdated.Add(bucket)
 				continue
 			}
 
@@ -142,7 +143,7 @@ func initFederatorBackend(buckets []BucketInfo, objLayer ObjectLayer) {
 			// bucket names are globally unique in federation at a given
 			// path prefix, name collision is not allowed. We simply log
 			// an error and continue.
-			bucketsInConflict.Add(bucket.Name)
+			bucketsInConflict.Add(bucket)
 		}
 	}
 
@@ -1340,22 +1341,6 @@ func (api objectAPIHandlers) PostPolicyBucketHandler(w http.ResponseWriter, r *h
 					continue
 				}
 
-				asize, err := objInfo.GetActualSize()
-				if err != nil {
-					asize = objInfo.Size
-				}
-
-				globalCacheConfig.Set(&cache.ObjectInfo{
-					Key:          objInfo.Name,
-					Bucket:       objInfo.Bucket,
-					ETag:         getDecryptedETag(formValues, objInfo, false),
-					ModTime:      objInfo.ModTime,
-					Expires:      objInfo.Expires.UTC().Format(http.TimeFormat),
-					CacheControl: objInfo.CacheControl,
-					Metadata:     cleanReservedKeys(objInfo.UserDefined),
-					Size:         asize,
-				})
-
 				fanOutResp = append(fanOutResp, minio.PutObjectFanOutResponse{
 					Key:          objInfo.Name,
 					ETag:         getDecryptedETag(formValues, objInfo, false),
@@ -1415,6 +1400,19 @@ func (api objectAPIHandlers) PostPolicyBucketHandler(w http.ResponseWriter, r *h
 		return
 	}
 
+	if formValues.Get(postPolicyBucketTagging) != "" {
+		tags, err := tags.ParseObjectXML(strings.NewReader(formValues.Get(postPolicyBucketTagging)))
+		if err != nil {
+			writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrMalformedPOSTRequest), r.URL)
+			return
+		}
+		tagsStr := tags.String()
+		opts.UserDefined[xhttp.AmzObjectTagging] = tagsStr
+	} else {
+		// avoid user set an invalid tag using `X-Amz-Tagging`
+		delete(opts.UserDefined, xhttp.AmzObjectTagging)
+	}
+
 	objInfo, err := objectAPI.PutObject(ctx, bucket, object, pReader, opts)
 	if err != nil {
 		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
@@ -1437,22 +1435,6 @@ func (api objectAPIHandlers) PostPolicyBucketHandler(w http.ResponseWriter, r *h
 		w.Header().Set(xhttp.Location, obj)
 	}
 
-	asize, err := objInfo.GetActualSize()
-	if err != nil {
-		asize = objInfo.Size
-	}
-
-	defer globalCacheConfig.Set(&cache.ObjectInfo{
-		Key:          objInfo.Name,
-		Bucket:       objInfo.Bucket,
-		ETag:         etag,
-		ModTime:      objInfo.ModTime,
-		Expires:      objInfo.ExpiresStr(),
-		CacheControl: objInfo.CacheControl,
-		Metadata:     cleanReservedKeys(objInfo.UserDefined),
-		Size:         asize,
-	})
-
 	// Notify object created event.
 	defer sendEvent(eventArgs{
 		EventName:    event.ObjectCreatedPost,

cmd/bucket-handlers_test.go

@@ -32,7 +32,7 @@ import (
 
 // Wrapper for calling RemoveBucket HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestRemoveBucketHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testRemoveBucketHandler, []string{"RemoveBucket"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testRemoveBucketHandler, endpoints: []string{"RemoveBucket"}})
 }
 
 func testRemoveBucketHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -78,7 +78,7 @@ func testRemoveBucketHandler(obj ObjectLayer, instanceType, bucketName string, a
 
 // Wrapper for calling GetBucketPolicy HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestGetBucketLocationHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testGetBucketLocationHandler, []string{"GetBucketLocation"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testGetBucketLocationHandler, endpoints: []string{"GetBucketLocation"}})
 }
 
 func testGetBucketLocationHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -220,7 +220,7 @@ func testGetBucketLocationHandler(obj ObjectLayer, instanceType, bucketName stri
 
 // Wrapper for calling HeadBucket HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestHeadBucketHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testHeadBucketHandler, []string{"HeadBucket"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testHeadBucketHandler, endpoints: []string{"HeadBucket"}})
 }
 
 func testHeadBucketHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -322,7 +322,7 @@ func testHeadBucketHandler(obj ObjectLayer, instanceType, bucketName string, api
 
 // Wrapper for calling TestListMultipartUploadsHandler tests for both Erasure multiple disks and single node setup.
 func TestListMultipartUploadsHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testListMultipartUploadsHandler, []string{"ListMultipartUploads"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testListMultipartUploadsHandler, endpoints: []string{"ListMultipartUploads"}})
 }
 
 // testListMultipartUploadsHandler - Tests validate listing of multipart uploads.
@@ -558,7 +558,7 @@ func testListMultipartUploadsHandler(obj ObjectLayer, instanceType, bucketName s
 
 // Wrapper for calling TestListBucketsHandler tests for both Erasure multiple disks and single node setup.
 func TestListBucketsHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testListBucketsHandler, []string{"ListBuckets"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testListBucketsHandler, endpoints: []string{"ListBuckets"}})
 }
 
 // testListBucketsHandler - Tests validate listing of buckets.
@@ -649,7 +649,7 @@ func testListBucketsHandler(obj ObjectLayer, instanceType, bucketName string, ap
 
 // Wrapper for calling DeleteMultipleObjects HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestAPIDeleteMultipleObjectsHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testAPIDeleteMultipleObjectsHandler, []string{"DeleteMultipleObjects", "PutBucketPolicy"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPIDeleteMultipleObjectsHandler, endpoints: []string{"DeleteMultipleObjects", "PutBucketPolicy"}})
 }
 
 func testAPIDeleteMultipleObjectsHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,

cmd/bucket-lifecycle-handlers_test.go

@@ -29,7 +29,7 @@ import (
 
 // Test S3 Bucket lifecycle APIs with wrong credentials
 func TestBucketLifecycleWrongCredentials(t *testing.T) {
-	ExecObjectLayerAPITest(t, testBucketLifecycleHandlersWrongCredentials, []string{"GetBucketLifecycle", "PutBucketLifecycle", "DeleteBucketLifecycle"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testBucketLifecycleHandlersWrongCredentials, endpoints: []string{"GetBucketLifecycle", "PutBucketLifecycle", "DeleteBucketLifecycle"}})
 }
 
 // Test for authentication
@@ -145,7 +145,7 @@ func testBucketLifecycleHandlersWrongCredentials(obj ObjectLayer, instanceType,
 
 // Test S3 Bucket lifecycle APIs
 func TestBucketLifecycle(t *testing.T) {
-	ExecObjectLayerAPITest(t, testBucketLifecycleHandlers, []string{"GetBucketLifecycle", "PutBucketLifecycle", "DeleteBucketLifecycle"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testBucketLifecycleHandlers, endpoints: []string{"GetBucketLifecycle", "PutBucketLifecycle", "DeleteBucketLifecycle"}})
 }
 
 // Simple tests of bucket lifecycle: PUT, GET, DELETE.

cmd/bucket-lifecycle.go

@@ -72,6 +72,7 @@ func NewLifecycleSys() *LifecycleSys {
 }
 
 func ilmTrace(startTime time.Time, duration time.Duration, oi ObjectInfo, event string) madmin.TraceInfo {
+	sz, _ := oi.GetActualSize()
 	return madmin.TraceInfo{
 		TraceType: madmin.TraceILM,
 		Time:      startTime,
@@ -79,6 +80,7 @@ func ilmTrace(startTime time.Time, duration time.Duration, oi ObjectInfo, event
 		FuncName:  event,
 		Duration:  duration,
 		Path:      pathJoin(oi.Bucket, oi.Name),
+		Bytes:     sz,
 		Error:     "",
 		Message:   getSource(4),
 		Custom:    map[string]string{"version-id": oi.VersionID},
@@ -277,6 +279,10 @@ func (es *expiryState) getWorkerCh(h uint64) chan<- expiryOp {
 }
 
 func (es *expiryState) ResizeWorkers(n int) {
+	if n == 0 {
+		n = 100
+	}
+
 	// Lock to avoid multiple resizes to happen at the same time.
 	es.mu.Lock()
 	defer es.mu.Unlock()
@@ -538,6 +544,10 @@ func (t *transitionState) UpdateWorkers(n int) {
 }
 
 func (t *transitionState) updateWorkers(n int) {
+	if n == 0 {
+		n = 100
+	}
+
 	for t.numWorkers < n {
 		go t.worker(t.objAPI)
 		t.numWorkers++
@@ -573,6 +583,10 @@ func enqueueTransitionImmediate(obj ObjectInfo, src lcEventSrc) {
 	if lc, err := globalLifecycleSys.Get(obj.Bucket); err == nil {
 		switch event := lc.Eval(obj.ToLifecycleOpts()); event.Action {
 		case lifecycle.TransitionAction, lifecycle.TransitionVersionAction:
+			if obj.DeleteMarker || obj.IsDir {
+				// nothing to transition
+				return
+			}
 			globalTransitionState.queueTransitionTask(obj, event, src)
 		}
 	}
@@ -727,7 +741,7 @@ func getTransitionedObjectReader(ctx context.Context, bucket, object string, rs
 		return nil, fmt.Errorf("transition storage class not configured: %w", err)
 	}
 
-	fn, off, length, err := NewGetObjectReader(rs, oi, opts)
+	fn, off, length, err := NewGetObjectReader(rs, oi, opts, h)
 	if err != nil {
 		return nil, ErrorRespToObjectError(err, bucket, object)
 	}

cmd/bucket-listobjects-handlers.go

@@ -202,7 +202,7 @@ func (api objectAPIHandlers) listObjectsV2Handler(ctx context.Context, w http.Re
 
 	if r.Header.Get(xMinIOExtract) == "true" && strings.Contains(prefix, archivePattern) {
 		// Initiate a list objects operation inside a zip file based in the input params
-		listObjectsV2Info, err = listObjectsV2InArchive(ctx, objectAPI, bucket, prefix, token, delimiter, maxKeys, fetchOwner, startAfter)
+		listObjectsV2Info, err = listObjectsV2InArchive(ctx, objectAPI, bucket, prefix, token, delimiter, maxKeys, startAfter, r.Header)
 	} else {
 		// Initiate a list objects operation based on the input params.
 		// On success would return back ListObjectsInfo object to be
@@ -231,7 +231,7 @@ func parseRequestToken(token string) (subToken string, nodeIndex int) {
 	if token == "" {
 		return token, -1
 	}
-	i := strings.Index(token, ":")
+	i := strings.Index(token, getKeySeparator())
 	if i < 0 {
 		return token, -1
 	}

cmd/bucket-metadata-sys.go

@@ -392,9 +392,7 @@ func (sys *BucketMetadataSys) GetReplicationConfig(ctx context.Context, bucket s
 		return nil, time.Time{}, BucketReplicationConfigNotFound{Bucket: bucket}
 	}
 	if reloaded {
-		globalBucketTargetSys.set(BucketInfo{
+		globalBucketTargetSys.set(bucket, meta)
-			Name: bucket,
-		}, meta)
 	}
 	return meta.replicationConfig, meta.ReplicationConfigUpdatedAt, nil
 }
@@ -413,9 +411,7 @@ func (sys *BucketMetadataSys) GetBucketTargetsConfig(bucket string) (*madmin.Buc
 		return nil, BucketRemoteTargetNotFound{Bucket: bucket}
 	}
 	if reloaded {
-		globalBucketTargetSys.set(BucketInfo{
+		globalBucketTargetSys.set(bucket, meta)
-			Name: bucket,
-		}, meta)
 	}
 	return meta.bucketTargetConfig, nil
 }
@@ -471,7 +467,7 @@ func (sys *BucketMetadataSys) GetConfig(ctx context.Context, bucket string) (met
 }
 
 // Init - initializes bucket metadata system for all buckets.
-func (sys *BucketMetadataSys) Init(ctx context.Context, buckets []BucketInfo, objAPI ObjectLayer) error {
+func (sys *BucketMetadataSys) Init(ctx context.Context, buckets []string, objAPI ObjectLayer) error {
 	if objAPI == nil {
 		return errServerNotInitialized
 	}
@@ -484,7 +480,7 @@ func (sys *BucketMetadataSys) Init(ctx context.Context, buckets []BucketInfo, ob
 }
 
 // concurrently load bucket metadata to speed up loading bucket metadata.
-func (sys *BucketMetadataSys) concurrentLoad(ctx context.Context, buckets []BucketInfo, failedBuckets map[string]struct{}) {
+func (sys *BucketMetadataSys) concurrentLoad(ctx context.Context, buckets []string, failedBuckets map[string]struct{}) {
 	g := errgroup.WithNErrs(len(buckets))
 	bucketMetas := make([]BucketMetadata, len(buckets))
 	for index := range buckets {
@@ -494,8 +490,8 @@ func (sys *BucketMetadataSys) concurrentLoad(ctx context.Context, buckets []Buck
 			// herd upon start up sequence.
 			time.Sleep(25*time.Millisecond + time.Duration(rand.Int63n(int64(100*time.Millisecond))))
 
-			_, _ = sys.objAPI.HealBucket(ctx, buckets[index].Name, madmin.HealOpts{Recreate: true})
+			_, _ = sys.objAPI.HealBucket(ctx, buckets[index], madmin.HealOpts{Recreate: true})
-			meta, err := loadBucketMetadata(ctx, sys.objAPI, buckets[index].Name)
+			meta, err := loadBucketMetadata(ctx, sys.objAPI, buckets[index])
 			if err != nil {
 				return err
 			}
@@ -508,7 +504,7 @@ func (sys *BucketMetadataSys) concurrentLoad(ctx context.Context, buckets []Buck
 	for index, err := range errs {
 		if err != nil {
 			internalLogOnceIf(ctx, fmt.Errorf("Unable to load bucket metadata, will be retried: %w", err),
-				"load-bucket-metadata-"+buckets[index].Name, logger.WarningKind)
+				"load-bucket-metadata-"+buckets[index], logger.WarningKind)
 		}
 	}
 
@@ -519,7 +515,7 @@ func (sys *BucketMetadataSys) concurrentLoad(ctx context.Context, buckets []Buck
 		if errs[i] != nil {
 			continue
 		}
-		sys.metadataMap[buckets[i].Name] = meta
+		sys.metadataMap[buckets[i]] = meta
 	}
 	sys.Unlock()
 
@@ -528,7 +524,7 @@ func (sys *BucketMetadataSys) concurrentLoad(ctx context.Context, buckets []Buck
 			if failedBuckets == nil {
 				failedBuckets = make(map[string]struct{})
 			}
-			failedBuckets[buckets[i].Name] = struct{}{}
+			failedBuckets[buckets[i]] = struct{}{}
 			continue
 		}
 		globalEventNotifier.set(buckets[i], meta)   // set notification targets
@@ -548,7 +544,7 @@ func (sys *BucketMetadataSys) refreshBucketsMetadataLoop(ctx context.Context, fa
 		case <-ctx.Done():
 			return
 		case <-t.C:
-			buckets, err := sys.objAPI.ListBuckets(ctx, BucketOptions{})
+			buckets, err := sys.objAPI.ListBuckets(ctx, BucketOptions{NoMetadata: true})
 			if err != nil {
 				internalLogIf(ctx, err, logger.WarningKind)
 				break
@@ -579,8 +575,8 @@ func (sys *BucketMetadataSys) refreshBucketsMetadataLoop(ctx context.Context, fa
 
 				// Initialize the failed buckets
 				if _, ok := failedBuckets[buckets[i].Name]; ok {
-					globalEventNotifier.set(buckets[i], meta)
+					globalEventNotifier.set(buckets[i].Name, meta)
-					globalBucketTargetSys.set(buckets[i], meta)
+					globalBucketTargetSys.set(buckets[i].Name, meta)
 					delete(failedBuckets, buckets[i].Name)
 				}
 
@@ -600,8 +596,8 @@ func (sys *BucketMetadataSys) Initialized() bool {
 }
 
 // Loads bucket metadata for all buckets into BucketMetadataSys.
-func (sys *BucketMetadataSys) init(ctx context.Context, buckets []BucketInfo) {
+func (sys *BucketMetadataSys) init(ctx context.Context, buckets []string) {
-	count := 100 // load 100 bucket metadata at a time.
+	count := globalEndpoints.ESCount() * 10
 	failedBuckets := make(map[string]struct{})
 	for {
 		if len(buckets) < count {

cmd/bucket-policy-handlers_test.go

@@ -107,7 +107,7 @@ func getAnonWriteOnlyObjectPolicy(bucketName, prefix string) *policy.BucketPolic
 
 // Wrapper for calling Create Bucket and ensure we get one and only one success.
 func TestCreateBucket(t *testing.T) {
-	ExecObjectLayerAPITest(t, testCreateBucket, []string{"MakeBucket"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testCreateBucket, endpoints: []string{"MakeBucket"}})
 }
 
 // testCreateBucket - Test for calling Create Bucket and ensure we get one and only one success.
@@ -154,7 +154,7 @@ func testCreateBucket(obj ObjectLayer, instanceType, bucketName string, apiRoute
 
 // Wrapper for calling Put Bucket Policy HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestPutBucketPolicyHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testPutBucketPolicyHandler, []string{"PutBucketPolicy"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testPutBucketPolicyHandler, endpoints: []string{"PutBucketPolicy"}})
 }
 
 // testPutBucketPolicyHandler - Test for Bucket policy end point.
@@ -373,7 +373,7 @@ func testPutBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 
 // Wrapper for calling Get Bucket Policy HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestGetBucketPolicyHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testGetBucketPolicyHandler, []string{"PutBucketPolicy", "GetBucketPolicy"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testGetBucketPolicyHandler, endpoints: []string{"PutBucketPolicy", "GetBucketPolicy"}})
 }
 
 // testGetBucketPolicyHandler - Test for end point which fetches the access policy json of the given bucket.
@@ -577,7 +577,7 @@ func testGetBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 
 // Wrapper for calling Delete Bucket Policy HTTP handler tests for both Erasure multiple disks and single node setup.
 func TestDeleteBucketPolicyHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testDeleteBucketPolicyHandler, []string{"PutBucketPolicy", "DeleteBucketPolicy"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testDeleteBucketPolicyHandler, endpoints: []string{"PutBucketPolicy", "DeleteBucketPolicy"}})
 }
 
 // testDeleteBucketPolicyHandler - Test for Delete bucket policy end point.

cmd/bucket-replication-utils.go

@@ -534,7 +534,7 @@ func getHealReplicateObjectInfo(oi ObjectInfo, rcfg replicationConfig) Replicate
 	rstate.ReplicateDecisionStr = dsc.String()
 	asz, _ := oi.GetActualSize()
 
-	return ReplicateObjectInfo{
+	r := ReplicateObjectInfo{
 		Name:                       oi.Name,
 		Size:                       oi.Size,
 		ActualSize:                 asz,
@@ -558,6 +558,10 @@ func getHealReplicateObjectInfo(oi ObjectInfo, rcfg replicationConfig) Replicate
 		SSEC:                 crypto.SSEC.IsEncrypted(oi.UserDefined),
 		UserTags:             oi.UserTags,
 	}
+	if r.SSEC {
+		r.Checksum = oi.Checksum
+	}
+	return r
 }
 
 // ReplicationState - returns replication state using other internal replication metadata in ObjectInfo

cmd/bucket-replication.go

@@ -19,6 +19,7 @@ package cmd
 
 import (
 	"context"
+	"encoding/base64"
 	"encoding/binary"
 	"errors"
 	"fmt"
@@ -74,8 +75,9 @@ const (
 	ObjectLockRetentionTimestamp = "objectlock-retention-timestamp"
 	// ObjectLockLegalHoldTimestamp - the last time a legal hold metadata modification happened on this cluster for this object version
 	ObjectLockLegalHoldTimestamp = "objectlock-legalhold-timestamp"
-	// ReplicationWorkerMultiplier is suggested worker multiplier if traffic exceeds replication worker capacity
+
-	ReplicationWorkerMultiplier = 1.5
+	// ReplicationSsecChecksumHeader - the encrypted checksum of the SSE-C encrypted object.
+	ReplicationSsecChecksumHeader = "X-Minio-Replication-Ssec-Crc"
 )
 
 // gets replication config associated to a given bucket name.
@@ -763,12 +765,34 @@ func (m caseInsensitiveMap) Lookup(key string) (string, bool) {
 	return "", false
 }
 
-func putReplicationOpts(ctx context.Context, sc string, objInfo ObjectInfo) (putOpts minio.PutObjectOptions, err error) {
+func getCRCMeta(oi ObjectInfo, partNum int, h http.Header) map[string]string {
 	meta := make(map[string]string)
+	cs := oi.decryptChecksums(partNum, h)
+	for k, v := range cs {
+		cksum := hash.NewChecksumString(k, v)
+		if cksum == nil {
+			continue
+		}
+		if cksum.Valid() {
+			meta[cksum.Type.Key()] = v
+		}
+	}
+	return meta
+}
+
+func putReplicationOpts(ctx context.Context, sc string, objInfo ObjectInfo, partNum int) (putOpts minio.PutObjectOptions, err error) {
+	meta := make(map[string]string)
+	isSSEC := crypto.SSEC.IsEncrypted(objInfo.UserDefined)
+
 	for k, v := range objInfo.UserDefined {
 		// In case of SSE-C objects copy the allowed internal headers as well
-		if !crypto.SSEC.IsEncrypted(objInfo.UserDefined) || !slices.Contains(maps.Keys(validSSEReplicationHeaders), k) {
+		if !isSSEC || !slices.Contains(maps.Keys(validSSEReplicationHeaders), k) {
 			if stringsHasPrefixFold(k, ReservedMetadataPrefixLower) {
+				if strings.EqualFold(k, ReservedMetadataPrefixLower+"crc") {
+					for k, v := range getCRCMeta(objInfo, partNum, nil) {
+						meta[k] = v
+					}
+				}
 				continue
 			}
 			if isStandardHeader(k) {
@@ -782,6 +806,17 @@ func putReplicationOpts(ctx context.Context, sc string, objInfo ObjectInfo) (put
 		}
 	}
 
+	if len(objInfo.Checksum) > 0 {
+		// Add encrypted CRC to metadata for SSE-C objects.
+		if isSSEC {
+			meta[ReplicationSsecChecksumHeader] = base64.StdEncoding.EncodeToString(objInfo.Checksum)
+		} else {
+			for k, v := range getCRCMeta(objInfo, 0, nil) {
+				meta[k] = v
+			}
+		}
+	}
+
 	if sc == "" && (objInfo.StorageClass == storageclass.STANDARD || objInfo.StorageClass == storageclass.RRS) {
 		sc = objInfo.StorageClass
 	}
@@ -1201,7 +1236,12 @@ func (ri ReplicateObjectInfo) replicateObject(ctx context.Context, objectAPI Obj
 	// make sure we have the latest metadata for metrics calculation
 	rinfo.PrevReplicationStatus = objInfo.TargetReplicationStatus(tgt.ARN)
 
-	size, err := objInfo.GetActualSize()
+	// Set the encrypted size for SSE-C objects
+	var size int64
+	if crypto.SSEC.IsEncrypted(objInfo.UserDefined) {
+		size = objInfo.Size
+	} else {
+		size, err = objInfo.GetActualSize()
 		if err != nil {
 			replLogIf(ctx, err)
 			sendEvent(eventArgs{
@@ -1213,6 +1253,7 @@ func (ri ReplicateObjectInfo) replicateObject(ctx context.Context, objectAPI Obj
 			})
 			return
 		}
+	}
 
 	if tgt.Bucket == "" {
 		replLogIf(ctx, fmt.Errorf("unable to replicate object %s(%s), bucket is empty for target %s", objInfo.Name, objInfo.VersionID, tgt.EndpointURL()))
@@ -1239,7 +1280,7 @@ func (ri ReplicateObjectInfo) replicateObject(ctx context.Context, objectAPI Obj
 	// use core client to avoid doing multipart on PUT
 	c := &minio.Core{Client: tgt.Client}
 
-	putOpts, err := putReplicationOpts(ctx, tgt.StorageClass, objInfo)
+	putOpts, err := putReplicationOpts(ctx, tgt.StorageClass, objInfo, 0)
 	if err != nil {
 		replLogIf(ctx, fmt.Errorf("failure setting options for replication bucket:%s err:%w", bucket, err))
 		sendEvent(eventArgs{
@@ -1377,7 +1418,8 @@ func (ri ReplicateObjectInfo) replicateAll(ctx context.Context, objectAPI Object
 	}
 
 	// Set the encrypted size for SSE-C objects
-	if crypto.SSEC.IsEncrypted(objInfo.UserDefined) {
+	isSSEC := crypto.SSEC.IsEncrypted(objInfo.UserDefined)
+	if isSSEC {
 		size = objInfo.Size
 	}
 
@@ -1437,6 +1479,13 @@ func (ri ReplicateObjectInfo) replicateAll(ctx context.Context, objectAPI Object
 			return
 		}
 	} else {
+		// SSEC objects will refuse HeadObject without the decryption key.
+		// Ignore the error, since we know the object exists and versioning prevents overwriting existing versions.
+		if isSSEC && strings.Contains(cerr.Error(), errorCodes[ErrSSEEncryptedObject].Description) {
+			rinfo.ReplicationStatus = replication.Completed
+			rinfo.ReplicationAction = replicateNone
+			goto applyAction
+		}
 		// if target returns error other than NoSuchKey, defer replication attempt
 		if minio.IsNetworkOrHostDown(cerr, true) && !globalBucketTargetSys.isOffline(tgt.EndpointURL()) {
 			globalBucketTargetSys.markOffline(tgt.EndpointURL())
@@ -1464,6 +1513,7 @@ func (ri ReplicateObjectInfo) replicateAll(ctx context.Context, objectAPI Object
 			return
 		}
 	}
+applyAction:
 	rinfo.ReplicationStatus = replication.Completed
 	rinfo.Size = size
 	rinfo.ReplicationAction = rAction
@@ -1506,7 +1556,7 @@ func (ri ReplicateObjectInfo) replicateAll(ctx context.Context, objectAPI Object
 		}
 	} else {
 		var putOpts minio.PutObjectOptions
-		putOpts, err = putReplicationOpts(ctx, tgt.StorageClass, objInfo)
+		putOpts, err = putReplicationOpts(ctx, tgt.StorageClass, objInfo, 0)
 		if err != nil {
 			replLogIf(ctx, fmt.Errorf("failed to set replicate options for object %s/%s(%s) (target %s) err:%w", bucket, objInfo.Name, objInfo.VersionID, tgt.EndpointURL(), err))
 			sendEvent(eventArgs{
@@ -1599,11 +1649,12 @@ func replicateObjectWithMultipart(ctx context.Context, c *minio.Core, bucket, ob
 	var (
 		hr     *hash.Reader
 		pInfo  minio.ObjectPart
+		isSSEC = crypto.SSEC.IsEncrypted(objInfo.UserDefined)
 	)
 
 	var objectSize int64
 	for _, partInfo := range objInfo.Parts {
-		if crypto.SSEC.IsEncrypted(objInfo.UserDefined) {
+		if isSSEC {
 			hr, err = hash.NewReader(ctx, io.LimitReader(r, partInfo.Size), partInfo.Size, "", "", partInfo.ActualSize)
 		} else {
 			hr, err = hash.NewReader(ctx, io.LimitReader(r, partInfo.ActualSize), partInfo.ActualSize, "", "", partInfo.ActualSize)
@@ -1614,12 +1665,18 @@ func replicateObjectWithMultipart(ctx context.Context, c *minio.Core, bucket, ob
 
 		cHeader := http.Header{}
 		cHeader.Add(xhttp.MinIOSourceReplicationRequest, "true")
+		if !isSSEC {
+			crc := getCRCMeta(objInfo, partInfo.Number, nil) // No SSE-C keys here.
+			for k, v := range crc {
+				cHeader.Add(k, v)
+			}
+		}
 		popts := minio.PutObjectPartOptions{
 			SSE:          opts.ServerSideEncryption,
 			CustomHeader: cHeader,
 		}
 
-		if crypto.SSEC.IsEncrypted(objInfo.UserDefined) {
+		if isSSEC {
 			objectSize += partInfo.Size
 			pInfo, err = c.PutObjectPart(ctx, bucket, object, uploadID, partInfo.Number, hr, partInfo.Size, popts)
 		} else {
@@ -1629,20 +1686,30 @@ func replicateObjectWithMultipart(ctx context.Context, c *minio.Core, bucket, ob
 		if err != nil {
 			return err
 		}
-		if !crypto.SSEC.IsEncrypted(objInfo.UserDefined) && pInfo.Size != partInfo.ActualSize {
+		if !isSSEC && pInfo.Size != partInfo.ActualSize {
 			return fmt.Errorf("Part size mismatch: got %d, want %d", pInfo.Size, partInfo.ActualSize)
 		}
 		uploadedParts = append(uploadedParts, minio.CompletePart{
 			PartNumber:     pInfo.PartNumber,
 			ETag:           pInfo.ETag,
+			ChecksumCRC32:  pInfo.ChecksumCRC32,
+			ChecksumCRC32C: pInfo.ChecksumCRC32C,
+			ChecksumSHA1:   pInfo.ChecksumSHA1,
+			ChecksumSHA256: pInfo.ChecksumSHA256,
 		})
 	}
+	userMeta := map[string]string{
+		validSSEReplicationHeaders[ReservedMetadataPrefix+"Actual-Object-Size"]: objInfo.UserDefined[ReservedMetadataPrefix+"actual-size"],
+	}
+	if isSSEC && objInfo.UserDefined[ReplicationSsecChecksumHeader] != "" {
+		userMeta[ReplicationSsecChecksumHeader] = objInfo.UserDefined[ReplicationSsecChecksumHeader]
+	}
 
 	// really big value but its okay on heavily loaded systems. This is just tail end timeout.
 	cctx, ccancel := context.WithTimeout(ctx, 10*time.Minute)
 	defer ccancel()
 	_, err = c.CompleteMultipartUpload(cctx, bucket, object, uploadID, uploadedParts, minio.PutObjectOptions{
-		UserMetadata: map[string]string{validSSEReplicationHeaders[ReservedMetadataPrefix+"Actual-Object-Size"]: objInfo.UserDefined[ReservedMetadataPrefix+"actual-size"]},
+		UserMetadata: userMeta,
 		Internal: minio.AdvancedPutOptions{
 			SourceMTime: objInfo.ModTime,
 			SourceETag:  objInfo.ETag,
@@ -2179,12 +2246,12 @@ type proxyResult struct {
 
 // get Reader from replication target if active-active replication is in place and
 // this node returns a 404
-func proxyGetToReplicationTarget(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, _ http.Header, opts ObjectOptions, proxyTargets *madmin.BucketTargets) (gr *GetObjectReader, proxy proxyResult, err error) {
+func proxyGetToReplicationTarget(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, opts ObjectOptions, proxyTargets *madmin.BucketTargets) (gr *GetObjectReader, proxy proxyResult, err error) {
 	tgt, oi, proxy := proxyHeadToRepTarget(ctx, bucket, object, rs, opts, proxyTargets)
 	if !proxy.Proxy {
 		return nil, proxy, nil
 	}
-	fn, _, _, err := NewGetObjectReader(nil, oi, opts)
+	fn, _, _, err := NewGetObjectReader(nil, oi, opts, h)
 	if err != nil {
 		return nil, proxy, err
 	}
@@ -2369,7 +2436,9 @@ func scheduleReplication(ctx context.Context, oi ObjectInfo, o ObjectLayer, dsc
 		SSEC:                 crypto.SSEC.IsEncrypted(oi.UserDefined),
 		UserTags:             oi.UserTags,
 	}
-
+	if ri.SSEC {
+		ri.Checksum = oi.Checksum
+	}
 	if dsc.Synchronous() {
 		replicateObject(ctx, ri, o)
 	} else {
@@ -2849,17 +2918,19 @@ func (s *replicationResyncer) resyncBucket(ctx context.Context, objectAPI Object
 						ReplicationProxyRequest: "false",
 					},
 				})
+				sz := roi.Size
 				if err != nil {
 					if roi.DeleteMarker && isErrMethodNotAllowed(ErrorRespToObjectError(err, opts.bucket, roi.Name)) {
 						st.ReplicatedCount++
 					} else {
 						st.FailedCount++
 					}
+					sz = 0
 				} else {
 					st.ReplicatedCount++
 					st.ReplicatedSize += roi.Size
 				}
-				traceFn(err)
+				traceFn(sz, err)
 				select {
 				case <-ctx.Done():
 					return
@@ -2974,17 +3045,17 @@ func (s *replicationResyncer) start(ctx context.Context, objAPI ObjectLayer, opt
 	return nil
 }
 
-func (s *replicationResyncer) trace(resyncID string, path string) func(err error) {
+func (s *replicationResyncer) trace(resyncID string, path string) func(sz int64, err error) {
 	startTime := time.Now()
-	return func(err error) {
+	return func(sz int64, err error) {
 		duration := time.Since(startTime)
 		if globalTrace.NumSubscribers(madmin.TraceReplicationResync) > 0 {
-			globalTrace.Publish(replicationResyncTrace(resyncID, startTime, duration, path, err))
+			globalTrace.Publish(replicationResyncTrace(resyncID, startTime, duration, path, err, sz))
 		}
 	}
 }
 
-func replicationResyncTrace(resyncID string, startTime time.Time, duration time.Duration, path string, err error) madmin.TraceInfo {
+func replicationResyncTrace(resyncID string, startTime time.Time, duration time.Duration, path string, err error, sz int64) madmin.TraceInfo {
 	var errStr string
 	if err != nil {
 		errStr = err.Error()
@@ -2998,6 +3069,7 @@ func replicationResyncTrace(resyncID string, startTime time.Time, duration time.
 		Duration:  duration,
 		Path:      path,
 		Error:     errStr,
+		Bytes:     sz,
 	}
 }
 
@@ -3014,7 +3086,7 @@ func (p *ReplicationPool) deleteResyncMetadata(ctx context.Context, bucket strin
 }
 
 // initResync - initializes bucket replication resync for all buckets.
-func (p *ReplicationPool) initResync(ctx context.Context, buckets []BucketInfo, objAPI ObjectLayer) error {
+func (p *ReplicationPool) initResync(ctx context.Context, buckets []string, objAPI ObjectLayer) error {
 	if objAPI == nil {
 		return errServerNotInitialized
 	}
@@ -3023,7 +3095,7 @@ func (p *ReplicationPool) initResync(ctx context.Context, buckets []BucketInfo,
 	return nil
 }
 
-func (p *ReplicationPool) startResyncRoutine(ctx context.Context, buckets []BucketInfo, objAPI ObjectLayer) {
+func (p *ReplicationPool) startResyncRoutine(ctx context.Context, buckets []string, objAPI ObjectLayer) {
 	r := rand.New(rand.NewSource(time.Now().UnixNano()))
 	// Run the replication resync in a loop
 	for {
@@ -3041,13 +3113,13 @@ func (p *ReplicationPool) startResyncRoutine(ctx context.Context, buckets []Buck
 }
 
 // Loads bucket replication resync statuses into memory.
-func (p *ReplicationPool) loadResync(ctx context.Context, buckets []BucketInfo, objAPI ObjectLayer) error {
+func (p *ReplicationPool) loadResync(ctx context.Context, buckets []string, objAPI ObjectLayer) error {
 	// Make sure only one node running resync on the cluster.
 	ctx, cancel := globalLeaderLock.GetLock(ctx)
 	defer cancel()
 
 	for index := range buckets {
-		bucket := buckets[index].Name
+		bucket := buckets[index]
 
 		meta, err := loadBucketResyncMetadata(ctx, bucket, objAPI)
 		if err != nil {

cmd/bucket-targets.go

@@ -612,7 +612,7 @@ func (sys *BucketTargetSys) UpdateAllTargets(bucket string, tgts *madmin.BucketT
 }
 
 // create minio-go clients for buckets having remote targets
-func (sys *BucketTargetSys) set(bucket BucketInfo, meta BucketMetadata) {
+func (sys *BucketTargetSys) set(bucket string, meta BucketMetadata) {
 	cfg := meta.bucketTargetConfig
 	if cfg == nil || cfg.Empty() {
 		return
@@ -626,9 +626,9 @@ func (sys *BucketTargetSys) set(bucket BucketInfo, meta BucketMetadata) {
 			continue
 		}
 		sys.arnRemotesMap[tgt.Arn] = arnTarget{Client: tgtClient}
-		sys.updateBandwidthLimit(bucket.Name, tgt.Arn, tgt.BandwidthLimit)
+		sys.updateBandwidthLimit(bucket, tgt.Arn, tgt.BandwidthLimit)
 	}
-	sys.targetsMap[bucket.Name] = cfg.Targets
+	sys.targetsMap[bucket] = cfg.Targets
 }
 
 // Returns a minio-go Client configured to access remote host described in replication target config.

cmd/common-main.go

@@ -402,6 +402,29 @@ func buildServerCtxt(ctx *cli.Context, ctxt *serverCtxt) (err error) {
 		ctxt.certsDirSet = true
 	}
 
+	memAvailable := availableMemory()
+	if ctx.IsSet("memlimit") || ctx.GlobalIsSet("memlimit") {
+		memlimit := ctx.String("memlimit")
+		if memlimit == "" {
+			memlimit = ctx.GlobalString("memlimit")
+		}
+		mlimit, err := humanize.ParseBytes(memlimit)
+		if err != nil {
+			return err
+		}
+		if mlimit > memAvailable {
+			logger.Info("WARNING: maximum memory available (%s) smaller than specified --memlimit=%s, ignoring --memlimit value",
+				humanize.IBytes(memAvailable), memlimit)
+		}
+		ctxt.MemLimit = mlimit
+	} else {
+		ctxt.MemLimit = memAvailable
+	}
+
+	if memAvailable < ctxt.MemLimit {
+		ctxt.MemLimit = memAvailable
+	}
+
 	ctxt.FTP = ctx.StringSlice("ftp")
 	ctxt.SFTP = ctx.StringSlice("sftp")
 	ctxt.Interface = ctx.String("interface")
@@ -410,7 +433,6 @@ func buildServerCtxt(ctx *cli.Context, ctxt *serverCtxt) (err error) {
 	ctxt.RecvBufSize = ctx.Int("recv-buf-size")
 	ctxt.IdleTimeout = ctx.Duration("idle-timeout")
 	ctxt.UserTimeout = ctx.Duration("conn-user-timeout")
-	ctxt.ShutdownTimeout = ctx.Duration("shutdown-timeout")
 
 	if conf := ctx.String("config"); len(conf) > 0 {
 		err = mergeServerCtxtFromConfigFile(conf, ctxt)

cmd/config-current.go

@@ -240,6 +240,11 @@ func initHelp() {
 			Description: "manage Browser HTTP specific features, such as Security headers, etc.",
 			Optional:    true,
 		},
+		config.HelpKV{
+			Key:         config.ILMSubSys,
+			Description: "manage ILM settings for expiration and transition workers",
+			Optional:    true,
+		},
 	}
 
 	if globalIsErasure {
@@ -288,6 +293,7 @@ func initHelp() {
 		config.DriveSubSys:          drive.HelpDrive,
 		config.CacheSubSys:          cache.Help,
 		config.BrowserSubSys:        browser.Help,
+		config.ILMSubSys:            ilm.Help,
 	}
 
 	config.RegisterHelpSubSys(helpMap)

cmd/data-scanner.go

@@ -31,6 +31,7 @@ import (
 	"strconv"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/minio/madmin-go/v3"
@@ -250,6 +251,7 @@ type folderScanner struct {
 	scanMode              madmin.HealScanMode
 
 	weSleep    func() bool
+	shouldHeal func() bool
 
 	disks       []StorageAPI
 	disksQuorum int
@@ -304,11 +306,12 @@ type folderScanner struct {
 // The returned cache will always be valid, but may not be updated from the existing.
 // Before each operation sleepDuration is called which can be used to temporarily halt the scanner.
 // If the supplied context is canceled the function will return at the first chance.
-func scanDataFolder(ctx context.Context, disks []StorageAPI, basePath string, cache dataUsageCache, getSize getSizeFn, scanMode madmin.HealScanMode, weSleep func() bool) (dataUsageCache, error) {
+func scanDataFolder(ctx context.Context, disks []StorageAPI, drive *xlStorage, cache dataUsageCache, getSize getSizeFn, scanMode madmin.HealScanMode, weSleep func() bool) (dataUsageCache, error) {
 	switch cache.Info.Name {
 	case "", dataUsageRoot:
 		return cache, errors.New("internal error: root scan attempted")
 	}
+	basePath := drive.drivePath
 	updatePath, closeDisk := globalScannerMetrics.currentPathUpdater(basePath, cache.Info.Name)
 	defer closeDisk()
 
@@ -328,6 +331,26 @@ func scanDataFolder(ctx context.Context, disks []StorageAPI, basePath string, ca
 		disksQuorum:           len(disks) / 2,
 	}
 
+	var skipHeal atomic.Bool
+	if globalIsErasure || cache.Info.SkipHealing {
+		skipHeal.Store(true)
+	}
+
+	// Check if we should do healing at all.
+	s.shouldHeal = func() bool {
+		if skipHeal.Load() {
+			return false
+		}
+		if s.healObjectSelect == 0 {
+			return false
+		}
+		if di, _ := drive.DiskInfo(ctx, DiskInfoOptions{}); di.Healing {
+			skipHeal.Store(true)
+			return false
+		}
+		return true
+	}
+
 	// Enable healing in XL mode.
 	if globalIsErasure && !cache.Info.SkipHealing {
 		// Do a heal check on an object once every n cycles. Must divide into healFolderInclude
@@ -482,14 +505,9 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 				replication: replicationCfg,
 			}
 
-			item.heal.enabled = thisHash.modAlt(f.oldCache.Info.NextCycle/folder.objectHealProbDiv, f.healObjectSelect/folder.objectHealProbDiv) && globalIsErasure
+			item.heal.enabled = thisHash.modAlt(f.oldCache.Info.NextCycle/folder.objectHealProbDiv, f.healObjectSelect/folder.objectHealProbDiv) && f.shouldHeal()
 			item.heal.bitrot = f.scanMode == madmin.HealDeepScan
 
-			// if the drive belongs to an erasure set
-			// that is already being healed, skip the
-			// healing attempt on this drive.
-			item.heal.enabled = item.heal.enabled && f.healObjectSelect > 0
-
 			sz, err := f.getSize(item)
 			if err != nil && err != errIgnoreFileContrib {
 				wait() // wait to proceed to next entry.
@@ -652,7 +670,7 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 		}
 
 		// Scan for healing
-		if f.healObjectSelect == 0 || len(abandonedChildren) == 0 {
+		if len(abandonedChildren) == 0 || !f.shouldHeal() {
 			// If we are not heal scanning, return now.
 			break
 		}
@@ -687,6 +705,9 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 
 		healObjectsPrefix := color.Green("healObjects:")
 		for k := range abandonedChildren {
+			if !f.shouldHeal() {
+				break
+			}
 			bucket, prefix := path2BucketObject(k)
 			stopFn := globalScannerMetrics.time(scannerMetricCheckMissing)
 			f.updateCurrentPath(k)
@@ -720,6 +741,10 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 				},
 				// Some disks have data for this.
 				partial: func(entries metaCacheEntries, errs []error) {
+					if !f.shouldHeal() {
+						cancel()
+						return
+					}
 					entry, ok := entries.resolve(&resolver)
 					if !ok {
 						// check if we can get one entry at least
@@ -1233,7 +1258,7 @@ func evalActionFromLifecycle(ctx context.Context, lc lifecycle.Lifecycle, lr loc
 }
 
 func applyTransitionRule(event lifecycle.Event, src lcEventSrc, obj ObjectInfo) bool {
-	if obj.DeleteMarker {
+	if obj.DeleteMarker || obj.IsDir {
 		return false
 	}
 	globalTransitionState.queueTransitionTask(obj, event, src)

cmd/data-usage_test.go

@@ -26,6 +26,9 @@ import (
 	"path"
 	"path/filepath"
 	"testing"
+	"time"
+
+	"github.com/minio/minio/internal/cachevalue"
 )
 
 type usageTestFile struct {
@@ -61,10 +64,13 @@ func TestDataUsageUpdate(t *testing.T) {
 		}
 		return
 	}
-
+	xls := xlStorage{drivePath: base, diskInfoCache: cachevalue.New[DiskInfo]()}
+	xls.diskInfoCache.InitOnce(time.Second, cachevalue.Opts{}, func(ctx context.Context) (DiskInfo, error) {
+		return DiskInfo{Total: 1 << 40, Free: 1 << 40}, nil
+	})
 	weSleep := func() bool { return false }
 
-	got, err := scanDataFolder(context.Background(), nil, base, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize, 0, weSleep)
+	got, err := scanDataFolder(context.Background(), nil, &xls, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize, 0, weSleep)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -174,7 +180,7 @@ func TestDataUsageUpdate(t *testing.T) {
 	}
 	// Changed dir must be picked up in this many cycles.
 	for i := 0; i < dataUsageUpdateDirCycles; i++ {
-		got, err = scanDataFolder(context.Background(), nil, base, got, getSize, 0, weSleep)
+		got, err = scanDataFolder(context.Background(), nil, &xls, got, getSize, 0, weSleep)
 		got.Info.NextCycle++
 		if err != nil {
 			t.Fatal(err)
@@ -283,8 +289,12 @@ func TestDataUsageUpdatePrefix(t *testing.T) {
 	}
 
 	weSleep := func() bool { return false }
+	xls := xlStorage{drivePath: base, diskInfoCache: cachevalue.New[DiskInfo]()}
+	xls.diskInfoCache.InitOnce(time.Second, cachevalue.Opts{}, func(ctx context.Context) (DiskInfo, error) {
+		return DiskInfo{Total: 1 << 40, Free: 1 << 40}, nil
+	})
 
-	got, err := scanDataFolder(context.Background(), nil, base, dataUsageCache{Info: dataUsageCacheInfo{Name: "bucket"}}, getSize, 0, weSleep)
+	got, err := scanDataFolder(context.Background(), nil, &xls, dataUsageCache{Info: dataUsageCacheInfo{Name: "bucket"}}, getSize, 0, weSleep)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -419,7 +429,7 @@ func TestDataUsageUpdatePrefix(t *testing.T) {
 	}
 	// Changed dir must be picked up in this many cycles.
 	for i := 0; i < dataUsageUpdateDirCycles; i++ {
-		got, err = scanDataFolder(context.Background(), nil, base, got, getSize, 0, weSleep)
+		got, err = scanDataFolder(context.Background(), nil, &xls, got, getSize, 0, weSleep)
 		got.Info.NextCycle++
 		if err != nil {
 			t.Fatal(err)
@@ -567,8 +577,12 @@ func TestDataUsageCacheSerialize(t *testing.T) {
 		}
 		return
 	}
+	xls := xlStorage{drivePath: base, diskInfoCache: cachevalue.New[DiskInfo]()}
+	xls.diskInfoCache.InitOnce(time.Second, cachevalue.Opts{}, func(ctx context.Context) (DiskInfo, error) {
+		return DiskInfo{Total: 1 << 40, Free: 1 << 40}, nil
+	})
 	weSleep := func() bool { return false }
-	want, err := scanDataFolder(context.Background(), nil, base, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize, 0, weSleep)
+	want, err := scanDataFolder(context.Background(), nil, &xls, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize, 0, weSleep)
 	if err != nil {
 		t.Fatal(err)
 	}

cmd/encryption-v1.go

@@ -1025,9 +1025,11 @@ func DecryptObjectInfo(info *ObjectInfo, r *http.Request) (encrypted bool, err e
 	if encrypted {
 		if crypto.SSEC.IsEncrypted(info.UserDefined) {
 			if !(crypto.SSEC.IsRequested(headers) || crypto.SSECopy.IsRequested(headers)) {
+				if r.Header.Get(xhttp.MinIOSourceReplicationRequest) != "true" {
 					return encrypted, errEncryptedObject
 				}
 			}
+		}
 
 		if crypto.S3.IsEncrypted(info.UserDefined) && r.Header.Get(xhttp.AmzCopySource) == "" {
 			if crypto.SSEC.IsRequested(headers) || crypto.SSECopy.IsRequested(headers) {
@@ -1077,13 +1079,16 @@ func metadataEncrypter(key crypto.ObjectKey) objectMetaEncryptFn {
 }
 
 // metadataDecrypter reverses metadataEncrypter.
-func (o *ObjectInfo) metadataDecrypter() objectMetaDecryptFn {
+func (o *ObjectInfo) metadataDecrypter(h http.Header) objectMetaDecryptFn {
 	return func(baseKey string, input []byte) ([]byte, error) {
 		if len(input) == 0 {
 			return input, nil
 		}
-
+		var key []byte
-		key, err := decryptObjectMeta(nil, o.Bucket, o.Name, o.UserDefined)
+		if k, err := crypto.SSEC.ParseHTTP(h); err == nil {
+			key = k[:]
+		}
+		key, err := decryptObjectMeta(key, o.Bucket, o.Name, o.UserDefined)
 		if err != nil {
 			return nil, err
 		}
@@ -1095,13 +1100,13 @@ func (o *ObjectInfo) metadataDecrypter() objectMetaDecryptFn {
 
 // decryptChecksums will attempt to decode checksums and return it/them if set.
 // if part > 0, and we have the checksum for the part that will be returned.
-func (o *ObjectInfo) decryptPartsChecksums() {
+func (o *ObjectInfo) decryptPartsChecksums(h http.Header) {
 	data := o.Checksum
 	if len(data) == 0 {
 		return
 	}
 	if _, encrypted := crypto.IsEncrypted(o.UserDefined); encrypted {
-		decrypted, err := o.metadataDecrypter()("object-checksum", data)
+		decrypted, err := o.metadataDecrypter(h)("object-checksum", data)
 		if err != nil {
 			encLogIf(GlobalContext, err)
 			return
@@ -1157,15 +1162,17 @@ func (o *ObjectInfo) metadataEncryptFn(headers http.Header) (objectMetaEncryptFn
 
 // decryptChecksums will attempt to decode checksums and return it/them if set.
 // if part > 0, and we have the checksum for the part that will be returned.
-func (o *ObjectInfo) decryptChecksums(part int) map[string]string {
+func (o *ObjectInfo) decryptChecksums(part int, h http.Header) map[string]string {
 	data := o.Checksum
 	if len(data) == 0 {
 		return nil
 	}
 	if _, encrypted := crypto.IsEncrypted(o.UserDefined); encrypted {
-		decrypted, err := o.metadataDecrypter()("object-checksum", data)
+		decrypted, err := o.metadataDecrypter(h)("object-checksum", data)
 		if err != nil {
+			if err != crypto.ErrSecretKeyMismatch {
 				encLogIf(GlobalContext, err)
+			}
 			return nil
 		}
 		data = decrypted

cmd/endpoint.go

@@ -253,6 +253,14 @@ type PoolEndpoints struct {
 // EndpointServerPools - list of list of endpoints
 type EndpointServerPools []PoolEndpoints
 
+// ESCount returns the total number of erasure sets in this cluster
+func (l EndpointServerPools) ESCount() (count int) {
+	for _, p := range l {
+		count += p.SetCount
+	}
+	return
+}
+
 // GetNodes returns a sorted list of nodes in this cluster
 func (l EndpointServerPools) GetNodes() (nodes []Node) {
 	nodesMap := make(map[string]Node)

cmd/erasure-healing-common.go

@@ -20,6 +20,7 @@ package cmd
 import (
 	"bytes"
 	"context"
+	"slices"
 	"time"
 
 	"github.com/minio/madmin-go/v3"
@@ -253,6 +254,33 @@ func listOnlineDisks(disks []StorageAPI, partsMetadata []FileInfo, errs []error,
 	return onlineDisks, modTime, ""
 }
 
+// Convert verify or check parts returned error to integer representation
+func convPartErrToInt(err error) int {
+	err = unwrapAll(err)
+	switch err {
+	case nil:
+		return checkPartSuccess
+	case errFileNotFound, errFileVersionNotFound:
+		return checkPartFileNotFound
+	case errFileCorrupt:
+		return checkPartFileCorrupt
+	case errVolumeNotFound:
+		return checkPartVolumeNotFound
+	case errDiskNotFound:
+		return checkPartDiskNotFound
+	default:
+		return checkPartUnknown
+	}
+}
+
+func partNeedsHealing(partErrs []int) bool {
+	return slices.IndexFunc(partErrs, func(i int) bool { return i != checkPartSuccess && i != checkPartUnknown }) > -1
+}
+
+func hasPartErr(partErrs []int) bool {
+	return slices.IndexFunc(partErrs, func(i int) bool { return i != checkPartSuccess }) > -1
+}
+
 // disksWithAllParts - This function needs to be called with
 // []StorageAPI returned by listOnlineDisks. Returns,
 //
@@ -262,10 +290,19 @@ func listOnlineDisks(disks []StorageAPI, partsMetadata []FileInfo, errs []error,
 //     a not-found error or a hash-mismatch error.
 func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetadata []FileInfo,
 	errs []error, latestMeta FileInfo, bucket, object string,
-	scanMode madmin.HealScanMode) ([]StorageAPI, []error, time.Time,
+	scanMode madmin.HealScanMode,
-) {
+) (availableDisks []StorageAPI, dataErrsByDisk map[int][]int, dataErrsByPart map[int][]int) {
-	availableDisks := make([]StorageAPI, len(onlineDisks))
+	availableDisks = make([]StorageAPI, len(onlineDisks))
-	dataErrs := make([]error, len(onlineDisks))
+
+	dataErrsByDisk = make(map[int][]int, len(onlineDisks))
+	for i := range onlineDisks {
+		dataErrsByDisk[i] = make([]int, len(latestMeta.Parts))
+	}
+
+	dataErrsByPart = make(map[int][]int, len(latestMeta.Parts))
+	for i := range latestMeta.Parts {
+		dataErrsByPart[i] = make([]int, len(onlineDisks))
+	}
 
 	inconsistent := 0
 	for i, meta := range partsMetadata {
@@ -295,19 +332,21 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 		erasureDistributionReliable = false
 	}
 
+	metaErrs := make([]error, len(errs))
+
 	for i, onlineDisk := range onlineDisks {
 		if errs[i] != nil {
-			dataErrs[i] = errs[i]
+			metaErrs[i] = errs[i]
 			continue
 		}
 		if onlineDisk == OfflineDisk {
-			dataErrs[i] = errDiskNotFound
+			metaErrs[i] = errDiskNotFound
 			continue
 		}
 
 		meta := partsMetadata[i]
 		if !meta.ModTime.Equal(latestMeta.ModTime) || meta.DataDir != latestMeta.DataDir {
-			dataErrs[i] = errFileCorrupt
+			metaErrs[i] = errFileCorrupt
 			partsMetadata[i] = FileInfo{}
 			continue
 		}
@@ -315,7 +354,7 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 		if erasureDistributionReliable {
 			if !meta.IsValid() {
 				partsMetadata[i] = FileInfo{}
-				dataErrs[i] = errFileCorrupt
+				metaErrs[i] = errFileCorrupt
 				continue
 			}
 
@@ -325,46 +364,79 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 					// attempt a fix if possible, assuming other entries
 					// might have the right erasure distribution.
 					partsMetadata[i] = FileInfo{}
-					dataErrs[i] = errFileCorrupt
+					metaErrs[i] = errFileCorrupt
 					continue
 				}
 			}
 		}
+	}
+
+	// Copy meta errors to part errors
+	for i, err := range metaErrs {
+		if err != nil {
+			partErr := convPartErrToInt(err)
+			for p := range latestMeta.Parts {
+				dataErrsByPart[p][i] = partErr
+			}
+		}
+	}
+
+	for i, onlineDisk := range onlineDisks {
+		if metaErrs[i] != nil {
+			continue
+		}
+		meta := partsMetadata[i]
+
+		if meta.Deleted || meta.IsRemote() {
+			continue
+		}
 
 		// Always check data, if we got it.
 		if (len(meta.Data) > 0 || meta.Size == 0) && len(meta.Parts) > 0 {
 			checksumInfo := meta.Erasure.GetChecksumInfo(meta.Parts[0].Number)
-			dataErrs[i] = bitrotVerify(bytes.NewReader(meta.Data),
+			verifyErr := bitrotVerify(bytes.NewReader(meta.Data),
 				int64(len(meta.Data)),
 				meta.Erasure.ShardFileSize(meta.Size),
 				checksumInfo.Algorithm,
 				checksumInfo.Hash, meta.Erasure.ShardSize())
-			if dataErrs[i] == nil {
+			dataErrsByPart[0][i] = convPartErrToInt(verifyErr)
-				// All parts verified, mark it as all data available.
-				availableDisks[i] = onlineDisk
-			} else {
-				// upon errors just make that disk's fileinfo invalid
-				partsMetadata[i] = FileInfo{}
-			}
 			continue
 		}
 
+		var (
+			verifyErr  error
+			verifyResp *CheckPartsResp
+		)
+
 		meta.DataDir = latestMeta.DataDir
 		switch scanMode {
 		case madmin.HealDeepScan:
 			// disk has a valid xl.meta but may not have all the
 			// parts. This is considered an outdated disk, since
 			// it needs healing too.
-			if !meta.Deleted && !meta.IsRemote() {
+			verifyResp, verifyErr = onlineDisk.VerifyFile(ctx, bucket, object, meta)
-				dataErrs[i] = onlineDisk.VerifyFile(ctx, bucket, object, meta)
+		default:
+			verifyResp, verifyErr = onlineDisk.CheckParts(ctx, bucket, object, meta)
+		}
+
+		for p := range latestMeta.Parts {
+			if verifyErr != nil {
+				dataErrsByPart[p][i] = convPartErrToInt(verifyErr)
+			} else {
+				dataErrsByPart[p][i] = verifyResp.Results[p]
 			}
-		case madmin.HealNormalScan:
-			if !meta.Deleted && !meta.IsRemote() {
-				dataErrs[i] = onlineDisk.CheckParts(ctx, bucket, object, meta)
 		}
 	}
 
-		if dataErrs[i] == nil {
+	// Build dataErrs by disk from dataErrs by part
+	for part, disks := range dataErrsByPart {
+		for disk := range disks {
+			dataErrsByDisk[disk][part] = dataErrsByPart[part][disk]
+		}
+	}
+
+	for i, onlineDisk := range onlineDisks {
+		if metaErrs[i] == nil && !hasPartErr(dataErrsByDisk[i]) {
 			// All parts verified, mark it as all data available.
 			availableDisks[i] = onlineDisk
 		} else {
@@ -373,5 +445,5 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 		}
 	}
 
-	return availableDisks, dataErrs, timeSentinel
+	return
 }

cmd/erasure-healing-common_test.go

@@ -308,9 +308,8 @@ func TestListOnlineDisks(t *testing.T) {
 				t.Fatalf("Expected modTime to be equal to %v but was found to be %v",
 					test.expectedTime, modTime)
 			}
-			availableDisks, newErrs, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
+			availableDisks, _, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
 				test.errs, fi, bucket, object, madmin.HealDeepScan)
-			test.errs = newErrs
 
 			if test._tamperBackend != noTamper {
 				if tamperedIndex != -1 && availableDisks[tamperedIndex] != nil {
@@ -491,9 +490,8 @@ func TestListOnlineDisksSmallObjects(t *testing.T) {
 					test.expectedTime, modTime)
 			}
 
-			availableDisks, newErrs, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
+			availableDisks, _, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
 				test.errs, fi, bucket, object, madmin.HealDeepScan)
-			test.errs = newErrs
 
 			if test._tamperBackend != noTamper {
 				if tamperedIndex != -1 && availableDisks[tamperedIndex] != nil {
@@ -554,7 +552,7 @@ func TestDisksWithAllParts(t *testing.T) {
 
 	erasureDisks, _, _ = listOnlineDisks(erasureDisks, partsMetadata, errs, readQuorum)
 
-	filteredDisks, errs, _ := disksWithAllParts(ctx, erasureDisks, partsMetadata,
+	filteredDisks, _, dataErrsPerDisk := disksWithAllParts(ctx, erasureDisks, partsMetadata,
 		errs, fi, bucket, object, madmin.HealDeepScan)
 
 	if len(filteredDisks) != len(erasureDisks) {
@@ -562,8 +560,8 @@ func TestDisksWithAllParts(t *testing.T) {
 	}
 
 	for diskIndex, disk := range filteredDisks {
-		if errs[diskIndex] != nil {
+		if partNeedsHealing(dataErrsPerDisk[diskIndex]) {
-			t.Errorf("Unexpected error %s", errs[diskIndex])
+			t.Errorf("Unexpected error: %v", dataErrsPerDisk[diskIndex])
 		}
 
 		if disk == nil {
@@ -634,7 +632,7 @@ func TestDisksWithAllParts(t *testing.T) {
 	}
 
 	errs = make([]error, len(erasureDisks))
-	filteredDisks, errs, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata,
+	filteredDisks, dataErrsPerDisk, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata,
 		errs, fi, bucket, object, madmin.HealDeepScan)
 
 	if len(filteredDisks) != len(erasureDisks) {
@@ -646,15 +644,15 @@ func TestDisksWithAllParts(t *testing.T) {
 			if disk != nil {
 				t.Errorf("Drive not filtered as expected, drive: %d", diskIndex)
 			}
-			if errs[diskIndex] == nil {
+			if !partNeedsHealing(dataErrsPerDisk[diskIndex]) {
-				t.Errorf("Expected error not received, driveIndex: %d", diskIndex)
+				t.Errorf("Disk expected to be healed, driveIndex: %d", diskIndex)
 			}
 		} else {
 			if disk == nil {
 				t.Errorf("Drive erroneously filtered, driveIndex: %d", diskIndex)
 			}
-			if errs[diskIndex] != nil {
+			if partNeedsHealing(dataErrsPerDisk[diskIndex]) {
-				t.Errorf("Unexpected error, %s, driveIndex: %d", errs[diskIndex], diskIndex)
+				t.Errorf("Disk not expected to be healed, driveIndex: %d", diskIndex)
 			}
 
 		}

cmd/erasure-healing.go

@@ -32,6 +32,7 @@ import (
 	"github.com/minio/minio/internal/grid"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/v3/sync/errgroup"
+	"golang.org/x/exp/slices"
 )
 
 //go:generate stringer -type=healingMetric -trimprefix=healingMetric $GOFILE
@@ -144,36 +145,41 @@ func listAllBuckets(ctx context.Context, storageDisks []StorageAPI, healBuckets
 	return reduceReadQuorumErrs(ctx, g.Wait(), bucketMetadataOpIgnoredErrs, readQuorum)
 }
 
+var errLegacyXLMeta = errors.New("legacy XL meta")
+
+var errOutdatedXLMeta = errors.New("outdated XL meta")
+
+var errPartMissingOrCorrupt = errors.New("part missing or corrupt")
+
 // Only heal on disks where we are sure that healing is needed. We can expand
 // this list as and when we figure out more errors can be added to this list safely.
-func shouldHealObjectOnDisk(erErr, dataErr error, meta FileInfo, latestMeta FileInfo) bool {
+func shouldHealObjectOnDisk(erErr error, partsErrs []int, meta FileInfo, latestMeta FileInfo) (bool, error) {
-	switch {
+	if errors.Is(erErr, errFileNotFound) || errors.Is(erErr, errFileVersionNotFound) || errors.Is(erErr, errFileCorrupt) {
-	case errors.Is(erErr, errFileNotFound) || errors.Is(erErr, errFileVersionNotFound):
+		return true, erErr
-		return true
-	case errors.Is(erErr, errFileCorrupt):
-		return true
 	}
 	if erErr == nil {
 		if meta.XLV1 {
 			// Legacy means heal always
 			// always check first.
-			return true
+			return true, errLegacyXLMeta
+		}
+		if !latestMeta.Equals(meta) {
+			return true, errOutdatedXLMeta
 		}
 		if !meta.Deleted && !meta.IsRemote() {
 			// If xl.meta was read fine but there may be problem with the part.N files.
-			if IsErr(dataErr, []error{
+			for _, partErr := range partsErrs {
-				errFileNotFound,
+				if slices.Contains([]int{
-				errFileVersionNotFound,
+					checkPartFileNotFound,
-				errFileCorrupt,
+					checkPartFileCorrupt,
-			}...) {
+				}, partErr) {
-				return true
+					return true, errPartMissingOrCorrupt
 				}
 			}
-		if !latestMeta.Equals(meta) {
-			return true
 		}
+		return false, nil
 	}
-	return false
+	return false, erErr
 }
 
 const (
@@ -332,7 +338,7 @@ func (er *erasureObjects) healObject(ctx context.Context, bucket string, object
 	// used here for reconstruction. This is done to ensure that
 	// we do not skip drives that have inconsistent metadata to be
 	// skipped from purging when they are stale.
-	availableDisks, dataErrs, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
+	availableDisks, dataErrsByDisk, dataErrsByPart := disksWithAllParts(ctx, onlineDisks, partsMetadata,
 		errs, latestMeta, bucket, object, scanMode)
 
 	var erasure Erasure
@@ -355,15 +361,20 @@ func (er *erasureObjects) healObject(ctx context.Context, bucket string, object
 	// to be healed.
 	outDatedDisks := make([]StorageAPI, len(storageDisks))
 	disksToHealCount := 0
-	for i, v := range availableDisks {
+	for i := range availableDisks {
+		yes, reason := shouldHealObjectOnDisk(errs[i], dataErrsByDisk[i], partsMetadata[i], latestMeta)
+		if yes {
+			outDatedDisks[i] = storageDisks[i]
+			disksToHealCount++
+		}
+
 		driveState := ""
 		switch {
-		case v != nil:
+		case reason == nil:
 			driveState = madmin.DriveStateOk
-		case errors.Is(errs[i], errDiskNotFound), errors.Is(dataErrs[i], errDiskNotFound):
+		case IsErr(reason, errDiskNotFound):
 			driveState = madmin.DriveStateOffline
-		case IsErr(errs[i], errFileNotFound, errFileVersionNotFound, errVolumeNotFound),
+		case IsErr(reason, errFileNotFound, errFileVersionNotFound, errVolumeNotFound, errPartMissingOrCorrupt, errOutdatedXLMeta, errLegacyXLMeta):
-			IsErr(dataErrs[i], errFileNotFound, errFileVersionNotFound, errVolumeNotFound):
 			driveState = madmin.DriveStateMissing
 		default:
 			// all remaining cases imply corrupt data/metadata
@@ -380,12 +391,6 @@ func (er *erasureObjects) healObject(ctx context.Context, bucket string, object
 			Endpoint: storageEndpoints[i].String(),
 			State:    driveState,
 		})
-
-		if shouldHealObjectOnDisk(errs[i], dataErrs[i], partsMetadata[i], latestMeta) {
-			outDatedDisks[i] = storageDisks[i]
-			disksToHealCount++
-			continue
-		}
 	}
 
 	if isAllNotFound(errs) {
@@ -412,7 +417,7 @@ func (er *erasureObjects) healObject(ctx context.Context, bucket string, object
 	if !latestMeta.XLV1 && !latestMeta.Deleted && disksToHealCount > latestMeta.Erasure.ParityBlocks {
 		// Allow for dangling deletes, on versions that have DataDir missing etc.
 		// this would end up restoring the correct readable versions.
-		m, err := er.deleteIfDangling(ctx, bucket, object, partsMetadata, errs, dataErrs, ObjectOptions{
+		m, err := er.deleteIfDangling(ctx, bucket, object, partsMetadata, errs, dataErrsByPart, ObjectOptions{
 			VersionID: versionID,
 		})
 		errs = make([]error, len(errs))
@@ -908,18 +913,7 @@ func isObjectDirDangling(errs []error) (ok bool) {
 	return found < notFound && found > 0
 }
 
-// Object is considered dangling/corrupted if and only
+func danglingMetaErrsCount(cerrs []error) (notFoundCount int, nonActionableCount int) {
-// if total disks - a combination of corrupted and missing
-// files is lesser than number of data blocks.
-func isObjectDangling(metaArr []FileInfo, errs []error, dataErrs []error) (validMeta FileInfo, ok bool) {
-	// We can consider an object data not reliable
-	// when xl.meta is not found in read quorum disks.
-	// or when xl.meta is not readable in read quorum disks.
-	danglingErrsCount := func(cerrs []error) (int, int) {
-		var (
-			notFoundCount      int
-			nonActionableCount int
-		)
 	for _, readErr := range cerrs {
 		if readErr == nil {
 			continue
@@ -932,11 +926,39 @@ func isObjectDangling(metaArr []FileInfo, errs []error, dataErrs []error) (valid
 			nonActionableCount++
 		}
 	}
-		return notFoundCount, nonActionableCount
+	return
 }
 
-	notFoundMetaErrs, nonActionableMetaErrs := danglingErrsCount(errs)
+func danglingPartErrsCount(results []int) (notFoundCount int, nonActionableCount int) {
-	notFoundPartsErrs, nonActionablePartsErrs := danglingErrsCount(dataErrs)
+	for _, partResult := range results {
+		switch partResult {
+		case checkPartSuccess:
+			continue
+		case checkPartFileNotFound:
+			notFoundCount++
+		default:
+			// All other errors are non-actionable
+			nonActionableCount++
+		}
+	}
+	return
+}
+
+// Object is considered dangling/corrupted if and only
+// if total disks - a combination of corrupted and missing
+// files is lesser than number of data blocks.
+func isObjectDangling(metaArr []FileInfo, errs []error, dataErrsByPart map[int][]int) (validMeta FileInfo, ok bool) {
+	// We can consider an object data not reliable
+	// when xl.meta is not found in read quorum disks.
+	// or when xl.meta is not readable in read quorum disks.
+	notFoundMetaErrs, nonActionableMetaErrs := danglingMetaErrsCount(errs)
+
+	notFoundPartsErrs, nonActionablePartsErrs := 0, 0
+	for _, dataErrs := range dataErrsByPart {
+		if nf, na := danglingPartErrsCount(dataErrs); nf > notFoundPartsErrs {
+			notFoundPartsErrs, nonActionablePartsErrs = nf, na
+		}
+	}
 
 	for _, m := range metaArr {
 		if m.IsValid() {
@@ -948,7 +970,7 @@ func isObjectDangling(metaArr []FileInfo, errs []error, dataErrs []error) (valid
 	if !validMeta.IsValid() {
 		// validMeta is invalid because all xl.meta is missing apparently
 		// we should figure out if dataDirs are also missing > dataBlocks.
-		dataBlocks := (len(dataErrs) + 1) / 2
+		dataBlocks := (len(metaArr) + 1) / 2
 		if notFoundPartsErrs > dataBlocks {
 			// Not using parity to ensure that we do not delete
 			// any valid content, if any is recoverable. But if
@@ -1065,6 +1087,7 @@ func healTrace(funcName healingMetric, startTime time.Time, bucket, object strin
 		if result != nil {
 			tr.Custom["version-id"] = result.VersionID
 			tr.Custom["disks"] = strconv.Itoa(result.DiskCount)
+			tr.Bytes = result.ObjectSize
 		}
 	}
 	if err != nil {

cmd/erasure-healing_test.go

@@ -49,7 +49,7 @@ func TestIsObjectDangling(t *testing.T) {
 		name             string
 		metaArr          []FileInfo
 		errs             []error
-		dataErrs         []error
+		dataErrs         map[int][]int
 		expectedMeta     FileInfo
 		expectedDangling bool
 	}{
@@ -165,11 +165,8 @@ func TestIsObjectDangling(t *testing.T) {
 				nil,
 				nil,
 			},
-			dataErrs: []error{
+			dataErrs: map[int][]int{
-				errFileCorrupt,
+				0: {checkPartFileCorrupt, checkPartFileNotFound, checkPartSuccess, checkPartFileCorrupt},
-				errFileNotFound,
-				nil,
-				errFileCorrupt,
 			},
 			expectedMeta:     fi,
 			expectedDangling: false,
@@ -188,11 +185,8 @@ func TestIsObjectDangling(t *testing.T) {
 				errFileNotFound,
 				nil,
 			},
-			dataErrs: []error{
+			dataErrs: map[int][]int{
-				errFileNotFound,
+				0: {checkPartFileNotFound, checkPartFileCorrupt, checkPartSuccess, checkPartSuccess},
-				errFileCorrupt,
-				nil,
-				nil,
 			},
 			expectedMeta:     fi,
 			expectedDangling: false,
@@ -247,15 +241,58 @@ func TestIsObjectDangling(t *testing.T) {
 				nil,
 				nil,
 			},
-			dataErrs: []error{
+			dataErrs: map[int][]int{
-				errFileNotFound,
+				0: {checkPartFileNotFound, checkPartFileNotFound, checkPartSuccess, checkPartFileNotFound},
-				errFileNotFound,
-				nil,
-				errFileNotFound,
 			},
 			expectedMeta:     fi,
 			expectedDangling: true,
 		},
+		{
+			name: "FileInfoDecided-case4-(missing data-dir for part 2)",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileNotFound,
+				nil,
+				nil,
+			},
+			dataErrs: map[int][]int{
+				0: {checkPartSuccess, checkPartSuccess, checkPartSuccess, checkPartSuccess},
+				1: {checkPartSuccess, checkPartFileNotFound, checkPartFileNotFound, checkPartFileNotFound},
+			},
+			expectedMeta:     fi,
+			expectedDangling: true,
+		},
+
+		{
+			name: "FileInfoDecided-case4-(enough data-dir existing for each part)",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileNotFound,
+				nil,
+				nil,
+			},
+			dataErrs: map[int][]int{
+				0: {checkPartFileNotFound, checkPartSuccess, checkPartSuccess, checkPartSuccess},
+				1: {checkPartSuccess, checkPartFileNotFound, checkPartSuccess, checkPartSuccess},
+				2: {checkPartSuccess, checkPartSuccess, checkPartFileNotFound, checkPartSuccess},
+				3: {checkPartSuccess, checkPartSuccess, checkPartSuccess, checkPartFileNotFound},
+			},
+			expectedMeta:     fi,
+			expectedDangling: false,
+		},
+
 		// Add new cases as seen
 	}
 	for _, testCase := range testCases {

cmd/erasure-multipart.go

@@ -483,6 +483,11 @@ func (er erasureObjects) newMultipartUpload(ctx context.Context, bucket string,
 	}
 	fi.DataDir = mustGetUUID()
 
+	if userDefined[ReplicationSsecChecksumHeader] != "" {
+		fi.Checksum, _ = base64.StdEncoding.DecodeString(userDefined[ReplicationSsecChecksumHeader])
+		delete(userDefined, ReplicationSsecChecksumHeader)
+	}
+
 	// Initialize erasure metadata.
 	for index := range partsMetadata {
 		partsMetadata[index] = fi
@@ -1294,9 +1299,18 @@ func (er erasureObjects) CompleteMultipartUpload(ctx context.Context, bucket str
 		defer lk.Unlock(lkctx)
 	}
 
+	// Accept encrypted checksum from incoming request.
+	if opts.UserDefined[ReplicationSsecChecksumHeader] != "" {
+		if v, err := base64.StdEncoding.DecodeString(opts.UserDefined[ReplicationSsecChecksumHeader]); err == nil {
+			fi.Checksum = v
+		}
+		delete(opts.UserDefined, ReplicationSsecChecksumHeader)
+	}
+
 	if checksumType.IsSet() {
 		checksumType |= hash.ChecksumMultipart | hash.ChecksumIncludesMultipart
-		cs := hash.NewChecksumFromData(checksumType, checksumCombined)
+		var cs *hash.Checksum
+		cs = hash.NewChecksumFromData(checksumType, checksumCombined)
 		fi.Checksum = cs.AppendTo(nil, checksumCombined)
 		if opts.EncryptFn != nil {
 			fi.Checksum = opts.EncryptFn("object-checksum", fi.Checksum)

cmd/erasure-object.go

@@ -20,6 +20,7 @@ package cmd
 import (
 	"bytes"
 	"context"
+	"encoding/base64"
 	"errors"
 	"fmt"
 	"io"
@@ -276,7 +277,7 @@ func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object stri
 		return gr.WithCleanupFuncs(nsUnlocker), nil
 	}
 
-	fn, off, length, err := NewGetObjectReader(rs, objInfo, opts)
+	fn, off, length, err := NewGetObjectReader(rs, objInfo, opts, h)
 	if err != nil {
 		return nil, err
 	}
@@ -483,8 +484,8 @@ func joinErrs(errs []error) []string {
 	return s
 }
 
-func (er erasureObjects) deleteIfDangling(ctx context.Context, bucket, object string, metaArr []FileInfo, errs []error, dataErrs []error, opts ObjectOptions) (FileInfo, error) {
+func (er erasureObjects) deleteIfDangling(ctx context.Context, bucket, object string, metaArr []FileInfo, errs []error, dataErrsByPart map[int][]int, opts ObjectOptions) (FileInfo, error) {
-	m, ok := isObjectDangling(metaArr, errs, dataErrs)
+	m, ok := isObjectDangling(metaArr, errs, dataErrsByPart)
 	if !ok {
 		// We only come here if we cannot figure out if the object
 		// can be deleted safely, in such a scenario return ReadQuorum error.
@@ -494,7 +495,7 @@ func (er erasureObjects) deleteIfDangling(ctx context.Context, bucket, object st
 	tags["set"] = er.setIndex
 	tags["pool"] = er.poolIndex
 	tags["merrs"] = joinErrs(errs)
-	tags["derrs"] = joinErrs(dataErrs)
+	tags["derrs"] = dataErrsByPart
 	if m.IsValid() {
 		tags["size"] = m.Size
 		tags["mtime"] = m.ModTime.Format(http.TimeFormat)
@@ -508,8 +509,20 @@ func (er erasureObjects) deleteIfDangling(ctx context.Context, bucket, object st
 
 	// count the number of offline disks
 	offline := 0
-	for i := 0; i < max(len(errs), len(dataErrs)); i++ {
+	for i := 0; i < len(errs); i++ {
-		if i < len(errs) && errors.Is(errs[i], errDiskNotFound) || i < len(dataErrs) && errors.Is(dataErrs[i], errDiskNotFound) {
+		var found bool
+		switch {
+		case errors.Is(errs[i], errDiskNotFound):
+			found = true
+		default:
+			for p := range dataErrsByPart {
+				if dataErrsByPart[p][i] == checkPartDiskNotFound {
+					found = true
+					break
+				}
+			}
+		}
+		if found {
 			offline++
 		}
 	}
@@ -563,35 +576,11 @@ func (er erasureObjects) deleteIfDangling(ctx context.Context, bucket, object st
 }
 
 func fileInfoFromRaw(ri RawFileInfo, bucket, object string, readData, inclFreeVers, allParts bool) (FileInfo, error) {
-	var xl xlMetaV2
+	return getFileInfo(ri.Buf, bucket, object, "", fileInfoOpts{
-	if err := xl.LoadOrConvert(ri.Buf); err != nil {
+		Data:             readData,
-		return FileInfo{}, errFileCorrupt
+		InclFreeVersions: inclFreeVers,
-	}
+		AllParts:         allParts,
-
+	})
-	fi, err := xl.ToFileInfo(bucket, object, "", inclFreeVers, allParts)
-	if err != nil {
-		return FileInfo{}, err
-	}
-
-	if !fi.IsValid() {
-		return FileInfo{}, errFileCorrupt
-	}
-
-	versionID := fi.VersionID
-	if versionID == "" {
-		versionID = nullVersionID
-	}
-
-	fileInfo, err := xl.ToFileInfo(bucket, object, versionID, inclFreeVers, allParts)
-	if err != nil {
-		return FileInfo{}, err
-	}
-
-	if readData {
-		fileInfo.Data = xl.data.find(versionID)
-	}
-
-	return fileInfo, nil
 }
 
 func readAllRawFileInfo(ctx context.Context, disks []StorageAPI, bucket, object string, readData bool) ([]RawFileInfo, []error) {
@@ -744,6 +733,7 @@ func (er erasureObjects) getObjectFileInfo(ctx context.Context, bucket, object s
 
 	ropts := ReadOptions{
 		ReadData:         readData,
+		InclFreeVersions: opts.InclFreeVersions,
 		Healing:          false,
 	}
 
@@ -1355,6 +1345,12 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 	if opts.EncryptFn != nil {
 		fi.Checksum = opts.EncryptFn("object-checksum", fi.Checksum)
 	}
+	if userDefined[ReplicationSsecChecksumHeader] != "" {
+		if v, err := base64.StdEncoding.DecodeString(userDefined[ReplicationSsecChecksumHeader]); err == nil {
+			fi.Checksum = v
+		}
+	}
+	delete(userDefined, ReplicationSsecChecksumHeader)
 	uniqueID := mustGetUUID()
 	tempObj := uniqueID
 

cmd/erasure-server-pool-decom.go

@@ -836,7 +836,7 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 				if filterLifecycle(bi.Name, version.Name, version) {
 					expired++
 					decommissioned++
-					stopFn(errors.New("ILM expired object/version will be skipped"))
+					stopFn(version.Size, errors.New("ILM expired object/version will be skipped"))
 					continue
 				}
 
@@ -846,7 +846,7 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 				remainingVersions := len(fivs.Versions) - expired
 				if version.Deleted && remainingVersions == 1 {
 					decommissioned++
-					stopFn(errors.New("DELETE marked object with no other non-current versions will be skipped"))
+					stopFn(version.Size, errors.New("DELETE marked object with no other non-current versions will be skipped"))
 					continue
 				}
 
@@ -877,7 +877,7 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 							err = nil
 						}
 					}
-					stopFn(err)
+					stopFn(version.Size, err)
 					if err != nil {
 						decomLogIf(ctx, err)
 						failure = true
@@ -902,12 +902,12 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 							MTime:       version.ModTime,
 							UserDefined: version.Metadata,
 						}); err != nil {
-							stopFn(err)
+							stopFn(version.Size, err)
 							failure = true
 							decomLogIf(ctx, err)
 							continue
 						}
-						stopFn(nil)
+						stopFn(version.Size, nil)
 						failure = false
 						break
 					}
@@ -925,14 +925,14 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 					if isErrObjectNotFound(err) || isErrVersionNotFound(err) {
 						// object deleted by the application, nothing to do here we move on.
 						ignore = true
-						stopFn(nil)
+						stopFn(version.Size, nil)
 						break
 					}
 					if err != nil && !ignore {
 						// if usage-cache.bin is not readable log and ignore it.
 						if bi.Name == minioMetaBucket && strings.Contains(version.Name, dataUsageCacheName) {
 							ignore = true
-							stopFn(err)
+							stopFn(version.Size, err)
 							decomLogIf(ctx, err)
 							break
 						}
@@ -940,16 +940,16 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 					if err != nil {
 						failure = true
 						decomLogIf(ctx, err)
-						stopFn(err)
+						stopFn(version.Size, err)
 						continue
 					}
 					if err = z.decommissionObject(ctx, bi.Name, gr); err != nil {
-						stopFn(err)
+						stopFn(version.Size, err)
 						failure = true
 						decomLogIf(ctx, err)
 						continue
 					}
-					stopFn(nil)
+					stopFn(version.Size, nil)
 					failure = false
 					break
 				}
@@ -977,7 +977,7 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 						NoAuditLog:         true,
 					},
 				)
-				stopFn(err)
+				stopFn(0, err)
 				auditLogDecom(ctx, "DecomDeleteObject", bi.Name, entry.name, "", err)
 				if err != nil {
 					decomLogIf(ctx, err)
@@ -1038,7 +1038,7 @@ const (
 	decomMetricDecommissionRemoveObject
 )
 
-func decomTrace(d decomMetric, poolIdx int, startTime time.Time, duration time.Duration, path string, err error) madmin.TraceInfo {
+func decomTrace(d decomMetric, poolIdx int, startTime time.Time, duration time.Duration, path string, err error, sz int64) madmin.TraceInfo {
 	var errStr string
 	if err != nil {
 		errStr = err.Error()
@@ -1051,15 +1051,16 @@ func decomTrace(d decomMetric, poolIdx int, startTime time.Time, duration time.D
 		Duration:  duration,
 		Path:      path,
 		Error:     errStr,
+		Bytes:     sz,
 	}
 }
 
-func (m *decomMetrics) log(d decomMetric, poolIdx int, paths ...string) func(err error) {
+func (m *decomMetrics) log(d decomMetric, poolIdx int, paths ...string) func(z int64, err error) {
 	startTime := time.Now()
-	return func(err error) {
+	return func(sz int64, err error) {
 		duration := time.Since(startTime)
 		if globalTrace.NumSubscribers(madmin.TraceDecommission) > 0 {
-			globalTrace.Publish(decomTrace(d, poolIdx, startTime, duration, strings.Join(paths, " "), err))
+			globalTrace.Publish(decomTrace(d, poolIdx, startTime, duration, strings.Join(paths, " "), err, sz))
 		}
 	}
 }
@@ -1092,10 +1093,10 @@ func (z *erasureServerPools) decommissionInBackground(ctx context.Context, idx i
 		}
 		stopFn := globalDecommissionMetrics.log(decomMetricDecommissionBucket, idx, bucket.Name)
 		if err := z.decommissionPool(ctx, idx, pool, bucket); err != nil {
-			stopFn(err)
+			stopFn(0, err)
 			return err
 		}
-		stopFn(nil)
+		stopFn(0, nil)
 
 		z.poolMetaMutex.Lock()
 		if z.poolMeta.BucketDone(idx, bucket) {
@@ -1326,11 +1327,7 @@ func (z *erasureServerPools) Status(ctx context.Context, idx int) (PoolStatus, e
 	poolInfo := z.poolMeta.Pools[idx].Clone()
 	if poolInfo.Decommission != nil {
 		poolInfo.Decommission.TotalSize = pi.Total
-		if poolInfo.Decommission.Failed || poolInfo.Decommission.Canceled {
 		poolInfo.Decommission.CurrentSize = pi.Free
-		} else {
-			poolInfo.Decommission.CurrentSize = poolInfo.Decommission.StartSize + poolInfo.Decommission.BytesDone
-		}
 	} else {
 		poolInfo.Decommission = &PoolDecommissionInfo{
 			TotalSize:   pi.Total,

cmd/erasure-server-pool-rebalance.go

@@ -427,7 +427,7 @@ func (z *erasureServerPools) rebalanceBuckets(ctx context.Context, poolIdx int)
 
 			stopFn := globalRebalanceMetrics.log(rebalanceMetricSaveMetadata, poolIdx, traceMsg)
 			err := z.saveRebalanceStats(GlobalContext, poolIdx, rebalSaveStats)
-			stopFn(err)
+			stopFn(0, err)
 			rebalanceLogIf(GlobalContext, err)
 
 			if quit {
@@ -456,7 +456,7 @@ func (z *erasureServerPools) rebalanceBuckets(ctx context.Context, poolIdx int)
 
 		stopFn := globalRebalanceMetrics.log(rebalanceMetricRebalanceBucket, poolIdx, bucket)
 		if err = z.rebalanceBucket(ctx, bucket, poolIdx); err != nil {
-			stopFn(err)
+			stopFn(0, err)
 			if errors.Is(err, errServerNotInitialized) || errors.Is(err, errBucketMetadataNotInitialized) {
 				continue
 			}
@@ -464,7 +464,7 @@ func (z *erasureServerPools) rebalanceBuckets(ctx context.Context, poolIdx int)
 			doneCh <- err
 			return
 		}
-		stopFn(nil)
+		stopFn(0, nil)
 		z.bucketRebalanceDone(bucket, poolIdx)
 	}
 
@@ -692,24 +692,24 @@ func (z *erasureServerPools) rebalanceBucket(ctx context.Context, bucket string,
 					if isErrObjectNotFound(err) || isErrVersionNotFound(err) {
 						// object deleted by the application, nothing to do here we move on.
 						ignore = true
-						stopFn(nil)
+						stopFn(0, nil)
 						break
 					}
 					if err != nil {
 						failure = true
 						rebalanceLogIf(ctx, err)
-						stopFn(err)
+						stopFn(0, err)
 						continue
 					}
 
 					if err = z.rebalanceObject(ctx, bucket, gr); err != nil {
 						failure = true
 						rebalanceLogIf(ctx, err)
-						stopFn(err)
+						stopFn(version.Size, err)
 						continue
 					}
 
-					stopFn(nil)
+					stopFn(version.Size, nil)
 					failure = false
 					break
 				}
@@ -735,7 +735,7 @@ func (z *erasureServerPools) rebalanceBucket(ctx context.Context, bucket string,
 						NoAuditLog:         true,
 					},
 				)
-				stopFn(err)
+				stopFn(0, err)
 				auditLogRebalance(ctx, "Rebalance:DeleteObject", bucket, entry.name, "", err)
 				if err != nil {
 					rebalanceLogIf(ctx, err)
@@ -935,7 +935,7 @@ func (z *erasureServerPools) StartRebalance() {
 		go func(idx int) {
 			stopfn := globalRebalanceMetrics.log(rebalanceMetricRebalanceBuckets, idx)
 			err := z.rebalanceBuckets(ctx, idx)
-			stopfn(err)
+			stopfn(0, err)
 		}(poolIdx)
 	}
 }
@@ -975,7 +975,7 @@ const (
 	rebalanceMetricSaveMetadata
 )
 
-func rebalanceTrace(r rebalanceMetric, poolIdx int, startTime time.Time, duration time.Duration, err error, path string) madmin.TraceInfo {
+func rebalanceTrace(r rebalanceMetric, poolIdx int, startTime time.Time, duration time.Duration, err error, path string, sz int64) madmin.TraceInfo {
 	var errStr string
 	if err != nil {
 		errStr = err.Error()
@@ -988,15 +988,16 @@ func rebalanceTrace(r rebalanceMetric, poolIdx int, startTime time.Time, duratio
 		Duration:  duration,
 		Path:      path,
 		Error:     errStr,
+		Bytes:     sz,
 	}
 }
 
-func (p *rebalanceMetrics) log(r rebalanceMetric, poolIdx int, paths ...string) func(err error) {
+func (p *rebalanceMetrics) log(r rebalanceMetric, poolIdx int, paths ...string) func(sz int64, err error) {
 	startTime := time.Now()
-	return func(err error) {
+	return func(sz int64, err error) {
 		duration := time.Since(startTime)
 		if globalTrace.NumSubscribers(madmin.TraceRebalance) > 0 {
-			globalTrace.Publish(rebalanceTrace(r, poolIdx, startTime, duration, err, strings.Join(paths, " ")))
+			globalTrace.Publish(rebalanceTrace(r, poolIdx, startTime, duration, err, strings.Join(paths, " "), sz))
 		}
 	}
 }

cmd/erasure-server-pool.go

@@ -104,7 +104,11 @@ func newErasureServerPools(ctx context.Context, endpointServerPools EndpointServ
 	// Initialize byte pool once for all sets, bpool size is set to
 	// setCount * setDriveCount with each memory upto blockSizeV2.
 	buffers := bpool.NewBytePoolCap(n, blockSizeV2, blockSizeV2*2)
+	if n >= 16384 {
+		// pre-populate buffers only n >= 16384 which is (32Gi/2Mi)
+		// for all setups smaller than this avoid pre-alloc.
 		buffers.Populate()
+	}
 	globalBytePoolCap.Store(buffers)
 
 	var localDrives []StorageAPI
@@ -165,6 +169,9 @@ func newErasureServerPools(ctx context.Context, endpointServerPools EndpointServ
 	if !globalIsDistErasure {
 		globalLocalDrivesMu.Lock()
 		globalLocalDrives = localDrives
+		for _, drive := range localDrives {
+			globalLocalDrivesMap[drive.Endpoint().String()] = drive
+		}
 		globalLocalDrivesMu.Unlock()
 	}
 
@@ -1535,7 +1542,7 @@ func (z *erasureServerPools) listObjectsGeneric(ctx context.Context, bucket, pre
 		return loi, nil
 	}
 	ri := logger.GetReqInfo(ctx)
-	hadoop := ri != nil && strings.Contains(ri.UserAgent, `Hadoop `) && strings.Contains(ri.UserAgent, "scala/")
+	hadoop := ri != nil && strings.Contains(ri.UserAgent, "Hadoop ") && strings.Contains(ri.UserAgent, "scala/")
 	matches := func() bool {
 		if prefix == "" {
 			return false
@@ -1600,7 +1607,7 @@ func (z *erasureServerPools) listObjectsGeneric(ctx context.Context, bucket, pre
 		// }
 		if matches() {
 			objInfo, err := z.GetObjectInfo(ctx, bucket, path.Dir(prefix), ObjectOptions{NoLock: true})
-			if err == nil {
+			if err == nil || objInfo.IsLatest && objInfo.DeleteMarker {
 				if opts.Lifecycle != nil {
 					evt := evalActionFromLifecycle(ctx, *opts.Lifecycle, opts.Retention, opts.Replication.Config, objInfo)
 					if evt.Action.Delete() {
@@ -1626,41 +1633,29 @@ func (z *erasureServerPools) listObjectsGeneric(ctx context.Context, bucket, pre
 		// and throw a 404 exception. This is especially a problem for spark jobs overwriting the same partition
 		// repeatedly. This workaround recursively lists the top 3 entries including delete markers to reflect the
 		// correct state of the directory in the list results.
-		opts.Recursive = true
+		if strings.HasSuffix(opts.Prefix, SlashSeparator) {
-		opts.InclDeleted = true
+			li, err := listFn(ctx, opts, maxKeys)
-		opts.Limit = maxKeys + 1
+			if err != nil {
-		li, err := listFn(ctx, opts, opts.Limit)
+				return loi, err
-		if err == nil {
+			}
-			switch {
+			if len(li.Objects) == 0 {
-			case len(li.Objects) == 0 && len(li.Prefixes) == 0:
+				prefixes := li.Prefixes[:0]
+				for _, prefix := range li.Prefixes {
+					objInfo, _ := z.GetObjectInfo(ctx, bucket, pathJoin(prefix, "_SUCCESS"), ObjectOptions{NoLock: true})
+					if objInfo.IsLatest && objInfo.DeleteMarker {
+						continue
+					}
+					prefixes = append(prefixes, prefix)
+				}
+				if len(prefixes) > 0 {
+					objInfo, _ := z.GetObjectInfo(ctx, bucket, pathJoin(opts.Prefix, "_SUCCESS"), ObjectOptions{NoLock: true})
+					if objInfo.IsLatest && objInfo.DeleteMarker {
 						return loi, nil
-			case len(li.Objects) > 0 || len(li.Prefixes) > 0:
-				var o ObjectInfo
-				var pfx string
-				if len(li.Objects) > 0 {
-					o = li.Objects[0]
-					p := strings.TrimPrefix(o.Name, opts.Prefix)
-					if p != "" {
-						sidx := strings.Index(p, "/")
-						if sidx > 0 {
-							pfx = p[:sidx]
 					}
 				}
+				li.Prefixes = prefixes
 			}
-				switch {
+			return li, nil
-				case o.DeleteMarker:
-					loi.Objects = append(loi.Objects, ObjectInfo{Bucket: bucket, IsDir: true, Name: prefix})
-					return loi, nil
-				case len(li.Objects) >= 1:
-					loi.Objects = append(loi.Objects, o)
-					if pfx != "" {
-						loi.Prefixes = append(loi.Prefixes, path.Join(opts.Prefix, pfx))
-					}
-				case len(li.Prefixes) > 0:
-					loi.Prefixes = append(loi.Prefixes, li.Prefixes...)
-				}
-			}
-			return loi, nil
 		}
 	}
 
@@ -2013,12 +2008,14 @@ func (z *erasureServerPools) ListBuckets(ctx context.Context, opts BucketOptions
 				if err != nil {
 					return nil, err
 				}
+				if !opts.NoMetadata {
 					for i := range buckets {
 						createdAt, err := globalBucketMetadataSys.CreatedAt(buckets[i].Name)
 						if err == nil {
 							buckets[i].Created = createdAt
 						}
 					}
+				}
 				return buckets, nil
 			},
 		)
@@ -2030,12 +2027,15 @@ func (z *erasureServerPools) ListBuckets(ctx context.Context, opts BucketOptions
 	if err != nil {
 		return nil, err
 	}
+
+	if !opts.NoMetadata {
 		for i := range buckets {
 			createdAt, err := globalBucketMetadataSys.CreatedAt(buckets[i].Name)
 			if err == nil {
 				buckets[i].Created = createdAt
 			}
 		}
+	}
 	return buckets, nil
 }
 
@@ -2087,7 +2087,6 @@ func (z *erasureServerPools) HealBucket(ctx context.Context, bucket string, opts
 
 // Walk a bucket, optionally prefix recursively, until we have returned
 // all the contents of the provided bucket+prefix.
-// TODO: Note that most errors will result in a truncated listing.
 func (z *erasureServerPools) Walk(ctx context.Context, bucket, prefix string, results chan<- itemOrErr[ObjectInfo], opts WalkOptions) error {
 	if err := checkListObjsArgs(ctx, bucket, prefix, ""); err != nil {
 		xioutil.SafeClose(results)
@@ -2204,9 +2203,8 @@ func (z *erasureServerPools) Walk(ctx context.Context, bucket, prefix string, re
 	// Convert and filter merged entries.
 	merged := make(chan metaCacheEntry, 100)
 	vcfg, _ := globalBucketVersioningSys.Get(bucket)
+	errCh := make(chan error, 1)
 	go func() {
-		defer cancelCause(nil)
-		defer xioutil.SafeClose(results)
 		sentErr := false
 		sendErr := func(err error) {
 			if !sentErr {
@@ -2217,6 +2215,15 @@ func (z *erasureServerPools) Walk(ctx context.Context, bucket, prefix string, re
 				}
 			}
 		}
+		defer func() {
+			select {
+			case <-ctx.Done():
+				sendErr(ctx.Err())
+			default:
+			}
+			xioutil.SafeClose(results)
+			cancelCause(nil)
+		}()
 		send := func(oi ObjectInfo) bool {
 			select {
 			case results <- itemOrErr[ObjectInfo]{Item: oi}:
@@ -2271,12 +2278,16 @@ func (z *erasureServerPools) Walk(ctx context.Context, bucket, prefix string, re
 				}
 			}
 		}
+		if err := <-errCh; err != nil {
+			sendErr(err)
+		}
 	}()
 	go func() {
+		defer close(errCh)
 		// Merge all entries from all disks.
 		// We leave quorum at 1, since entries are already resolved to have the desired quorum.
 		// mergeEntryChannels will close 'merged' channel upon completion or cancellation.
-		storageLogIf(ctx, mergeEntryChannels(ctx, entries, merged, 1))
+		errCh <- mergeEntryChannels(ctx, entries, merged, 1)
 	}()
 
 	return nil
@@ -2454,6 +2465,24 @@ type HealthResult struct {
 	UsingDefaults bool
 }
 
+func (hr HealthResult) String() string {
+	var str strings.Builder
+	for i, es := range hr.ESHealth {
+		str.WriteString("(Pool: ")
+		str.WriteString(strconv.Itoa(es.PoolID))
+		str.WriteString(" Set: ")
+		str.WriteString(strconv.Itoa(es.SetID))
+		str.WriteString(" Healthy: ")
+		str.WriteString(strconv.FormatBool(es.Healthy))
+		if i == 0 {
+			str.WriteString(")")
+		} else {
+			str.WriteString("), ")
+		}
+	}
+	return str.String()
+}
+
 // Health - returns current status of the object layer health,
 // provides if write access exists across sets, additionally
 // can be used to query scenarios if health may be lost
@@ -2573,16 +2602,16 @@ func (z *erasureServerPools) Health(ctx context.Context, opts HealthOptions) Hea
 			healthy := erasureSetUpCount[poolIdx][setIdx].online >= poolWriteQuorums[poolIdx]
 			if !healthy {
 				storageLogIf(logger.SetReqInfo(ctx, reqInfo),
-					fmt.Errorf("Write quorum may be lost on pool: %d, set: %d, expected write quorum: %d",
+					fmt.Errorf("Write quorum could not be established on pool: %d, set: %d, expected write quorum: %d, drives-online: %d",
-						poolIdx, setIdx, poolWriteQuorums[poolIdx]), logger.FatalKind)
+						poolIdx, setIdx, poolWriteQuorums[poolIdx], erasureSetUpCount[poolIdx][setIdx].online), logger.FatalKind)
 			}
 			result.Healthy = result.Healthy && healthy
 
 			healthyRead := erasureSetUpCount[poolIdx][setIdx].online >= poolReadQuorums[poolIdx]
 			if !healthyRead {
 				storageLogIf(logger.SetReqInfo(ctx, reqInfo),
-					fmt.Errorf("Read quorum may be lost on pool: %d, set: %d, expected read quorum: %d",
+					fmt.Errorf("Read quorum could not be established on pool: %d, set: %d, expected read quorum: %d, drives-online: %d",
-						poolIdx, setIdx, poolReadQuorums[poolIdx]))
+						poolIdx, setIdx, poolReadQuorums[poolIdx], erasureSetUpCount[poolIdx][setIdx].online))
 			}
 			result.HealthyRead = result.HealthyRead && healthyRead
 		}

cmd/erasure-sets.go

@@ -36,7 +36,6 @@ import (
 	"github.com/minio/minio-go/v7/pkg/set"
 	"github.com/minio/minio-go/v7/pkg/tags"
 	"github.com/minio/minio/internal/dsync"
-	xioutil "github.com/minio/minio/internal/ioutil"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/v3/console"
 	"github.com/minio/pkg/v3/sync/errgroup"
@@ -80,10 +79,6 @@ type erasureSets struct {
 
 	poolIndex int
 
-	// A channel to send the set index to the MRF when
-	// any disk belonging to that set is connected
-	setReconnectEvent chan int
-
 	// Distribution algorithm of choice.
 	distributionAlgo string
 	deploymentID     [16]byte
@@ -380,7 +375,6 @@ func newErasureSets(ctx context.Context, endpoints PoolEndpoints, storageDisks [
 		setDriveCount:      setDriveCount,
 		defaultParityCount: defaultParityCount,
 		format:             format,
-		setReconnectEvent:  make(chan int),
 		distributionAlgo:   format.Erasure.DistributionAlgo,
 		deploymentID:       uuid.MustParse(format.ID),
 		poolIndex:          poolIdx,
@@ -662,14 +656,6 @@ func (s *erasureSets) Shutdown(ctx context.Context) error {
 			return err
 		}
 	}
-	select {
-	case _, ok := <-s.setReconnectEvent:
-		if ok {
-			xioutil.SafeClose(s.setReconnectEvent)
-		}
-	default:
-		xioutil.SafeClose(s.setReconnectEvent)
-	}
 	return nil
 }
 

cmd/event-notification.go

@@ -49,23 +49,18 @@ func NewEventNotifier(ctx context.Context) *EventNotifier {
 }
 
 // GetARNList - returns available ARNs.
-func (evnot *EventNotifier) GetARNList(onlyActive bool) []string {
+func (evnot *EventNotifier) GetARNList() []string {
 	arns := []string{}
 	if evnot == nil {
 		return arns
 	}
 	region := globalSite.Region()
-	for targetID, target := range evnot.targetList.TargetMap() {
+	for targetID := range evnot.targetList.TargetMap() {
 		// httpclient target is part of ListenNotification
 		// which doesn't need to be listed as part of the ARN list
 		// This list is only meant for external targets, filter
 		// this out pro-actively.
 		if !strings.HasPrefix(targetID.ID, "httpclient+") {
-			if onlyActive {
-				if _, err := target.IsActive(); err != nil {
-					continue
-				}
-			}
 			arns = append(arns, targetID.ToARN(region).String())
 		}
 	}
@@ -74,7 +69,7 @@ func (evnot *EventNotifier) GetARNList(onlyActive bool) []string {
 }
 
 // Loads notification policies for all buckets into EventNotifier.
-func (evnot *EventNotifier) set(bucket BucketInfo, meta BucketMetadata) {
+func (evnot *EventNotifier) set(bucket string, meta BucketMetadata) {
 	config := meta.notificationConfig
 	if config == nil {
 		return
@@ -86,7 +81,7 @@ func (evnot *EventNotifier) set(bucket BucketInfo, meta BucketMetadata) {
 			internalLogIf(GlobalContext, err)
 		}
 	}
-	evnot.AddRulesMap(bucket.Name, config.ToRulesMap())
+	evnot.AddRulesMap(bucket, config.ToRulesMap())
 }
 
 // Targets returns all the registered targets

cmd/ftp-server-driver.go

@@ -105,7 +105,7 @@ type ftpMetrics struct{}
 
 var globalFtpMetrics ftpMetrics
 
-func ftpTrace(s *ftp.Context, startTime time.Time, source, objPath string, err error) madmin.TraceInfo {
+func ftpTrace(s *ftp.Context, startTime time.Time, source, objPath string, err error, sz int64) madmin.TraceInfo {
 	var errStr string
 	if err != nil {
 		errStr = err.Error()
@@ -114,25 +114,33 @@ func ftpTrace(s *ftp.Context, startTime time.Time, source, objPath string, err e
 		TraceType: madmin.TraceFTP,
 		Time:      startTime,
 		NodeName:  globalLocalNodeName,
-		FuncName:  fmt.Sprintf("ftp USER=%s COMMAND=%s PARAM=%s ISLOGIN=%t, Source=%s", s.Sess.LoginUser(), s.Cmd, s.Param, s.Sess.IsLogin(), source),
+		FuncName:  fmt.Sprintf(s.Cmd),
 		Duration:  time.Since(startTime),
 		Path:      objPath,
 		Error:     errStr,
+		Bytes:     sz,
+		Custom: map[string]string{
+			"user":   s.Sess.LoginUser(),
+			"cmd":    s.Cmd,
+			"param":  s.Param,
+			"login":  fmt.Sprintf("%t", s.Sess.IsLogin()),
+			"source": source,
+		},
 	}
 }
 
-func (m *ftpMetrics) log(s *ftp.Context, paths ...string) func(err error) {
+func (m *ftpMetrics) log(s *ftp.Context, paths ...string) func(sz int64, err error) {
 	startTime := time.Now()
 	source := getSource(2)
-	return func(err error) {
+	return func(sz int64, err error) {
-		globalTrace.Publish(ftpTrace(s, startTime, source, strings.Join(paths, " "), err))
+		globalTrace.Publish(ftpTrace(s, startTime, source, strings.Join(paths, " "), err, sz))
 	}
 }
 
 // Stat implements ftpDriver
 func (driver *ftpDriver) Stat(ctx *ftp.Context, objPath string) (fi os.FileInfo, err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	if objPath == SlashSeparator {
 		return &minioFileInfo{
@@ -190,7 +198,7 @@ func (driver *ftpDriver) Stat(ctx *ftp.Context, objPath string) (fi os.FileInfo,
 // ListDir implements ftpDriver
 func (driver *ftpDriver) ListDir(ctx *ftp.Context, objPath string, callback func(os.FileInfo) error) (err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	clnt, err := driver.getMinIOClient(ctx)
 	if err != nil {
@@ -252,7 +260,7 @@ func (driver *ftpDriver) ListDir(ctx *ftp.Context, objPath string, callback func
 
 func (driver *ftpDriver) CheckPasswd(c *ftp.Context, username, password string) (ok bool, err error) {
 	stopFn := globalFtpMetrics.log(c, username)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	if globalIAMSys.LDAPConfig.Enabled() {
 		sa, _, err := globalIAMSys.getServiceAccount(context.Background(), username)
@@ -376,7 +384,7 @@ func (driver *ftpDriver) getMinIOClient(ctx *ftp.Context) (*minio.Client, error)
 // DeleteDir implements ftpDriver
 func (driver *ftpDriver) DeleteDir(ctx *ftp.Context, objPath string) (err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	bucket, prefix := path2BucketObject(objPath)
 	if bucket == "" {
@@ -426,7 +434,7 @@ func (driver *ftpDriver) DeleteDir(ctx *ftp.Context, objPath string) (err error)
 // DeleteFile implements ftpDriver
 func (driver *ftpDriver) DeleteFile(ctx *ftp.Context, objPath string) (err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	bucket, object := path2BucketObject(objPath)
 	if bucket == "" {
@@ -444,7 +452,7 @@ func (driver *ftpDriver) DeleteFile(ctx *ftp.Context, objPath string) (err error
 // Rename implements ftpDriver
 func (driver *ftpDriver) Rename(ctx *ftp.Context, fromObjPath string, toObjPath string) (err error) {
 	stopFn := globalFtpMetrics.log(ctx, fromObjPath, toObjPath)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	return NotImplemented{}
 }
@@ -452,7 +460,7 @@ func (driver *ftpDriver) Rename(ctx *ftp.Context, fromObjPath string, toObjPath
 // MakeDir implements ftpDriver
 func (driver *ftpDriver) MakeDir(ctx *ftp.Context, objPath string) (err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	bucket, prefix := path2BucketObject(objPath)
 	if bucket == "" {
@@ -479,7 +487,7 @@ func (driver *ftpDriver) MakeDir(ctx *ftp.Context, objPath string) (err error) {
 // GetFile implements ftpDriver
 func (driver *ftpDriver) GetFile(ctx *ftp.Context, objPath string, offset int64) (n int64, rc io.ReadCloser, err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(n, err)
 
 	bucket, object := path2BucketObject(objPath)
 	if bucket == "" {
@@ -511,14 +519,14 @@ func (driver *ftpDriver) GetFile(ctx *ftp.Context, objPath string, offset int64)
 	if err != nil {
 		return 0, nil, err
 	}
-
+	n = info.Size - offset
-	return info.Size - offset, obj, nil
+	return n, obj, nil
 }
 
 // PutFile implements ftpDriver
 func (driver *ftpDriver) PutFile(ctx *ftp.Context, objPath string, data io.Reader, offset int64) (n int64, err error) {
 	stopFn := globalFtpMetrics.log(ctx, objPath)
-	defer stopFn(err)
+	defer stopFn(n, err)
 
 	bucket, object := path2BucketObject(objPath)
 	if bucket == "" {
@@ -539,5 +547,6 @@ func (driver *ftpDriver) PutFile(ctx *ftp.Context, objPath string, data io.Reade
 		ContentType:          mimedb.TypeByExtension(path.Ext(object)),
 		DisableContentSha256: true,
 	})
-	return info.Size, err
+	n = info.Size
+	return n, err
 }

cmd/generic-handlers.go

@@ -601,13 +601,6 @@ func setUploadForwardingMiddleware(h http.Handler) http.Handler {
 				h.ServeHTTP(w, r)
 				return
 			}
-			// forward request to peer handling this upload
-			if globalBucketTargetSys.isOffline(remote.EndpointURL) {
-				defer logger.AuditLog(r.Context(), w, r, mustGetClaimsFromToken(r))
-				writeErrorResponse(r.Context(), w, errorCodes.ToAPIErr(ErrReplicationRemoteConnectionError), r.URL)
-				return
-			}
-
 			r.URL.Scheme = remote.EndpointURL.Scheme
 			r.URL.Host = remote.EndpointURL.Host
 			// Make sure we remove any existing headers before

cmd/global-heal.go

@@ -163,7 +163,10 @@ func (er *erasureObjects) healErasureSet(ctx context.Context, buckets []string,
 	}
 
 	for _, bucket := range healBuckets {
-		_, err := objAPI.HealBucket(ctx, bucket, madmin.HealOpts{ScanMode: scanMode})
+		_, err := objAPI.HealBucket(ctx, bucket, madmin.HealOpts{
+			Recreate: true,
+			ScanMode: scanMode,
+		})
 		if err != nil {
 			// Log bucket healing error if any, we shall retry again.
 			healingLogIf(ctx, err)
@@ -262,6 +265,7 @@ func (er *erasureObjects) healErasureSet(ctx context.Context, buckets []string,
 		// Heal current bucket again in case if it is failed
 		// in the beginning of erasure set healing
 		if _, err := objAPI.HealBucket(ctx, bucket, madmin.HealOpts{
+			Recreate: true,
 			ScanMode: scanMode,
 		}); err != nil {
 			// Set this such that when we return this function
@@ -513,7 +517,7 @@ func (er *erasureObjects) healErasureSet(ctx context.Context, buckets []string,
 			// we let the caller retry this disk again for the
 			// buckets it failed to list.
 			retErr = err
-			healingLogIf(ctx, err)
+			healingLogIf(ctx, fmt.Errorf("listing failed with: %v on bucket: %v", err, bucket))
 			continue
 		}
 
@@ -530,9 +534,18 @@ func (er *erasureObjects) healErasureSet(ctx context.Context, buckets []string,
 	tracker.setObject("")
 	tracker.setBucket("")
 
+	if retErr != nil {
 		return retErr
 	}
 
+	// Last sanity check
+	if len(tracker.QueuedBuckets) > 0 {
+		return fmt.Errorf("not all buckets were healed: %v", tracker.QueuedBuckets)
+	}
+
+	return nil
+}
+
 func healBucket(bucket string, scan madmin.HealScanMode) error {
 	// Get background heal sequence to send elements to heal
 	bgSeq, ok := globalBackgroundHealState.getHealSequenceByToken(bgHealingUUID)

cmd/globals.go

@@ -160,8 +160,9 @@ type serverCtxt struct {
 	FTP  []string
 	SFTP []string
 
+	MemLimit uint64
+
 	UserTimeout         time.Duration
-	ShutdownTimeout     time.Duration
 	IdleTimeout         time.Duration
 	ReadHeaderTimeout   time.Duration
 	MaxIdleConnsPerHost int

cmd/handler-api.go

@@ -91,11 +91,11 @@ func availableMemory() (available uint64) {
 	available = 2048 * blockSizeV2 * 2 // Default to 4 GiB when we can't find the limits.
 
 	if runtime.GOOS == "linux" {
-		// Useful in container mode
+		// Honor cgroup limits if set.
 		limit := cgroupMemLimit()
 		if limit > 0 {
-			// A valid value is found, return its 75%
+			// A valid value is found, return its 90%
-			available = (limit * 3) / 4
+			available = (limit * 9) / 10
 			return
 		}
 	} // for all other platforms limits are based on virtual memory.
@@ -104,8 +104,9 @@ func availableMemory() (available uint64) {
 	if err != nil {
 		return
 	}
-	// A valid value is available return its 75%
+
-	available = (memStats.Available * 3) / 4
+	// A valid value is available return its 90%
+	available = (memStats.Available * 9) / 10
 	return
 }
 
@@ -129,7 +130,7 @@ func (t *apiConfig) init(cfg api.Config, setDriveCounts []int, legacy bool) {
 		maxSetDrives := slices.Max(setDriveCounts)
 
 		// Returns 75% of max memory allowed
-		maxMem := availableMemory()
+		maxMem := globalServerCtxt.MemLimit
 
 		// max requests per node is calculated as
 		// total_ram / ram_per_request
@@ -322,10 +323,9 @@ func maxClients(f http.HandlerFunc) http.HandlerFunc {
 		}
 
 		globalHTTPStats.addRequestsInQueue(1)
-		defer globalHTTPStats.addRequestsInQueue(-1)
-
 		pool, deadline := globalAPIConfig.getRequestsPool()
 		if pool == nil {
+			globalHTTPStats.addRequestsInQueue(-1)
 			f.ServeHTTP(w, r)
 			return
 		}
@@ -333,6 +333,7 @@ func maxClients(f http.HandlerFunc) http.HandlerFunc {
 		// No deadline to wait, there is nothing to queue
 		// perform the API call immediately.
 		if deadline <= 0 {
+			globalHTTPStats.addRequestsInQueue(-1)
 			f.ServeHTTP(w, r)
 			return
 		}
@@ -348,12 +349,14 @@ func maxClients(f http.HandlerFunc) http.HandlerFunc {
 		select {
 		case pool <- struct{}{}:
 			defer func() { <-pool }()
+			globalHTTPStats.addRequestsInQueue(-1)
 			if contextCanceled(ctx) {
 				w.WriteHeader(499)
 				return
 			}
 			f.ServeHTTP(w, r)
 		case <-deadlineTimer.C:
+			globalHTTPStats.addRequestsInQueue(-1)
 			if contextCanceled(ctx) {
 				w.WriteHeader(499)
 				return
@@ -363,6 +366,7 @@ func maxClients(f http.HandlerFunc) http.HandlerFunc {
 				errorCodes.ToAPIErr(ErrTooManyRequests),
 				r.URL)
 		case <-r.Context().Done():
+			globalHTTPStats.addRequestsInQueue(-1)
 			// When the client disconnects before getting the S3 handler
 			// status code response, set the status code to 499 so this request
 			// will be properly audited and traced.

cmd/handler-utils.go

@@ -33,8 +33,6 @@ import (
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/minio/internal/mcontext"
 	xnet "github.com/minio/pkg/v3/net"
-	"golang.org/x/exp/maps"
-	"golang.org/x/exp/slices"
 )
 
 const (
@@ -90,6 +88,7 @@ var supportedHeaders = []string{
 	"X-Minio-Replication-Server-Side-Encryption-Iv",
 	"X-Minio-Replication-Encrypted-Multipart",
 	"X-Minio-Replication-Actual-Object-Size",
+	ReplicationSsecChecksumHeader,
 	// Add more supported headers here.
 }
 
@@ -110,6 +109,7 @@ var replicationToInternalHeaders = map[string]string{
 	"X-Minio-Replication-Server-Side-Encryption-Iv":             "X-Minio-Internal-Server-Side-Encryption-Iv",
 	"X-Minio-Replication-Encrypted-Multipart":                   "X-Minio-Internal-Encrypted-Multipart",
 	"X-Minio-Replication-Actual-Object-Size":                    "X-Minio-Internal-Actual-Object-Size",
+	ReplicationSsecChecksumHeader:                               ReplicationSsecChecksumHeader,
 	// Add more supported headers here.
 }
 
@@ -206,8 +206,8 @@ func extractMetadataFromMime(ctx context.Context, v textproto.MIMEHeader, m map[
 	for _, supportedHeader := range supportedHeaders {
 		value, ok := nv[http.CanonicalHeaderKey(supportedHeader)]
 		if ok {
-			if slices.Contains(maps.Keys(replicationToInternalHeaders), supportedHeader) {
+			if v, ok := replicationToInternalHeaders[supportedHeader]; ok {
-				m[replicationToInternalHeaders[supportedHeader]] = strings.Join(value, ",")
+				m[v] = strings.Join(value, ",")
 			} else {
 				m[supportedHeader] = strings.Join(value, ",")
 			}

cmd/http-tracer.go

@@ -142,6 +142,7 @@ func httpTracerMiddleware(h http.Handler) http.Handler {
 			Time:      reqStartTime,
 			Duration:  reqEndTime.Sub(respRecorder.StartTime),
 			Path:      reqPath,
+			Bytes:     int64(inputBytes + respRecorder.Size()),
 			HTTP: &madmin.TraceHTTPStats{
 				ReqInfo: madmin.TraceRequestInfo{
 					Time:     reqStartTime,

cmd/iam-etcd-store.go

@@ -236,9 +236,8 @@ func (ies *IAMEtcdStore) addUser(ctx context.Context, user string, userType IAMU
 				// for the expiring credentials.
 				deleteKeyEtcd(ctx, ies.client, getUserIdentityPath(user, userType))
 				deleteKeyEtcd(ctx, ies.client, getMappedPolicyPath(user, userType, false))
-				return nil
 			}
-			return err
+			return nil
 		}
 		u.Credentials.Claims = jwtClaims.Map()
 	}

cmd/iam-object-store.go

@@ -25,7 +25,6 @@ import (
 	"path"
 	"strings"
 	"sync"
-	"sync/atomic"
 	"time"
 	"unicode/utf8"
 
@@ -34,6 +33,7 @@ import (
 	"github.com/minio/minio/internal/config"
 	xioutil "github.com/minio/minio/internal/ioutil"
 	"github.com/minio/minio/internal/kms"
+	"github.com/minio/minio/internal/logger"
 	"github.com/puzpuzpuz/xsync/v3"
 )
 
@@ -44,8 +44,6 @@ type IAMObjectStore struct {
 
 	*iamCache
 
-	cachedIAMListing atomic.Value
-
 	usersSysType UsersSysType
 
 	objAPI ObjectLayer
@@ -256,9 +254,8 @@ func (iamOS *IAMObjectStore) loadUser(ctx context.Context, user string, userType
 				// for the expiring credentials.
 				iamOS.deleteIAMConfig(ctx, getUserIdentityPath(user, userType))
 				iamOS.deleteIAMConfig(ctx, getMappedPolicyPath(user, userType, false))
-				return nil
 			}
-			return err
+			return nil
 
 		}
 		u.Credentials.Claims = jwtClaims.Map()
@@ -421,8 +418,8 @@ func splitPath(s string, lastIndex bool) (string, string) {
 	return s[:i+1], s[i+1:]
 }
 
-func (iamOS *IAMObjectStore) listAllIAMConfigItems(ctx context.Context) (map[string][]string, error) {
+func (iamOS *IAMObjectStore) listAllIAMConfigItems(ctx context.Context) (res map[string][]string, err error) {
-	res := make(map[string][]string)
+	res = make(map[string][]string)
 	ctx, cancel := context.WithCancel(ctx)
 	defer cancel()
 	for item := range listIAMConfigItems(ctx, iamOS.objAPI, iamConfigPrefix+SlashSeparator) {
@@ -438,62 +435,10 @@ func (iamOS *IAMObjectStore) listAllIAMConfigItems(ctx context.Context) (map[str
 
 		res[listKey] = append(res[listKey], trimmedItem)
 	}
-	// Store the listing for later re-use.
+
-	iamOS.cachedIAMListing.Store(res)
 	return res, nil
 }
 
-// PurgeExpiredSTS - purge expired STS credentials from object store.
-func (iamOS *IAMObjectStore) PurgeExpiredSTS(ctx context.Context) error {
-	if iamOS.objAPI == nil {
-		return errServerNotInitialized
-	}
-
-	bootstrapTraceMsg("purging expired STS credentials")
-
-	iamListing, ok := iamOS.cachedIAMListing.Load().(map[string][]string)
-	if !ok {
-		// There has been no store yet. This should never happen!
-		iamLogIf(GlobalContext, errors.New("WARNING: no cached IAM listing found"))
-		return nil
-	}
-
-	// Scan STS users on disk and purge expired ones. We do not need to hold a
-	// lock with store.lock() here.
-	stsAccountsFromStore := map[string]UserIdentity{}
-	stsAccPoliciesFromStore := xsync.NewMapOf[string, MappedPolicy]()
-	for _, item := range iamListing[stsListKey] {
-		userName := path.Dir(item)
-		// loadUser() will delete expired user during the load.
-		err := iamOS.loadUser(ctx, userName, stsUser, stsAccountsFromStore)
-		if err != nil && !errors.Is(err, errNoSuchUser) {
-			iamLogIf(GlobalContext,
-				fmt.Errorf("unable to load user during STS purge: %w (%s)", err, item))
-		}
-
-	}
-	// Loading the STS policy mappings from disk ensures that stale entries
-	// (removed during loadUser() in the loop above) are removed from memory.
-	for _, item := range iamListing[policyDBSTSUsersListKey] {
-		stsName := strings.TrimSuffix(item, ".json")
-		err := iamOS.loadMappedPolicy(ctx, stsName, stsUser, false, stsAccPoliciesFromStore)
-		if err != nil && !errors.Is(err, errNoSuchPolicy) {
-			iamLogIf(GlobalContext,
-				fmt.Errorf("unable to load policies during STS purge: %w (%s)", err, item))
-		}
-
-	}
-
-	// Store the newly populated map in the iam cache. This takes care of
-	// removing stale entries from the existing map.
-	iamOS.Lock()
-	defer iamOS.Unlock()
-	iamOS.iamCache.iamSTSAccountsMap = stsAccountsFromStore
-	iamOS.iamCache.iamSTSPolicyMap = stsAccPoliciesFromStore
-
-	return nil
-}
-
 // Assumes cache is locked by caller.
 func (iamOS *IAMObjectStore) loadAllFromObjStore(ctx context.Context, cache *iamCache) error {
 	if iamOS.objAPI == nil {
@@ -594,6 +539,45 @@ func (iamOS *IAMObjectStore) loadAllFromObjStore(ctx context.Context, cache *iam
 	}
 
 	cache.buildUserGroupMemberships()
+
+	purgeStart := time.Now()
+
+	// Purge expired STS credentials.
+
+	// Scan STS users on disk and purge expired ones.
+	stsAccountsFromStore := map[string]UserIdentity{}
+	stsAccPoliciesFromStore := xsync.NewMapOf[string, MappedPolicy]()
+	for _, item := range listedConfigItems[stsListKey] {
+		userName := path.Dir(item)
+		// loadUser() will delete expired user during the load.
+		err := iamOS.loadUser(ctx, userName, stsUser, stsAccountsFromStore)
+		if err != nil && !errors.Is(err, errNoSuchUser) {
+			return fmt.Errorf("unable to load user during STS purge: %w (%s)", err, item)
+		}
+
+	}
+	// Loading the STS policy mappings from disk ensures that stale entries
+	// (removed during loadUser() in the loop above) are removed from memory.
+	for _, item := range listedConfigItems[policyDBSTSUsersListKey] {
+		stsName := strings.TrimSuffix(item, ".json")
+		err := iamOS.loadMappedPolicy(ctx, stsName, stsUser, false, stsAccPoliciesFromStore)
+		if err != nil && !errors.Is(err, errNoSuchPolicy) {
+			return fmt.Errorf("unable to load policies during STS purge: %w (%s)", err, item)
+		}
+
+	}
+
+	took := time.Since(purgeStart).Seconds()
+	if took > maxDurationSecondsForLog {
+		// Log if we took a lot of time to load.
+		logger.Info("IAM expired STS purge took %.2fs", took)
+	}
+
+	// Store the newly populated map in the iam cache. This takes care of
+	// removing stale entries from the existing map.
+	cache.iamSTSAccountsMap = stsAccountsFromStore
+	cache.iamSTSPolicyMap = stsAccPoliciesFromStore
+
 	return nil
 }
 

cmd/iam-store.go

@@ -33,8 +33,10 @@ import (
 	"github.com/minio/minio/internal/auth"
 	"github.com/minio/minio/internal/config/identity/openid"
 	"github.com/minio/minio/internal/jwt"
+	"github.com/minio/pkg/v3/console"
 	"github.com/minio/pkg/v3/policy"
 	"github.com/puzpuzpuz/xsync/v3"
+	"golang.org/x/sync/singleflight"
 )
 
 const (
@@ -532,16 +534,6 @@ func setDefaultCannedPolicies(policies map[string]PolicyDoc) {
 	}
 }
 
-// PurgeExpiredSTS - purges expired STS credentials.
-func (store *IAMStoreSys) PurgeExpiredSTS(ctx context.Context) error {
-	iamOS, ok := store.IAMStorageAPI.(*IAMObjectStore)
-	if !ok {
-		// No purging is done for non-object storage.
-		return nil
-	}
-	return iamOS.PurgeExpiredSTS(ctx)
-}
-
 // LoadIAMCache reads all IAM items and populates a new iamCache object and
 // replaces the in-memory cache object.
 func (store *IAMStoreSys) LoadIAMCache(ctx context.Context, firstTime bool) error {
@@ -643,6 +635,8 @@ func (store *IAMStoreSys) LoadIAMCache(ctx context.Context, firstTime bool) erro
 // layer.
 type IAMStoreSys struct {
 	IAMStorageAPI
+
+	group *singleflight.Group
 }
 
 // HasWatcher - returns if the storage system has a watcher.
@@ -1810,6 +1804,7 @@ func (store *IAMStoreSys) DeleteUser(ctx context.Context, accessKey string, user
 		err = nil
 	}
 	delete(cache.iamUsersMap, accessKey)
+	store.group.Forget(accessKey)
 
 	cache.updatedAt = time.Now()
 
@@ -1885,6 +1880,7 @@ func (store *IAMStoreSys) DeleteUsers(ctx context.Context, users []string) error
 			err := store.deleteUserIdentity(ctx, user, userType)
 			iamLogIf(GlobalContext, err)
 			delete(cache.iamUsersMap, user)
+			store.group.Forget(user)
 
 			deleted = true
 		}
@@ -2371,7 +2367,7 @@ func (store *IAMStoreSys) UpdateServiceAccount(ctx context.Context, accessKey st
 			return updatedAt, err
 		}
 
-		if len(policyBuf) > 2048 {
+		if len(policyBuf) > maxSVCSessionPolicySize {
 			return updatedAt, errSessionPolicyTooLarge
 		}
 
@@ -2572,37 +2568,43 @@ func (store *IAMStoreSys) UpdateUserIdentity(ctx context.Context, cred auth.Cred
 }
 
 // LoadUser - attempts to load user info from storage and updates cache.
-func (store *IAMStoreSys) LoadUser(ctx context.Context, accessKey string) {
+func (store *IAMStoreSys) LoadUser(ctx context.Context, accessKey string) error {
+	// We use singleflight to de-duplicate requests when server
+	// is coming up and loading accessKey and its associated assets
+	val, err, shared := store.group.Do(accessKey, func() (val interface{}, err error) {
 		cache := store.lock()
-	defer store.unlock()
+		defer func() {
-
 			cache.updatedAt = time.Now()
+			store.unlock()
+		}()
 
 		_, found := cache.iamUsersMap[accessKey]
 
 		// Check for regular user access key
 		if !found {
-		store.loadUser(ctx, accessKey, regUser, cache.iamUsersMap)
+			err = store.loadUser(ctx, accessKey, regUser, cache.iamUsersMap)
 			if _, found = cache.iamUsersMap[accessKey]; found {
 				// load mapped policies
-			store.loadMappedPolicyWithRetry(ctx, accessKey, regUser, false, cache.iamUserPolicyMap, 3)
+				err = store.loadMappedPolicyWithRetry(ctx, accessKey, regUser, false, cache.iamUserPolicyMap, 3)
 			}
 		}
 
 		// Check for service account
 		if !found {
-		store.loadUser(ctx, accessKey, svcUser, cache.iamUsersMap)
+			err = store.loadUser(ctx, accessKey, svcUser, cache.iamUsersMap)
 			var svc UserIdentity
 			svc, found = cache.iamUsersMap[accessKey]
 			if found {
 				// Load parent user and mapped policies.
 				if store.getUsersSysType() == MinIOUsersSysType {
-				store.loadUser(ctx, svc.Credentials.ParentUser, regUser, cache.iamUsersMap)
+					err = store.loadUser(ctx, svc.Credentials.ParentUser, regUser, cache.iamUsersMap)
-				store.loadMappedPolicyWithRetry(ctx, svc.Credentials.ParentUser, regUser, false, cache.iamUserPolicyMap, 3)
+					if err == nil {
+						err = store.loadMappedPolicyWithRetry(ctx, svc.Credentials.ParentUser, regUser, false, cache.iamUserPolicyMap, 3)
+					}
 				} else {
 					// In case of LDAP the parent user's policy mapping needs to be
 					// loaded into sts map
-				store.loadMappedPolicyWithRetry(ctx, svc.Credentials.ParentUser, stsUser, false, cache.iamSTSPolicyMap, 3)
+					err = store.loadMappedPolicyWithRetry(ctx, svc.Credentials.ParentUser, stsUser, false, cache.iamSTSPolicyMap, 3)
 				}
 			}
 		}
@@ -2611,10 +2613,10 @@ func (store *IAMStoreSys) LoadUser(ctx context.Context, accessKey string) {
 		stsAccountFound := false
 		var stsUserCred UserIdentity
 		if !found {
-		store.loadUser(ctx, accessKey, stsUser, cache.iamSTSAccountsMap)
+			err = store.loadUser(ctx, accessKey, stsUser, cache.iamSTSAccountsMap)
 			if stsUserCred, found = cache.iamSTSAccountsMap[accessKey]; found {
 				// Load mapped policy
-			store.loadMappedPolicyWithRetry(ctx, stsUserCred.Credentials.ParentUser, stsUser, false, cache.iamSTSPolicyMap, 3)
+				err = store.loadMappedPolicyWithRetry(ctx, stsUserCred.Credentials.ParentUser, stsUser, false, cache.iamSTSPolicyMap, 3)
 				stsAccountFound = true
 			}
 		}
@@ -2624,17 +2626,30 @@ func (store *IAMStoreSys) LoadUser(ctx context.Context, accessKey string) {
 			pols, _ := cache.iamUserPolicyMap.Load(accessKey)
 			for _, policy := range pols.toSlice() {
 				if _, found = cache.iamPolicyDocsMap[policy]; !found {
-				store.loadPolicyDocWithRetry(ctx, policy, cache.iamPolicyDocsMap, 3)
+					err = store.loadPolicyDocWithRetry(ctx, policy, cache.iamPolicyDocsMap, 3)
 				}
 			}
 		} else {
 			pols, _ := cache.iamSTSPolicyMap.Load(stsUserCred.Credentials.AccessKey)
 			for _, policy := range pols.toSlice() {
 				if _, found = cache.iamPolicyDocsMap[policy]; !found {
-				store.loadPolicyDocWithRetry(ctx, policy, cache.iamPolicyDocsMap, 3)
+					err = store.loadPolicyDocWithRetry(ctx, policy, cache.iamPolicyDocsMap, 3)
 				}
 			}
 		}
+
+		return "done", err
+	})
+
+	if serverDebugLog {
+		console.Debugln("loadUser: loading shared", val, err, shared)
+	}
+
+	if IsErr(err, errNoSuchUser, errNoSuchPolicy, errNoSuchGroup) {
+		return nil
+	}
+
+	return err
 }
 
 func extractJWTClaims(u UserIdentity) (*jwt.MapClaims, error) {

cmd/iam.go

@@ -52,6 +52,7 @@ import (
 	"github.com/minio/pkg/v3/ldap"
 	"github.com/minio/pkg/v3/policy"
 	etcd "go.etcd.io/etcd/client/v3"
+	"golang.org/x/sync/singleflight"
 )
 
 // UsersSysType - defines the type of users and groups system that is
@@ -78,6 +79,10 @@ const (
 	inheritedPolicyType = "inherited-policy"
 )
 
+const (
+	maxSVCSessionPolicySize = 4096
+)
+
 // IAMSys - config system.
 type IAMSys struct {
 	// Need to keep them here to keep alignment - ref: https://golang.org/pkg/sync/atomic/#pkg-note-BUG
@@ -173,9 +178,9 @@ func (sys *IAMSys) initStore(objAPI ObjectLayer, etcdClient *etcd.Client) {
 	}
 
 	if etcdClient == nil {
-		sys.store = &IAMStoreSys{newIAMObjectStore(objAPI, sys.usersSysType)}
+		sys.store = &IAMStoreSys{newIAMObjectStore(objAPI, sys.usersSysType), &singleflight.Group{}}
 	} else {
-		sys.store = &IAMStoreSys{newIAMEtcdStore(etcdClient, sys.usersSysType)}
+		sys.store = &IAMStoreSys{newIAMEtcdStore(etcdClient, sys.usersSysType), &singleflight.Group{}}
 	}
 }
 
@@ -348,6 +353,8 @@ func (sys *IAMSys) Init(ctx context.Context, objAPI ObjectLayer, etcdClient *etc
 	bootstrapTraceMsg("finishing IAM loading")
 }
 
+const maxDurationSecondsForLog = 5
+
 func (sys *IAMSys) periodicRoutines(ctx context.Context, baseInterval time.Duration) {
 	// Watch for IAM config changes for iamStorageWatcher.
 	watcher, isWatcher := sys.store.IAMStorageAPI.(iamStorageWatcher)
@@ -380,7 +387,6 @@ func (sys *IAMSys) periodicRoutines(ctx context.Context, baseInterval time.Durat
 		return baseInterval/2 + randAmt
 	}
 
-	var maxDurationSecondsForLog float64 = 5
 	timer := time.NewTimer(waitInterval())
 	defer timer.Stop()
 
@@ -399,18 +405,6 @@ func (sys *IAMSys) periodicRoutines(ctx context.Context, baseInterval time.Durat
 				}
 			}
 
-			// Purge expired STS credentials.
-			purgeStart := time.Now()
-			if err := sys.store.PurgeExpiredSTS(ctx); err != nil {
-				iamLogIf(ctx, fmt.Errorf("Failure in periodic STS purge for IAM (took %.2fs): %v", time.Since(purgeStart).Seconds(), err))
-			} else {
-				took := time.Since(purgeStart).Seconds()
-				if took > maxDurationSecondsForLog {
-					// Log if we took a lot of time to load.
-					logger.Info("IAM expired STS purge took %.2fs", took)
-				}
-			}
-
 			// The following actions are performed about once in 4 times that
 			// IAM is refreshed:
 			if r.Intn(4) == 0 {
@@ -977,7 +971,7 @@ func (sys *IAMSys) NewServiceAccount(ctx context.Context, parentUser string, gro
 		if err != nil {
 			return auth.Credentials{}, time.Time{}, err
 		}
-		if len(policyBuf) > 2048 {
+		if len(policyBuf) > maxSVCSessionPolicySize {
 			return auth.Credentials{}, time.Time{}, errSessionPolicyTooLarge
 		}
 	}
@@ -1273,6 +1267,10 @@ func (sys *IAMSys) CreateUser(ctx context.Context, accessKey string, ureq madmin
 		return updatedAt, auth.ErrInvalidAccessKeyLength
 	}
 
+	if auth.ContainsReservedChars(accessKey) {
+		return updatedAt, auth.ErrContainsReservedChars
+	}
+
 	if !auth.IsSecretKeyValid(ureq.SecretKey) {
 		return updatedAt, auth.ErrInvalidSecretKeyLength
 	}
@@ -1570,20 +1568,6 @@ func (sys *IAMSys) NormalizeLDAPAccessKeypairs(ctx context.Context, accessKeyMap
 
 func (sys *IAMSys) getStoredLDAPPolicyMappingKeys(ctx context.Context, isGroup bool) set.StringSet {
 	entityKeysInStorage := set.NewStringSet()
-	if iamOS, ok := sys.store.IAMStorageAPI.(*IAMObjectStore); ok {
-		// Load existing mapping keys from the cached listing for
-		// `IAMObjectStore`.
-		iamFilesListing := iamOS.cachedIAMListing.Load().(map[string][]string)
-		listKey := policyDBSTSUsersListKey
-		if isGroup {
-			listKey = policyDBGroupsListKey
-		}
-		for _, item := range iamFilesListing[listKey] {
-			stsUserName := strings.TrimSuffix(item, ".json")
-			entityKeysInStorage.Add(stsUserName)
-		}
-	} else {
-		// For non-iam object store, we copy the mapping keys from the cache.
 	cache := sys.store.rlock()
 	defer sys.store.runlock()
 	cachedPolicyMap := cache.iamSTSPolicyMap
@@ -1594,7 +1578,6 @@ func (sys *IAMSys) getStoredLDAPPolicyMappingKeys(ctx context.Context, isGroup b
 		entityKeysInStorage.Add(k)
 		return true
 	})
-	}
 
 	return entityKeysInStorage
 }
@@ -1720,31 +1703,46 @@ func (sys *IAMSys) NormalizeLDAPMappingImport(ctx context.Context, isGroup bool,
 	return nil
 }
 
-// GetUser - get user credentials
+// CheckKey validates the incoming accessKey
-func (sys *IAMSys) GetUser(ctx context.Context, accessKey string) (u UserIdentity, ok bool) {
+func (sys *IAMSys) CheckKey(ctx context.Context, accessKey string) (u UserIdentity, ok bool, err error) {
 	if !sys.Initialized() {
-		return u, false
+		return u, false, nil
 	}
 
 	if accessKey == globalActiveCred.AccessKey {
-		return newUserIdentity(globalActiveCred), true
+		return newUserIdentity(globalActiveCred), true, nil
 	}
 
 	loadUserCalled := false
 	select {
 	case <-sys.configLoaded:
 	default:
-		sys.store.LoadUser(ctx, accessKey)
+		err = sys.store.LoadUser(ctx, accessKey)
 		loadUserCalled = true
 	}
 
 	u, ok = sys.store.GetUser(accessKey)
 	if !ok && !loadUserCalled {
-		sys.store.LoadUser(ctx, accessKey)
+		err = sys.store.LoadUser(ctx, accessKey)
+		loadUserCalled = true
+
 		u, ok = sys.store.GetUser(accessKey)
 	}
 
-	return u, ok && u.Credentials.IsValid()
+	if !ok && loadUserCalled && err != nil {
+		iamLogOnceIf(ctx, err, accessKey)
+
+		// return 503 to application
+		return u, false, errIAMNotInitialized
+	}
+
+	return u, ok && u.Credentials.IsValid(), nil
+}
+
+// GetUser - get user credentials
+func (sys *IAMSys) GetUser(ctx context.Context, accessKey string) (u UserIdentity, ok bool) {
+	u, ok, _ = sys.CheckKey(ctx, accessKey)
+	return u, ok
 }
 
 // Notify all other MinIO peers to load group.
@@ -1766,6 +1764,10 @@ func (sys *IAMSys) AddUsersToGroup(ctx context.Context, group string, members []
 		return updatedAt, errServerNotInitialized
 	}
 
+	if auth.ContainsReservedChars(group) {
+		return updatedAt, errGroupNameContainsReservedChars
+	}
+
 	updatedAt, err = sys.store.AddUsersToGroup(ctx, group, members)
 	if err != nil {
 		return updatedAt, err
@@ -1974,20 +1976,22 @@ func (sys *IAMSys) PolicyDBUpdateLDAP(ctx context.Context, isAttach bool,
 		}
 		isGroup = false
 	} else {
-		if isAttach {
 		var underBaseDN bool
 		if dnResult, underBaseDN, err = sys.LDAPConfig.GetValidatedGroupDN(nil, r.Group); err != nil {
 			iamLogIf(ctx, err)
 			return
-			} else if dnResult == nil || !underBaseDN {
+		}
+		if dnResult == nil || !underBaseDN {
+			if !isAttach {
+				dn = r.Group
+			} else {
 				err = errNoSuchGroup
 				return
 			}
+		} else {
 			// We use the group DN returned by the LDAP server (this may not
 			// equal the input group name, but we assume it is canonical).
 			dn = dnResult.NormDN
-		} else {
-			dn = r.Group
 		}
 		isGroup = true
 	}

cmd/logging.go

@@ -8,6 +8,10 @@ import (
 	"github.com/minio/minio/internal/logger"
 )
 
+func proxyLogIf(ctx context.Context, err error, errKind ...interface{}) {
+	logger.LogIf(ctx, "proxy", err, errKind...)
+}
+
 func replLogIf(ctx context.Context, err error, errKind ...interface{}) {
 	logger.LogIf(ctx, "replication", err, errKind...)
 }
@@ -16,6 +20,10 @@ func replLogOnceIf(ctx context.Context, err error, id string, errKind ...interfa
 	logger.LogOnceIf(ctx, "replication", err, id, errKind...)
 }
 
+func iamLogOnceIf(ctx context.Context, err error, id string, errKind ...interface{}) {
+	logger.LogOnceIf(ctx, "iam", err, id, errKind...)
+}
+
 func iamLogIf(ctx context.Context, err error, errKind ...interface{}) {
 	if !errors.Is(err, grid.ErrDisconnected) {
 		logger.LogIf(ctx, "iam", err, errKind...)

cmd/main.go

@@ -189,12 +189,14 @@ func printMinIOVersion(c *cli.Context) {
 	io.Copy(c.App.Writer, versionBanner(c))
 }
 
+var debugNoExit = env.Get("_MINIO_DEBUG_NO_EXIT", "") != ""
+
 // Main main for minio server.
 func Main(args []string) {
 	// Set the minio app name.
 	appName := filepath.Base(args[0])
 
-	if env.Get("_MINIO_DEBUG_NO_EXIT", "") != "" {
+	if debugNoExit {
 		freeze := func(_ int) {
 			// Infinite blocking op
 			<-make(chan struct{})

cmd/metacache-entries.go

@@ -20,7 +20,6 @@ package cmd
 import (
 	"bytes"
 	"context"
-	"errors"
 	"os"
 	"path"
 	"sort"
@@ -260,7 +259,7 @@ func (e *metaCacheEntry) fileInfo(bucket string) (FileInfo, error) {
 		}
 		return e.cached.ToFileInfo(bucket, e.name, "", false, false)
 	}
-	return getFileInfo(e.metadata, bucket, e.name, "", false, false)
+	return getFileInfo(e.metadata, bucket, e.name, "", fileInfoOpts{})
 }
 
 // xlmeta returns the decoded metadata.
@@ -301,7 +300,7 @@ func (e *metaCacheEntry) fileInfoVersions(bucket string) (FileInfoVersions, erro
 		}, nil
 	}
 	// Too small gains to reuse cache here.
-	return getFileInfoVersions(e.metadata, bucket, e.name, false)
+	return getFileInfoVersions(e.metadata, bucket, e.name, false, true)
 }
 
 // metaCacheEntries is a slice of metacache entries.
@@ -385,9 +384,6 @@ func (m metaCacheEntries) resolve(r *metadataResolutionParams) (selected *metaCa
 		// shallow decode.
 		xl, err := entry.xlmeta()
 		if err != nil {
-			if !errors.Is(err, errFileNotFound) {
-				internalLogIf(GlobalContext, err)
-			}
 			continue
 		}
 		objsValid++
@@ -752,7 +748,9 @@ func mergeEntryChannels(ctx context.Context, in []chan metaCacheEntry, out chan<
 				if !dirMatches {
 					// We have an object `name` or 'name/' and a directory `name/`.
 					if other.isDir() {
+						if serverDebugLog {
 							console.Debugln("mergeEntryChannels: discarding directory", other.name, "for object", best.name)
+						}
 						// Discard the directory.
 						if err := selectFrom(otherIdx); err != nil {
 							return err
@@ -760,7 +758,9 @@ func mergeEntryChannels(ctx context.Context, in []chan metaCacheEntry, out chan<
 						continue
 					}
 					// Replace directory with object.
+					if serverDebugLog {
 						console.Debugln("mergeEntryChannels: discarding directory", best.name, "for object", other.name)
+					}
 					toMerge = toMerge[:0]
 					best = other
 					bestIdx = otherIdx

cmd/metacache-set.go

@@ -391,7 +391,7 @@ func (r *metacacheReader) filter(o listPathOptions) (entries metaCacheEntriesSor
 			if !o.InclDeleted && entry.isObject() && entry.isLatestDeletemarker() && !entry.isObjectDir() {
 				return true
 			}
-			if entry.isAllFreeVersions() {
+			if !o.InclDeleted && entry.isAllFreeVersions() {
 				return true
 			}
 			entries.o = append(entries.o, entry)

cmd/metacache-walk.go

@@ -405,7 +405,7 @@ func (p *xlStorageDiskIDCheck) WalkDir(ctx context.Context, opts WalkDirOptions,
 	if err != nil {
 		return err
 	}
-	defer done(&err)
+	defer done(0, &err)
 
 	return p.storage.WalkDir(ctx, opts, wr)
 }

cmd/metrics-realtime.go

@@ -132,6 +132,15 @@ func collectLocalMetrics(types madmin.MetricType, opts collectMetricsOpts) (m ma
 			m.Aggregated.CPU.LoadStat = loadStat
 		}
 	}
+	if types.Contains(madmin.MetricsRPC) {
+		gr := globalGrid.Load()
+		if gr == nil {
+			m.Errors = append(m.Errors, fmt.Sprintf("%s: Grid not initialized", byHostName))
+		} else {
+			stats := gr.ConnStats()
+			m.Aggregated.RPC = &stats
+		}
+	}
 	// Add types...
 
 	// ByHost is a shallow reference, so careful about sharing.

cmd/metrics-v2.go

@@ -273,13 +273,10 @@ const (
 	vmemory          = "virtual_memory_bytes"
 	cpu              = "cpu_total_seconds"
 
-	expiryPendingTasks           MetricName = "expiry_pending_tasks"
 	expiryMissedTasks            MetricName = "expiry_missed_tasks"
 	expiryMissedFreeVersions     MetricName = "expiry_missed_freeversions"
 	expiryMissedTierJournalTasks MetricName = "expiry_missed_tierjournal_tasks"
 	expiryNumWorkers             MetricName = "expiry_num_workers"
-	transitionPendingTasks       MetricName = "transition_pending_tasks"
-	transitionActiveTasks        MetricName = "transition_active_tasks"
 	transitionMissedTasks        MetricName = "transition_missed_immediate_tasks"
 
 	transitionedBytes    MetricName = "transitioned_bytes"
@@ -820,7 +817,7 @@ func getClusterObjectVersionsMD() MetricDescription {
 
 func getClusterRepLinkLatencyCurrMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      currLinkLatency,
 		Help:      "Replication current link latency in milliseconds",
@@ -830,7 +827,7 @@ func getClusterRepLinkLatencyCurrMD() MetricDescription {
 
 func getClusterRepLinkOnlineMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      linkOnline,
 		Help:      "Reports whether replication link is online (1) or offline(0)",
@@ -840,7 +837,7 @@ func getClusterRepLinkOnlineMD() MetricDescription {
 
 func getClusterRepLinkCurrOfflineDurationMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      linkOfflineDuration,
 		Help:      "Duration of replication link being offline in seconds since last offline event",
@@ -850,7 +847,7 @@ func getClusterRepLinkCurrOfflineDurationMD() MetricDescription {
 
 func getClusterRepLinkTotalOfflineDurationMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      linkDowntimeTotalDuration,
 		Help:      "Total downtime of replication link in seconds since server uptime",
@@ -978,7 +975,7 @@ func getRepReceivedOperationsMD(namespace MetricNamespace) MetricDescription {
 
 func getClusterReplMRFFailedOperationsMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      recentBacklogCount,
 		Help:      "Total number of objects seen in replication backlog in the last 5 minutes",
@@ -998,7 +995,7 @@ func getClusterRepCredentialErrorsMD(namespace MetricNamespace) MetricDescriptio
 
 func getClusterReplCurrQueuedOperationsMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      currInQueueCount,
 		Help:      "Total number of objects queued for replication in the last full minute",
@@ -1008,7 +1005,7 @@ func getClusterReplCurrQueuedOperationsMD() MetricDescription {
 
 func getClusterReplCurrQueuedBytesMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      currInQueueBytes,
 		Help:      "Total number of bytes queued for replication in the last full minute",
@@ -1018,7 +1015,7 @@ func getClusterReplCurrQueuedBytesMD() MetricDescription {
 
 func getClusterReplActiveWorkersCountMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      currActiveWorkers,
 		Help:      "Total number of active replication workers",
@@ -1028,7 +1025,7 @@ func getClusterReplActiveWorkersCountMD() MetricDescription {
 
 func getClusterReplAvgActiveWorkersCountMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      avgActiveWorkers,
 		Help:      "Average number of active replication workers",
@@ -1038,7 +1035,7 @@ func getClusterReplAvgActiveWorkersCountMD() MetricDescription {
 
 func getClusterReplMaxActiveWorkersCountMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      maxActiveWorkers,
 		Help:      "Maximum number of active replication workers seen since server uptime",
@@ -1048,7 +1045,7 @@ func getClusterReplMaxActiveWorkersCountMD() MetricDescription {
 
 func getClusterReplCurrentTransferRateMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      currTransferRate,
 		Help:      "Current replication transfer rate in bytes/sec",
@@ -1058,7 +1055,7 @@ func getClusterReplCurrentTransferRateMD() MetricDescription {
 
 func getClusterRepLinkLatencyMaxMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      maxLinkLatency,
 		Help:      "Maximum replication link latency in milliseconds seen since server uptime",
@@ -1068,7 +1065,7 @@ func getClusterRepLinkLatencyMaxMD() MetricDescription {
 
 func getClusterRepLinkLatencyAvgMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      avgLinkLatency,
 		Help:      "Average replication link latency in milliseconds",
@@ -1078,7 +1075,7 @@ func getClusterRepLinkLatencyAvgMD() MetricDescription {
 
 func getClusterReplAvgQueuedOperationsMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      avgInQueueCount,
 		Help:      "Average number of objects queued for replication since server uptime",
@@ -1088,7 +1085,7 @@ func getClusterReplAvgQueuedOperationsMD() MetricDescription {
 
 func getClusterReplAvgQueuedBytesMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      avgInQueueBytes,
 		Help:      "Average number of bytes queued for replication since server uptime",
@@ -1098,7 +1095,7 @@ func getClusterReplAvgQueuedBytesMD() MetricDescription {
 
 func getClusterReplMaxQueuedOperationsMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      maxInQueueCount,
 		Help:      "Maximum number of objects queued for replication since server uptime",
@@ -1108,7 +1105,7 @@ func getClusterReplMaxQueuedOperationsMD() MetricDescription {
 
 func getClusterReplMaxQueuedBytesMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      maxInQueueBytes,
 		Help:      "Maximum number of bytes queued for replication since server uptime",
@@ -1118,7 +1115,7 @@ func getClusterReplMaxQueuedBytesMD() MetricDescription {
 
 func getClusterReplAvgTransferRateMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      avgTransferRate,
 		Help:      "Average replication transfer rate in bytes/sec",
@@ -1128,7 +1125,7 @@ func getClusterReplAvgTransferRateMD() MetricDescription {
 
 func getClusterReplMaxTransferRateMD() MetricDescription {
 	return MetricDescription{
-		Namespace: clusterMetricNamespace,
+		Namespace: nodeMetricNamespace,
 		Subsystem: replicationSubsystem,
 		Name:      maxTransferRate,
 		Help:      "Maximum replication transfer rate in bytes/sec seen since server uptime",
@@ -1691,7 +1688,7 @@ func getMinioProcMetrics() *MetricsGroupV2 {
 		cacheInterval: 10 * time.Second,
 	}
 	mg.RegisterRead(func(ctx context.Context) (metrics []MetricV2) {
-		if runtime.GOOS == "windows" {
+		if runtime.GOOS == globalWindowsOSName || runtime.GOOS == globalMacOSName {
 			return nil
 		}
 

cmd/metrics-v3-ilm.go

@@ -0,0 +1,53 @@
+// Copyright (c) 2024 MinIO, Inc.
+//
+// # This file is part of MinIO Object Storage stack
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package cmd
+
+import (
+	"context"
+)
+
+const (
+	expiryPendingTasks             = "expiry_pending_tasks"
+	transitionActiveTasks          = "transition_active_tasks"
+	transitionPendingTasks         = "transition_pending_tasks"
+	transitionMissedImmediateTasks = "transition_missed_immediate_tasks"
+	versionsScanned                = "versions_scanned"
+)
+
+var (
+	ilmExpiryPendingTasksMD             = NewGaugeMD(expiryPendingTasks, "Number of pending ILM expiry tasks in the queue")
+	ilmTransitionActiveTasksMD          = NewGaugeMD(transitionActiveTasks, "Number of active ILM transition tasks")
+	ilmTransitionPendingTasksMD         = NewGaugeMD(transitionPendingTasks, "Number of pending ILM transition tasks in the queue")
+	ilmTransitionMissedImmediateTasksMD = NewCounterMD(transitionMissedImmediateTasks, "Number of missed immediate ILM transition tasks")
+	ilmVersionsScannedMD                = NewCounterMD(versionsScanned, "Total number of object versions checked for ILM actions since server start")
+)
+
+// loadILMMetrics - `MetricsLoaderFn` for ILM metrics.
+func loadILMMetrics(_ context.Context, m MetricValues, _ *metricsCache) error {
+	if globalExpiryState != nil {
+		m.Set(expiryPendingTasks, float64(globalExpiryState.PendingTasks()))
+	}
+	if globalTransitionState != nil {
+		m.Set(transitionActiveTasks, float64(globalTransitionState.ActiveTasks()))
+		m.Set(transitionPendingTasks, float64(globalTransitionState.PendingTasks()))
+		m.Set(transitionMissedImmediateTasks, float64(globalTransitionState.MissedImmediateTasks()))
+	}
+	m.Set(versionsScanned, float64(globalScannerMetrics.lifetime(scannerMetricILM)))
+
+	return nil
+}

cmd/metrics-v3-system-cpu.go

@@ -55,6 +55,7 @@ func loadCPUMetrics(ctx context.Context, m MetricValues, c *metricsCache) error
 	}
 
 	ts := cpuMetrics.TimesStat
+	if ts != nil {
 		tot := ts.User + ts.System + ts.Idle + ts.Iowait + ts.Nice + ts.Steal
 		cpuUserVal := math.Round(ts.User/tot*100*100) / 100
 		m.Set(sysCPUUser, cpuUserVal)
@@ -64,6 +65,7 @@ func loadCPUMetrics(ctx context.Context, m MetricValues, c *metricsCache) error
 		m.Set(sysCPUNice, cpuNiceVal)
 		cpuStealVal := math.Round(ts.Steal/tot*100*100) / 100
 		m.Set(sysCPUSteal, cpuStealVal)
+	}
 
 	// metrics-resource.go runs a job to collect resource metrics including their Avg values and
 	// stores them in resourceMetricsMap. We can use it to get the Avg values of CPU idle and IOWait.

cmd/metrics-v3-system-process.go

@@ -156,12 +156,14 @@ func loadProcessMetrics(ctx context.Context, m MetricValues, c *metricsCache) er
 		m.Set(processUptimeSeconds, time.Since(globalBootTime).Seconds())
 	}
 
+	if runtime.GOOS != globalWindowsOSName && runtime.GOOS != globalMacOSName {
 		p, err := procfs.Self()
 		if err != nil {
 			metricsLogIf(ctx, err)
 		} else {
 			loadProcFSMetrics(ctx, p, m)
 		}
+	}
 
 	if globalIsDistErasure && globalLockServer != nil {
 		st := globalLockServer.stats()

cmd/metrics-v3.go

@@ -54,6 +54,7 @@ const (
 	clusterIAMCollectorPath          collectorPath = "/cluster/iam"
 	clusterConfigCollectorPath       collectorPath = "/cluster/config"
 
+	ilmCollectorPath           collectorPath = "/ilm"
 	auditCollectorPath         collectorPath = "/audit"
 	loggerWebhookCollectorPath collectorPath = "/logger/webhook"
 	replicationCollectorPath   collectorPath = "/replication"
@@ -382,6 +383,17 @@ func newMetricGroups(r *prometheus.Registry) *metricsV3Collection {
 		loadAuditMetrics,
 	)
 
+	ilmMG := NewMetricsGroup(ilmCollectorPath,
+		[]MetricDescriptor{
+			ilmExpiryPendingTasksMD,
+			ilmTransitionActiveTasksMD,
+			ilmTransitionPendingTasksMD,
+			ilmTransitionMissedImmediateTasksMD,
+			ilmVersionsScannedMD,
+		},
+		loadILMMetrics,
+	)
+
 	allMetricGroups := []*MetricsGroup{
 		apiRequestsMG,
 		bucketAPIMG,
@@ -402,6 +414,7 @@ func newMetricGroups(r *prometheus.Registry) *metricsV3Collection {
 		clusterReplicationMG,
 		clusterConfigMG,
 
+		ilmMG,
 		scannerMG,
 		auditMG,
 		loggerWebhookMG,

cmd/naughty-disk_test.go

@@ -215,9 +215,9 @@ func (d *naughtyDisk) RenameFile(ctx context.Context, srcVolume, srcPath, dstVol
 	return d.disk.RenameFile(ctx, srcVolume, srcPath, dstVolume, dstPath)
 }
 
-func (d *naughtyDisk) CheckParts(ctx context.Context, volume string, path string, fi FileInfo) (err error) {
+func (d *naughtyDisk) CheckParts(ctx context.Context, volume string, path string, fi FileInfo) (*CheckPartsResp, error) {
 	if err := d.calcError(); err != nil {
-		return err
+		return nil, err
 	}
 	return d.disk.CheckParts(ctx, volume, path, fi)
 }
@@ -289,9 +289,9 @@ func (d *naughtyDisk) ReadXL(ctx context.Context, volume string, path string, re
 	return d.disk.ReadXL(ctx, volume, path, readData)
 }
 
-func (d *naughtyDisk) VerifyFile(ctx context.Context, volume, path string, fi FileInfo) error {
+func (d *naughtyDisk) VerifyFile(ctx context.Context, volume, path string, fi FileInfo) (*CheckPartsResp, error) {
 	if err := d.calcError(); err != nil {
-		return err
+		return nil, err
 	}
 	return d.disk.VerifyFile(ctx, volume, path, fi)
 }

cmd/net.go

@@ -19,7 +19,6 @@ package cmd
 
 import (
 	"errors"
-	"fmt"
 	"net"
 	"net/url"
 	"runtime"
@@ -32,8 +31,16 @@ import (
 	xnet "github.com/minio/pkg/v3/net"
 )
 
+var (
 	// IPv4 addresses of localhost.
-var localIP4 = mustGetLocalIP4()
+	localIP4 = mustGetLocalIP4()
+
+	// IPv6 addresses of localhost.
+	localIP6 = mustGetLocalIP6()
+
+	// List of all local loopback addresses.
+	localLoopbacks = mustGetLocalLoopbacks()
+)
 
 // mustSplitHostPort is a wrapper to net.SplitHostPort() where error is assumed to be a fatal.
 func mustSplitHostPort(hostPort string) (host, port string) {
@@ -74,6 +81,16 @@ func mustGetLocalIPs() (ipList []net.IP) {
 	return ipList
 }
 
+func mustGetLocalLoopbacks() (ipList set.StringSet) {
+	ipList = set.NewStringSet()
+	for _, ip := range mustGetLocalIPs() {
+		if ip != nil && ip.IsLoopback() {
+			ipList.Add(ip.String())
+		}
+	}
+	return
+}
+
 // mustGetLocalIP4 returns IPv4 addresses of localhost.  It panics on error.
 func mustGetLocalIP4() (ipList set.StringSet) {
 	ipList = set.NewStringSet()
@@ -160,15 +177,15 @@ func getConsoleEndpoints() (consoleEndpoints []string) {
 	}
 	var ipList []string
 	if globalMinioConsoleHost == "" {
-		ipList = sortIPs(mustGetLocalIP4().ToSlice())
+		ipList = sortIPs(localIP4.ToSlice())
-		ipList = append(ipList, mustGetLocalIP6().ToSlice()...)
+		ipList = append(ipList, localIP6.ToSlice()...)
 	} else {
 		ipList = []string{globalMinioConsoleHost}
 	}
 
+	consoleEndpoints = make([]string, 0, len(ipList))
 	for _, ip := range ipList {
-		endpoint := fmt.Sprintf("%s://%s", getURLScheme(globalIsTLS), net.JoinHostPort(ip, globalMinioConsolePort))
+		consoleEndpoints = append(consoleEndpoints, getURLScheme(globalIsTLS)+"://"+net.JoinHostPort(ip, globalMinioConsolePort))
-		consoleEndpoints = append(consoleEndpoints, endpoint)
 	}
 
 	return consoleEndpoints
@@ -180,15 +197,15 @@ func getAPIEndpoints() (apiEndpoints []string) {
 	}
 	var ipList []string
 	if globalMinioHost == "" {
-		ipList = sortIPs(mustGetLocalIP4().ToSlice())
+		ipList = sortIPs(localIP4.ToSlice())
-		ipList = append(ipList, mustGetLocalIP6().ToSlice()...)
+		ipList = append(ipList, localIP6.ToSlice()...)
 	} else {
 		ipList = []string{globalMinioHost}
 	}
 
+	apiEndpoints = make([]string, 0, len(ipList))
 	for _, ip := range ipList {
-		endpoint := fmt.Sprintf("%s://%s", getURLScheme(globalIsTLS), net.JoinHostPort(ip, globalMinioPort))
+		apiEndpoints = append(apiEndpoints, getURLScheme(globalIsTLS)+"://"+net.JoinHostPort(ip, globalMinioPort))
-		apiEndpoints = append(apiEndpoints, endpoint)
 	}
 
 	return apiEndpoints

cmd/notification.go

@@ -424,7 +424,7 @@ func (sys *NotificationSys) SignalConfigReload(subSys string) []NotificationPeer
 		}
 		client := client
 		ng.Go(GlobalContext, func() error {
-			return client.SignalService(serviceReloadDynamic, subSys, false)
+			return client.SignalService(serviceReloadDynamic, subSys, false, nil)
 		}, idx, *client.host)
 	}
 	return ng.Wait()
@@ -440,14 +440,14 @@ func (sys *NotificationSys) SignalService(sig serviceSignal) []NotificationPeerE
 		client := client
 		ng.Go(GlobalContext, func() error {
 			// force == true preserves the current behavior
-			return client.SignalService(sig, "", false)
+			return client.SignalService(sig, "", false, nil)
 		}, idx, *client.host)
 	}
 	return ng.Wait()
 }
 
 // SignalServiceV2 - calls signal service RPC call on all peers with v2 API
-func (sys *NotificationSys) SignalServiceV2(sig serviceSignal, dryRun bool) []NotificationPeerErr {
+func (sys *NotificationSys) SignalServiceV2(sig serviceSignal, dryRun bool, execAt *time.Time) []NotificationPeerErr {
 	ng := WithNPeers(len(sys.peerClients))
 	for idx, client := range sys.peerClients {
 		if client == nil {
@@ -455,7 +455,7 @@ func (sys *NotificationSys) SignalServiceV2(sig serviceSignal, dryRun bool) []No
 		}
 		client := client
 		ng.Go(GlobalContext, func() error {
-			return client.SignalService(sig, "", dryRun)
+			return client.SignalService(sig, "", dryRun, execAt)
 		}, idx, *client.host)
 	}
 	return ng.Wait()
@@ -1314,7 +1314,7 @@ func (sys *NotificationSys) ServiceFreeze(ctx context.Context, freeze bool) []No
 		}
 		client := client
 		ng.Go(GlobalContext, func() error {
-			return client.SignalService(serviceSig, "", false)
+			return client.SignalService(serviceSig, "", false, nil)
 		}, idx, *client.host)
 	}
 	nerrs := ng.Wait()

cmd/object-api-datatypes.go

@@ -233,7 +233,7 @@ func (o ObjectInfo) ExpiresStr() string {
 
 // ArchiveInfo returns any saved zip archive meta information.
 // It will be decrypted if needed.
-func (o *ObjectInfo) ArchiveInfo() []byte {
+func (o *ObjectInfo) ArchiveInfo(h http.Header) []byte {
 	if len(o.UserDefined) == 0 {
 		return nil
 	}
@@ -243,7 +243,7 @@ func (o *ObjectInfo) ArchiveInfo() []byte {
 	}
 	data := []byte(z)
 	if v, ok := o.UserDefined[archiveTypeMetadataKey]; ok && v == archiveTypeEnc {
-		decrypted, err := o.metadataDecrypter()(archiveTypeEnc, data)
+		decrypted, err := o.metadataDecrypter(h)(archiveTypeEnc, data)
 		if err != nil {
 			encLogIf(GlobalContext, err)
 			return nil
@@ -326,6 +326,7 @@ func (ri ReplicateObjectInfo) ToObjectInfo() ObjectInfo {
 		VersionPurgeStatusInternal: ri.VersionPurgeStatusInternal,
 		DeleteMarker:               true,
 		UserDefined:                map[string]string{},
+		Checksum:                   ri.Checksum,
 	}
 }
 
@@ -357,6 +358,7 @@ type ReplicateObjectInfo struct {
 	TargetStatuses       map[string]replication.StatusType
 	TargetPurgeStatuses  map[string]VersionPurgeStatusType
 	ReplicationTimestamp time.Time
+	Checksum             []byte
 }
 
 // MultipartInfo captures metadata information about the uploadId

cmd/object-api-datatypes_gen.go

@@ -1896,9 +1896,9 @@ func (z *PartInfo) Msgsize() (s int) {
 // MarshalMsg implements msgp.Marshaler
 func (z *ReplicateObjectInfo) MarshalMsg(b []byte) (o []byte, err error) {
 	o = msgp.Require(b, z.Msgsize())
-	// map header, size 25
+	// map header, size 26
 	// string "Name"
-	o = append(o, 0xde, 0x0, 0x19, 0xa4, 0x4e, 0x61, 0x6d, 0x65)
+	o = append(o, 0xde, 0x0, 0x1a, 0xa4, 0x4e, 0x61, 0x6d, 0x65)
 	o = msgp.AppendString(o, z.Name)
 	// string "Bucket"
 	o = append(o, 0xa6, 0x42, 0x75, 0x63, 0x6b, 0x65, 0x74)
@@ -2012,6 +2012,9 @@ func (z *ReplicateObjectInfo) MarshalMsg(b []byte) (o []byte, err error) {
 	// string "ReplicationTimestamp"
 	o = append(o, 0xb4, 0x52, 0x65, 0x70, 0x6c, 0x69, 0x63, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x54, 0x69, 0x6d, 0x65, 0x73, 0x74, 0x61, 0x6d, 0x70)
 	o = msgp.AppendTime(o, z.ReplicationTimestamp)
+	// string "Checksum"
+	o = append(o, 0xa8, 0x43, 0x68, 0x65, 0x63, 0x6b, 0x73, 0x75, 0x6d)
+	o = msgp.AppendBytes(o, z.Checksum)
 	return
 }
 
@@ -2231,6 +2234,12 @@ func (z *ReplicateObjectInfo) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "ReplicationTimestamp")
 				return
 			}
+		case "Checksum":
+			z.Checksum, bts, err = msgp.ReadBytesBytes(bts, z.Checksum)
+			if err != nil {
+				err = msgp.WrapError(err, "Checksum")
+				return
+			}
 		default:
 			bts, err = msgp.Skip(bts)
 			if err != nil {
@@ -2259,7 +2268,7 @@ func (z *ReplicateObjectInfo) Msgsize() (s int) {
 			s += msgp.StringPrefixSize + len(za0003) + za0004.Msgsize()
 		}
 	}
-	s += 21 + msgp.TimeSize
+	s += 21 + msgp.TimeSize + 9 + msgp.BytesPrefixSize + len(z.Checksum)
 	return
 }
 

cmd/object-api-interface.go

@@ -181,6 +181,7 @@ type DeleteBucketOptions struct {
 type BucketOptions struct {
 	Deleted    bool // true only when site replication is enabled
 	Cached     bool // true only when we are requesting a cached response instead of hitting the disk for example ListBuckets() call.
+	NoMetadata bool
 }
 
 // SetReplicaStatus sets replica status and timestamp for delete operations in ObjectOptions

cmd/object-api-interface_gen.go

@@ -9,13 +9,16 @@ import (
 // MarshalMsg implements msgp.Marshaler
 func (z BucketOptions) MarshalMsg(b []byte) (o []byte, err error) {
 	o = msgp.Require(b, z.Msgsize())
-	// map header, size 2
+	// map header, size 3
 	// string "Deleted"
-	o = append(o, 0x82, 0xa7, 0x44, 0x65, 0x6c, 0x65, 0x74, 0x65, 0x64)
+	o = append(o, 0x83, 0xa7, 0x44, 0x65, 0x6c, 0x65, 0x74, 0x65, 0x64)
 	o = msgp.AppendBool(o, z.Deleted)
 	// string "Cached"
 	o = append(o, 0xa6, 0x43, 0x61, 0x63, 0x68, 0x65, 0x64)
 	o = msgp.AppendBool(o, z.Cached)
+	// string "NoMetadata"
+	o = append(o, 0xaa, 0x4e, 0x6f, 0x4d, 0x65, 0x74, 0x61, 0x64, 0x61, 0x74, 0x61)
+	o = msgp.AppendBool(o, z.NoMetadata)
 	return
 }
 
@@ -49,6 +52,12 @@ func (z *BucketOptions) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "Cached")
 				return
 			}
+		case "NoMetadata":
+			z.NoMetadata, bts, err = msgp.ReadBoolBytes(bts)
+			if err != nil {
+				err = msgp.WrapError(err, "NoMetadata")
+				return
+			}
 		default:
 			bts, err = msgp.Skip(bts)
 			if err != nil {
@@ -63,7 +72,7 @@ func (z *BucketOptions) UnmarshalMsg(bts []byte) (o []byte, err error) {
 
 // Msgsize returns an upper bound estimate of the number of bytes occupied by the serialized message
 func (z BucketOptions) Msgsize() (s int) {
-	s = 1 + 8 + msgp.BoolSize + 7 + msgp.BoolSize
+	s = 1 + 8 + msgp.BoolSize + 7 + msgp.BoolSize + 11 + msgp.BoolSize
 	return
 }
 

cmd/object-api-options.go

@@ -37,6 +37,15 @@ func getDefaultOpts(header http.Header, copySource bool, metadata map[string]str
 	var sse encrypt.ServerSide
 
 	opts = ObjectOptions{UserDefined: metadata}
+	if v, ok := header[xhttp.MinIOSourceProxyRequest]; ok {
+		opts.ProxyHeaderSet = true
+		opts.ProxyRequest = strings.Join(v, "") == "true"
+	}
+	if _, ok := header[xhttp.MinIOSourceReplicationRequest]; ok {
+		opts.ReplicationRequest = true
+	}
+	opts.Speedtest = header.Get(globalObjectPerfUserMetadata) != ""
+
 	if copySource {
 		if crypto.SSECopy.IsRequested(header) {
 			clientKey, err = crypto.SSECopy.ParseHTTP(header)
@@ -66,14 +75,7 @@ func getDefaultOpts(header http.Header, copySource bool, metadata map[string]str
 	if crypto.S3.IsRequested(header) || (metadata != nil && crypto.S3.IsEncrypted(metadata)) {
 		opts.ServerSideEncryption = encrypt.NewSSE()
 	}
-	if v, ok := header[xhttp.MinIOSourceProxyRequest]; ok {
+
-		opts.ProxyHeaderSet = true
-		opts.ProxyRequest = strings.Join(v, "") == "true"
-	}
-	if _, ok := header[xhttp.MinIOSourceReplicationRequest]; ok {
-		opts.ReplicationRequest = true
-	}
-	opts.Speedtest = header.Get(globalObjectPerfUserMetadata) != ""
 	return
 }
 
@@ -494,5 +496,8 @@ func completeMultipartOpts(ctx context.Context, r *http.Request, bucket, object
 	if _, ok := r.Header[xhttp.MinIOSourceReplicationRequest]; ok {
 		opts.ReplicationRequest = true
 	}
+	if r.Header.Get(ReplicationSsecChecksumHeader) != "" {
+		opts.UserDefined[ReplicationSsecChecksumHeader] = r.Header.Get(ReplicationSsecChecksumHeader)
+	}
 	return opts, nil
 }

cmd/object-api-utils.go

@@ -80,6 +80,18 @@ const (
 	compMinIndexSize = 8 << 20
 )
 
+// getkeyeparator - returns the separator to be used for
+// persisting on drive.
+//
+// - ":" is used on non-windows platforms
+// - "_" is used on windows platforms
+func getKeySeparator() string {
+	if runtime.GOOS == globalWindowsOSName {
+		return "_"
+	}
+	return ":"
+}
+
 // isMinioBucket returns true if given bucket is a MinIO internal
 // bucket and false otherwise.
 func isMinioMetaBucketName(bucket string) bool {
@@ -755,7 +767,7 @@ type ObjReaderFn func(inputReader io.Reader, h http.Header, cleanupFns ...func()
 // are called on Close() in FIFO order as passed in ObjReadFn(). NOTE: It is
 // assumed that clean up functions do not panic (otherwise, they may
 // not all run!).
-func NewGetObjectReader(rs *HTTPRangeSpec, oi ObjectInfo, opts ObjectOptions) (
+func NewGetObjectReader(rs *HTTPRangeSpec, oi ObjectInfo, opts ObjectOptions, h http.Header) (
 	fn ObjReaderFn, off, length int64, err error,
 ) {
 	if opts.CheckPrecondFn != nil && opts.CheckPrecondFn(oi) {
@@ -810,7 +822,9 @@ func NewGetObjectReader(rs *HTTPRangeSpec, oi ObjectInfo, opts ObjectOptions) (
 				return b, nil
 			}
 			if isEncrypted {
-				decrypt = oi.compressionIndexDecrypt
+				decrypt = func(b []byte) ([]byte, error) {
+					return oi.compressionIndexDecrypt(b, h)
+				}
 			}
 			// In case of range based queries on multiparts, the offset and length are reduced.
 			off, decOff, firstPart, decryptSkip, seqNum = getCompressedOffsets(oi, off, decrypt)
@@ -982,8 +996,8 @@ func compressionIndexEncrypter(key crypto.ObjectKey, input func() []byte) func()
 }
 
 // compressionIndexDecrypt reverses compressionIndexEncrypter.
-func (o *ObjectInfo) compressionIndexDecrypt(input []byte) ([]byte, error) {
+func (o *ObjectInfo) compressionIndexDecrypt(input []byte, h http.Header) ([]byte, error) {
-	return o.metadataDecrypter()("compression-index", input)
+	return o.metadataDecrypter(h)("compression-index", input)
 }
 
 // SealMD5CurrFn seals md5sum with object encryption key and returns sealed
@@ -1206,7 +1220,20 @@ func hasSpaceFor(di []*DiskInfo, size int64) (bool, error) {
 	}
 
 	if nDisks < len(di)/2 || nDisks <= 0 {
-		return false, fmt.Errorf("not enough online disks to calculate the available space, expected (%d)/(%d)", (len(di)/2)+1, nDisks)
+		var errs []error
+		for index, disk := range di {
+			switch {
+			case disk == nil:
+				errs = append(errs, fmt.Errorf("disk[%d]: offline", index))
+			case disk.Error != "":
+				errs = append(errs, fmt.Errorf("disk %s: %s", disk.Endpoint, disk.Error))
+			case disk.Total == 0:
+				errs = append(errs, fmt.Errorf("disk %s: total is zero", disk.Endpoint))
+			}
+		}
+		// Log disk errors.
+		peersLogIf(context.Background(), errors.Join(errs...))
+		return false, fmt.Errorf("not enough online disks to calculate the available space, need %d, found %d", (len(di)/2)+1, nDisks)
 	}
 
 	// Check we have enough on each disk, ignoring diskFillFraction.

cmd/object-handlers-common.go

@@ -335,7 +335,7 @@ func isETagEqual(left, right string) bool {
 // setPutObjHeaders sets all the necessary headers returned back
 // upon a success Put/Copy/CompleteMultipart/Delete requests
 // to activate delete only headers set delete as true
-func setPutObjHeaders(w http.ResponseWriter, objInfo ObjectInfo, delete bool) {
+func setPutObjHeaders(w http.ResponseWriter, objInfo ObjectInfo, delete bool, h http.Header) {
 	// We must not use the http.Header().Set method here because some (broken)
 	// clients expect the ETag header key to be literally "ETag" - not "Etag" (case-sensitive).
 	// Therefore, we have to set the ETag directly as map entry.
@@ -357,7 +357,7 @@ func setPutObjHeaders(w http.ResponseWriter, objInfo ObjectInfo, delete bool) {
 			lc.SetPredictionHeaders(w, objInfo.ToLifecycleOpts())
 		}
 	}
-	hash.AddChecksumHeader(w, objInfo.decryptChecksums(0))
+	hash.AddChecksumHeader(w, objInfo.decryptChecksums(0, h))
 }
 
 func deleteObjectVersions(ctx context.Context, o ObjectLayer, bucket string, toDel []ObjectToDelete, lcEvent lifecycle.Event) {

cmd/object-handlers.go

@@ -384,6 +384,10 @@ func (api objectAPIHandlers) getObjectHandler(ctx context.Context, objectAPI Obj
 	}
 
 	cachedResult := globalCacheConfig.Enabled() && opts.VersionID == ""
+	if _, ok := crypto.IsRequested(r.Header); ok {
+		// No need to check cache for encrypted objects.
+		cachedResult = false
+	}
 
 	var update bool
 	if cachedResult {
@@ -611,40 +615,7 @@ func (api objectAPIHandlers) getObjectHandler(ctx context.Context, objectAPI Obj
 
 	if r.Header.Get(xhttp.AmzChecksumMode) == "ENABLED" && rs == nil {
 		// AWS S3 silently drops checksums on range requests.
-		hash.AddChecksumHeader(w, objInfo.decryptChecksums(opts.PartNumber))
+		hash.AddChecksumHeader(w, objInfo.decryptChecksums(opts.PartNumber, r.Header))
-	}
-
-	var buf *bytebufferpool.ByteBuffer
-	if update {
-		if globalCacheConfig.MatchesSize(objInfo.Size) {
-			buf = bytebufferpool.Get()
-			defer bytebufferpool.Put(buf)
-		}
-		defer func() {
-			var data []byte
-			if buf != nil {
-				data = buf.Bytes()
-			}
-
-			asize, err := objInfo.GetActualSize()
-			if err != nil {
-				asize = objInfo.Size
-			}
-
-			globalCacheConfig.Set(&cache.ObjectInfo{
-				Key:          objInfo.Name,
-				Bucket:       objInfo.Bucket,
-				ETag:         objInfo.ETag,
-				ModTime:      objInfo.ModTime,
-				Expires:      objInfo.ExpiresStr(),
-				CacheControl: objInfo.CacheControl,
-				Metadata:     cleanReservedKeys(objInfo.UserDefined),
-				Range:        rangeHeader,
-				PartNumber:   opts.PartNumber,
-				Size:         asize,
-				Data:         data,
-			})
-		}()
 	}
 
 	if err = setObjectHeaders(ctx, w, objInfo, rs, opts); err != nil {
@@ -659,6 +630,12 @@ func (api objectAPIHandlers) getObjectHandler(ctx context.Context, objectAPI Obj
 
 	setHeadGetRespHeaders(w, r.Form)
 
+	var buf *bytebufferpool.ByteBuffer
+	if update && globalCacheConfig.MatchesSize(objInfo.Size) {
+		buf = bytebufferpool.Get()
+		defer bytebufferpool.Put(buf)
+	}
+
 	var iw io.Writer
 	iw = w
 	if buf != nil {
@@ -690,6 +667,31 @@ func (api objectAPIHandlers) getObjectHandler(ctx context.Context, objectAPI Obj
 		return
 	}
 
+	defer func() {
+		var data []byte
+		if buf != nil {
+			data = buf.Bytes()
+		}
+
+		if len(data) == 0 {
+			return
+		}
+
+		globalCacheConfig.Set(&cache.ObjectInfo{
+			Key:          objInfo.Name,
+			Bucket:       objInfo.Bucket,
+			ETag:         objInfo.ETag,
+			ModTime:      objInfo.ModTime,
+			Expires:      objInfo.ExpiresStr(),
+			CacheControl: objInfo.CacheControl,
+			Metadata:     cleanReservedKeys(objInfo.UserDefined),
+			Range:        rangeHeader,
+			PartNumber:   opts.PartNumber,
+			Size:         int64(len(data)),
+			Data:         data,
+		})
+	}()
+
 	// Notify object accessed via a GET request.
 	sendEvent(eventArgs{
 		EventName:    event.ObjectAccessedGet,
@@ -758,7 +760,7 @@ func (api objectAPIHandlers) getObjectAttributesHandler(ctx context.Context, obj
 	w.Header().Del(xhttp.ContentType)
 
 	if _, ok := opts.ObjectAttributes[xhttp.Checksum]; ok {
-		chkSums := objInfo.decryptChecksums(0)
+		chkSums := objInfo.decryptChecksums(0, r.Header)
 		// AWS does not appear to append part number on this API call.
 		switch {
 		case chkSums["CRC32"] != "":
@@ -789,7 +791,7 @@ func (api objectAPIHandlers) getObjectAttributesHandler(ctx context.Context, obj
 		OA.StorageClass = filterStorageClass(ctx, objInfo.StorageClass)
 	}
 
-	objInfo.decryptPartsChecksums()
+	objInfo.decryptPartsChecksums(r.Header)
 
 	if _, ok := opts.ObjectAttributes[xhttp.ObjectParts]; ok {
 		OA.ObjectParts = new(objectAttributesParts)
@@ -949,8 +951,10 @@ func (api objectAPIHandlers) headObjectHandler(ctx context.Context, objectAPI Ob
 	}
 
 	cachedResult := globalCacheConfig.Enabled() && opts.VersionID == ""
-
+	if _, ok := crypto.IsRequested(r.Header); ok {
-	var update bool
+		// No need to check cache for encrypted objects.
+		cachedResult = false
+	}
 	if cachedResult {
 		rc := &cache.CondCheck{}
 		h := r.Header.Clone()
@@ -959,7 +963,7 @@ func (api objectAPIHandlers) headObjectHandler(ctx context.Context, objectAPI Ob
 		}
 		rc.Init(bucket, object, h)
 
-		ci, err := globalCacheConfig.Get(rc)
+		ci, _ := globalCacheConfig.Get(rc)
 		if ci != nil {
 			tgs, ok := ci.Metadata[xhttp.AmzObjectTagging]
 			if ok {
@@ -1018,9 +1022,6 @@ func (api objectAPIHandlers) headObjectHandler(ctx context.Context, objectAPI Ob
 				return
 			}
 		}
-		if errors.Is(err, cache.ErrKeyMissing) {
-			update = true
-		}
 	}
 
 	opts.FastGetObjInfo = true
@@ -1124,24 +1125,6 @@ func (api objectAPIHandlers) headObjectHandler(ctx context.Context, objectAPI Ob
 		return
 	}
 
-	if update {
-		asize, err := objInfo.GetActualSize()
-		if err != nil {
-			asize = objInfo.Size
-		}
-
-		defer globalCacheConfig.Set(&cache.ObjectInfo{
-			Key:          objInfo.Name,
-			Bucket:       objInfo.Bucket,
-			ETag:         objInfo.ETag,
-			ModTime:      objInfo.ModTime,
-			Expires:      objInfo.ExpiresStr(),
-			CacheControl: objInfo.CacheControl,
-			Size:         asize,
-			Metadata:     cleanReservedKeys(objInfo.UserDefined),
-		})
-	}
-
 	// Validate pre-conditions if any.
 	if checkPreconditions(ctx, w, r, objInfo, opts) {
 		return
@@ -1169,7 +1152,7 @@ func (api objectAPIHandlers) headObjectHandler(ctx context.Context, objectAPI Ob
 
 	if r.Header.Get(xhttp.AmzChecksumMode) == "ENABLED" && rs == nil {
 		// AWS S3 silently drops checksums on range requests.
-		hash.AddChecksumHeader(w, objInfo.decryptChecksums(opts.PartNumber))
+		hash.AddChecksumHeader(w, objInfo.decryptChecksums(opts.PartNumber, r.Header))
 	}
 
 	// Set standard object headers.
@@ -1472,6 +1455,7 @@ func (api objectAPIHandlers) CopyObjectHandler(w http.ResponseWriter, r *http.Re
 		VersionID:          srcOpts.VersionID,
 		Versioned:          srcOpts.Versioned,
 		VersionSuspended:   srcOpts.VersionSuspended,
+		ReplicationRequest: r.Header.Get(xhttp.MinIOSourceReplicationRequest) == "true",
 	}
 	getSSE := encrypt.SSE(srcOpts.ServerSideEncryption)
 	if getSSE != srcOpts.ServerSideEncryption {
@@ -1567,7 +1551,7 @@ func (api objectAPIHandlers) CopyObjectHandler(w http.ResponseWriter, r *http.Re
 	// Pass the decompressed stream to such calls.
 	isDstCompressed := isCompressible(r.Header, dstObject) &&
 		length > minCompressibleSize &&
-		!isRemoteCopyRequired(ctx, srcBucket, dstBucket, objectAPI) && !cpSrcDstSame && !objectEncryption
+		!isRemoteCopyRequired(ctx, srcBucket, dstBucket, objectAPI) && !objectEncryption
 	if isDstCompressed {
 		compressMetadata = make(map[string]string, 2)
 		// Preserving the compression metadata.
@@ -1603,7 +1587,7 @@ func (api objectAPIHandlers) CopyObjectHandler(w http.ResponseWriter, r *http.Re
 		return
 	}
 	// Encryption parameters not present for this object.
-	if crypto.SSEC.IsEncrypted(srcInfo.UserDefined) && !crypto.SSECopy.IsRequested(r.Header) {
+	if crypto.SSEC.IsEncrypted(srcInfo.UserDefined) && !crypto.SSECopy.IsRequested(r.Header) && r.Header.Get(xhttp.MinIOSourceReplicationRequest) != "true" {
 		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrInvalidSSECustomerAlgorithm), r.URL)
 		return
 	}
@@ -1929,7 +1913,7 @@ func (api objectAPIHandlers) CopyObjectHandler(w http.ResponseWriter, r *http.Re
 		scheduleReplication(ctx, objInfo, objectAPI, dsc, replication.ObjectReplicationType)
 	}
 
-	setPutObjHeaders(w, objInfo, false)
+	setPutObjHeaders(w, objInfo, false, r.Header)
 	// We must not use the http.Header().Set method here because some (broken)
 	// clients expect the x-amz-copy-source-version-id header key to be literally
 	// "x-amz-copy-source-version-id"- not in canonicalized form, preserve it.
@@ -1951,22 +1935,6 @@ func (api objectAPIHandlers) CopyObjectHandler(w http.ResponseWriter, r *http.Re
 		Host:         handlers.GetSourceIP(r),
 	})
 
-	asize, err := objInfo.GetActualSize()
-	if err != nil {
-		asize = objInfo.Size
-	}
-
-	defer globalCacheConfig.Set(&cache.ObjectInfo{
-		Key:          objInfo.Name,
-		Bucket:       objInfo.Bucket,
-		ETag:         objInfo.ETag,
-		ModTime:      objInfo.ModTime,
-		Expires:      objInfo.ExpiresStr(),
-		CacheControl: objInfo.CacheControl,
-		Size:         asize,
-		Metadata:     cleanReservedKeys(objInfo.UserDefined),
-	})
-
 	if !remoteCallRequired && !globalTierConfigMgr.Empty() {
 		// Schedule object for immediate transition if eligible.
 		objInfo.ETag = origETag
@@ -2139,7 +2107,8 @@ func (api objectAPIHandlers) PutObjectHandler(w http.ResponseWriter, r *http.Req
 	})
 
 	var buf *bytebufferpool.ByteBuffer
-	if globalCacheConfig.MatchesSize(size) {
+	// Disable cache for encrypted objects - headers are applied with sseConfig.Apply if auto encrypted.
+	if globalCacheConfig.MatchesSize(size) && !crypto.Requested(r.Header) {
 		buf = bytebufferpool.Get()
 		defer bytebufferpool.Put(buf)
 	}
@@ -2262,11 +2231,6 @@ func (api objectAPIHandlers) PutObjectHandler(w http.ResponseWriter, r *http.Req
 			return
 		}
 
-		if crypto.SSEC.IsRequested(r.Header) && isCompressible(r.Header, object) {
-			writeErrorResponse(ctx, w, toAPIError(ctx, crypto.ErrIncompatibleEncryptionWithCompression), r.URL)
-			return
-		}
-
 		reader, objectEncryptionKey, err = EncryptRequest(hashReader, r, bucket, object, metadata)
 		if err != nil {
 			writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
@@ -2351,7 +2315,7 @@ func (api objectAPIHandlers) PutObjectHandler(w http.ResponseWriter, r *http.Req
 		scheduleReplication(ctx, objInfo, objectAPI, dsc, replication.ObjectReplicationType)
 	}
 
-	setPutObjHeaders(w, objInfo, false)
+	setPutObjHeaders(w, objInfo, false, r.Header)
 
 	defer func() {
 		var data []byte
@@ -2359,9 +2323,8 @@ func (api objectAPIHandlers) PutObjectHandler(w http.ResponseWriter, r *http.Req
 			data = buf.Bytes()
 		}
 
-		asize, err := objInfo.GetActualSize()
+		if len(data) == 0 {
-		if err != nil {
+			return
-			asize = objInfo.Size
 		}
 
 		globalCacheConfig.Set(&cache.ObjectInfo{
@@ -2371,7 +2334,7 @@ func (api objectAPIHandlers) PutObjectHandler(w http.ResponseWriter, r *http.Req
 			ModTime:      objInfo.ModTime,
 			Expires:      objInfo.ExpiresStr(),
 			CacheControl: objInfo.CacheControl,
-			Size:         asize,
+			Size:         int64(len(data)),
 			Metadata:     cleanReservedKeys(objInfo.UserDefined),
 			Data:         data,
 		})
@@ -2728,22 +2691,6 @@ func (api objectAPIHandlers) PutObjectExtractHandler(w http.ResponseWriter, r *h
 			scheduleReplication(ctx, objInfo, objectAPI, dsc, replication.ObjectReplicationType)
 		}
 
-		asize, err := objInfo.GetActualSize()
-		if err != nil {
-			asize = objInfo.Size
-		}
-
-		defer globalCacheConfig.Set(&cache.ObjectInfo{
-			Key:          objInfo.Name,
-			Bucket:       objInfo.Bucket,
-			ETag:         objInfo.ETag,
-			ModTime:      objInfo.ModTime,
-			Expires:      objInfo.ExpiresStr(),
-			CacheControl: objInfo.CacheControl,
-			Size:         asize,
-			Metadata:     cleanReservedKeys(objInfo.UserDefined),
-		})
-
 		// Notify object created event.
 		evt := eventArgs{
 			EventName:    event.ObjectCreatedPut,
@@ -2943,7 +2890,7 @@ func (api objectAPIHandlers) DeleteObjectHandler(w http.ResponseWriter, r *http.
 
 	defer globalCacheConfig.Delete(bucket, object)
 
-	setPutObjHeaders(w, objInfo, true)
+	setPutObjHeaders(w, objInfo, true, r.Header)
 	writeSuccessNoContent(w)
 
 	eventName := event.ObjectRemovedDelete

cmd/object-handlers_test.go

@@ -61,7 +61,7 @@ const (
 
 // Wrapper for calling HeadObject API handler tests for both Erasure multiple disks and FS single drive setup.
 func TestAPIHeadObjectHandler(t *testing.T) {
-	ExecObjectLayerAPITest(t, testAPIHeadObjectHandler, []string{"HeadObject"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPIHeadObjectHandler, endpoints: []string{"HeadObject"}})
 }
 
 func testAPIHeadObjectHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -207,7 +207,7 @@ func TestAPIHeadObjectHandlerWithEncryption(t *testing.T) {
 	defer func() { globalPolicySys = nil }()
 
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPIHeadObjectHandlerWithEncryption, []string{"NewMultipart", "PutObjectPart", "CompleteMultipart", "GetObject", "PutObject", "HeadObject"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPIHeadObjectHandlerWithEncryption, endpoints: []string{"NewMultipart", "PutObjectPart", "CompleteMultipart", "GetObject", "PutObject", "HeadObject"}})
 }
 
 func testAPIHeadObjectHandlerWithEncryption(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -707,7 +707,13 @@ func testAPIGetObjectWithMPHandler(obj ObjectLayer, instanceType, bucketName str
 		{"enc-mp-2", []int64{5487701, 5487799, 3}, mapCopy(metaWithSSEC)},
 		{"enc-mp-3", []int64{10499807, 10499963, 7}, mapCopy(metaWithSSEC)},
 	}
-
+	// SSEC can't be used with compression
+	globalCompressConfigMu.Lock()
+	globalCompressEnabled := globalCompressConfig.Enabled
+	globalCompressConfigMu.Unlock()
+	if globalCompressEnabled {
+		objectInputs = objectInputs[0:8]
+	}
 	// iterate through the above set of inputs and upload the object.
 	for _, input := range objectInputs {
 		uploadTestObject(t, apiRouter, credentials, bucketName, input.objectName, input.partLengths, input.metaData, false)
@@ -898,6 +904,14 @@ func testAPIGetObjectWithPartNumberHandler(obj ObjectLayer, instanceType, bucket
 		{"enc-mp-3", []int64{10499807, 10499963, 7}, mapCopy(metaWithSSEC)},
 	}
 
+	// SSEC can't be used with compression
+	globalCompressConfigMu.Lock()
+	globalCompressEnabled := globalCompressConfig.Enabled
+	globalCompressConfigMu.Unlock()
+	if globalCompressEnabled {
+		objectInputs = objectInputs[0:9]
+	}
+
 	// iterate through the above set of inputs and upload the object.
 	for _, input := range objectInputs {
 		uploadTestObject(t, apiRouter, credentials, bucketName, input.objectName, input.partLengths, input.metaData, false)
@@ -2119,7 +2133,7 @@ func testAPICopyObjectPartHandler(obj ObjectLayer, instanceType, bucketName stri
 // Wrapper for calling Copy Object API handler tests for both Erasure multiple disks and single node setup.
 func TestAPICopyObjectHandler(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecExtendedObjectLayerAPITest(t, testAPICopyObjectHandler, []string{"CopyObject"})
+	ExecExtendedObjectLayerAPITest(t, testAPICopyObjectHandler, []string{"CopyObject", "PutObject"})
 }
 
 func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -2150,7 +2164,10 @@ func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string,
 		new(bytes.Buffer),
 		new(bytes.Buffer),
 	}
-
+	bucketInfo, err := obj.GetBucketInfo(context.Background(), bucketName, BucketOptions{})
+	if err != nil {
+		t.Fatalf("Test -1: %s: Failed to get bucket info: <ERROR> %s", instanceType, err)
+	}
 	// set of inputs for uploading the objects before tests for downloading is done.
 	putObjectInputs := []struct {
 		bucketName    string
@@ -2167,21 +2184,29 @@ func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string,
 		// used for anonymous HTTP request test.
 		{bucketName, anonObject, int64(len(bytesData[0].byteData)), bytesData[0].byteData, bytesData[0].md5sum, make(map[string]string)},
 	}
-
 	// iterate through the above set of inputs and upload the object.
 	for i, input := range putObjectInputs {
-		// uploading the object.
+		rec := httptest.NewRecorder()
-		var objInfo ObjectInfo
+		req, err := newTestSignedRequestV4(http.MethodPut, getPutObjectURL("", input.bucketName, input.objectName),
-		objInfo, err = obj.PutObject(context.Background(), input.bucketName, input.objectName, mustGetPutObjReader(t, bytes.NewReader(input.textData), input.contentLength, input.md5sum, ""), ObjectOptions{UserDefined: input.metaData})
+			input.contentLength, bytes.NewReader(input.textData), credentials.AccessKey, credentials.SecretKey, nil)
-		// if object upload fails stop the test.
 		if err != nil {
-			t.Fatalf("Put Object case %d:  Error uploading object: <ERROR> %v", i+1, err)
+			t.Fatalf("Test %d: Failed to create HTTP request for Put Object: <ERROR> %v", i, err)
 		}
-		if objInfo.ETag != input.md5sum {
+		apiRouter.ServeHTTP(rec, req)
-			t.Fatalf("Put Object case %d:  Checksum mismatched: <ERROR> got %s, expected %s", i+1, input.md5sum, objInfo.ETag)
+		if rec.Code != http.StatusOK {
+			b, err := io.ReadAll(rec.Body)
+			if err != nil {
+				t.Fatal(err)
+			}
+			var apiErr APIErrorResponse
+			err = xml.Unmarshal(b, &apiErr)
+			if err != nil {
+				t.Fatal(err)
+			}
+			gotErr := apiErr.Code
+			t.Errorf("test %d: want api got %q", i, gotErr)
 		}
 	}
-
 	// test cases with inputs and expected result for Copy Object.
 	testCases := []struct {
 		bucketName           string
@@ -2434,6 +2459,22 @@ func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string,
 	}
 
 	for i, testCase := range testCases {
+		if bucketInfo.Versioning {
+			if strings.Contains(testCase.copySourceHeader, "versionId=null") {
+				testCase.expectedRespStatus = http.StatusNotFound
+			}
+		}
+		values := url.Values{}
+		if testCase.expectedRespStatus == http.StatusOK {
+			r, err := obj.GetObjectNInfo(context.Background(), testCase.bucketName, objectName, nil, nil, opts)
+			if err != nil {
+				t.Fatalf("Test %d: %s reading completed file failed: <ERROR> %v", i, instanceType, err)
+			}
+			r.Close()
+			if r.ObjInfo.VersionID != "" {
+				values.Set(xhttp.VersionID, r.ObjInfo.VersionID)
+			}
+		}
 		var req *http.Request
 		var reqV2 *http.Request
 		// initialize HTTP NewRecorder, this records any mutations to response writer inside the handler.
@@ -2446,8 +2487,12 @@ func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string,
 		}
 		// "X-Amz-Copy-Source" header contains the information about the source bucket and the object to copied.
 		if testCase.copySourceHeader != "" {
+			if values.Encode() != "" && !strings.Contains(testCase.copySourceHeader, "?") {
+				req.Header.Set("X-Amz-Copy-Source", testCase.copySourceHeader+"?"+values.Encode())
+			} else {
 				req.Header.Set("X-Amz-Copy-Source", testCase.copySourceHeader)
 			}
+		}
 		if testCase.copyModifiedHeader != "" {
 			req.Header.Set("X-Amz-Copy-Source-If-Modified-Since", testCase.copyModifiedHeader)
 		}
@@ -2552,10 +2597,10 @@ func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string,
 			if testCase.copySourceSame {
 				// encryption will rotate creds, so fail only for non-encryption scenario.
 				if GlobalKMS == nil {
-					t.Errorf("Test %d: %s:  Expected the response status to be `%d`, but instead found `%d`", i, instanceType, testCase.expectedRespStatus, rec.Code)
+					t.Errorf("Test %d: %s:  Expected the response status to be `%d`, but instead found `%d`", i, instanceType, testCase.expectedRespStatus, recV2.Code)
 				}
 			} else {
-				t.Errorf("Test %d: %s:  Expected the response status to be `%d`, but instead found `%d`", i+1, instanceType, testCase.expectedRespStatus, rec.Code)
+				t.Errorf("Test %d: %s:  Expected the response status to be `%d`, but instead found `%d`", i+1, instanceType, testCase.expectedRespStatus, recV2.Code)
 			}
 		}
 	}
@@ -2588,7 +2633,7 @@ func testAPICopyObjectHandler(obj ObjectLayer, instanceType, bucketName string,
 // The UploadID from the response body is parsed and its existence is asserted with an attempt to ListParts using it.
 func TestAPINewMultipartHandler(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPINewMultipartHandler, []string{"NewMultipart"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPINewMultipartHandler, endpoints: []string{"NewMultipart"}})
 }
 
 func testAPINewMultipartHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -2727,7 +2772,7 @@ func testAPINewMultipartHandler(obj ObjectLayer, instanceType, bucketName string
 // The UploadID from the response body is parsed and its existence is asserted with an attempt to ListParts using it.
 func TestAPINewMultipartHandlerParallel(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPINewMultipartHandlerParallel, []string{"NewMultipart"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPINewMultipartHandlerParallel, endpoints: []string{"NewMultipart"}})
 }
 
 func testAPINewMultipartHandlerParallel(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -2790,7 +2835,7 @@ func testAPINewMultipartHandlerParallel(obj ObjectLayer, instanceType, bucketNam
 // The UploadID from the response body is parsed and its existence is asserted with an attempt to ListParts using it.
 func TestAPICompleteMultipartHandler(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPICompleteMultipartHandler, []string{"CompleteMultipart"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPICompleteMultipartHandler, endpoints: []string{"CompleteMultipart"}})
 }
 
 func testAPICompleteMultipartHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -2914,7 +2959,7 @@ func testAPICompleteMultipartHandler(obj ObjectLayer, instanceType, bucketName s
 	s3MD5 := getCompleteMultipartMD5(inputParts[3].parts)
 
 	// generating the response body content for the success case.
-	successResponse := generateCompleteMultpartUploadResponse(bucketName, objectName, getGetObjectURL("", bucketName, objectName), ObjectInfo{ETag: s3MD5})
+	successResponse := generateCompleteMultipartUploadResponse(bucketName, objectName, getGetObjectURL("", bucketName, objectName), ObjectInfo{ETag: s3MD5}, nil)
 	encodedSuccessResponse := encodeResponse(successResponse)
 
 	ctx := context.Background()
@@ -3159,7 +3204,7 @@ func testAPICompleteMultipartHandler(obj ObjectLayer, instanceType, bucketName s
 // The UploadID from the response body is parsed and its existence is asserted with an attempt to ListParts using it.
 func TestAPIAbortMultipartHandler(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPIAbortMultipartHandler, []string{"AbortMultipart"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPIAbortMultipartHandler, endpoints: []string{"AbortMultipart"}})
 }
 
 func testAPIAbortMultipartHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -3319,7 +3364,7 @@ func testAPIAbortMultipartHandler(obj ObjectLayer, instanceType, bucketName stri
 // Wrapper for calling Delete Object API handler tests for both Erasure multiple disks and FS single drive setup.
 func TestAPIDeleteObjectHandler(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPIDeleteObjectHandler, []string{"DeleteObject"})
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPIDeleteObjectHandler, endpoints: []string{"DeleteObject"}})
 }
 
 func testAPIDeleteObjectHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
@@ -3881,8 +3926,7 @@ func testAPIPutObjectPartHandler(obj ObjectLayer, instanceType, bucketName strin
 //	when signature type of the HTTP request is `Presigned`.
 func TestAPIListObjectPartsHandlerPreSign(t *testing.T) {
 	defer DetectTestLeak(t)()
-	ExecObjectLayerAPITest(t, testAPIListObjectPartsHandlerPreSign,
+	ExecObjectLayerAPITest(ExecObjectLayerAPITestArgs{t: t, objAPITest: testAPIListObjectPartsHandlerPreSign, endpoints: []string{"PutObjectPart", "NewMultipart", "ListObjectParts"}})
-		[]string{"PutObjectPart", "NewMultipart", "ListObjectParts"})
 }
 
 func testAPIListObjectPartsHandlerPreSign(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,

cmd/object-multipart-handlers.go

@@ -36,7 +36,6 @@ import (
 	sse "github.com/minio/minio/internal/bucket/encryption"
 	objectlock "github.com/minio/minio/internal/bucket/object/lock"
 	"github.com/minio/minio/internal/bucket/replication"
-	"github.com/minio/minio/internal/config/cache"
 	"github.com/minio/minio/internal/config/dns"
 	"github.com/minio/minio/internal/config/storageclass"
 	"github.com/minio/minio/internal/crypto"
@@ -116,11 +115,6 @@ func (api objectAPIHandlers) NewMultipartUploadHandler(w http.ResponseWriter, r
 			return
 		}
 
-		if crypto.SSEC.IsRequested(r.Header) && isCompressible(r.Header, object) {
-			writeErrorResponse(ctx, w, toAPIError(ctx, crypto.ErrIncompatibleEncryptionWithCompression), r.URL)
-			return
-		}
-
 		_, sourceReplReq := r.Header[xhttp.MinIOSourceReplicationRequest]
 		ssecRepHeaders := []string{
 			"X-Minio-Replication-Server-Side-Encryption-Seal-Algorithm",
@@ -773,6 +767,7 @@ func (api objectAPIHandlers) PutObjectPartHandler(w http.ResponseWriter, r *http
 
 	_, isEncrypted := crypto.IsEncrypted(mi.UserDefined)
 	_, replicationStatus := mi.UserDefined[xhttp.AmzBucketReplicationStatus]
+	_, sourceReplReq := r.Header[xhttp.MinIOSourceReplicationRequest]
 	var objectEncryptionKey crypto.ObjectKey
 	if isEncrypted {
 		if !crypto.SSEC.IsRequested(r.Header) && crypto.SSEC.IsEncrypted(mi.UserDefined) && !replicationStatus {
@@ -795,7 +790,6 @@ func (api objectAPIHandlers) PutObjectPartHandler(w http.ResponseWriter, r *http
 			}
 		}
 
-		_, sourceReplReq := r.Header[xhttp.MinIOSourceReplicationRequest]
 		if !(sourceReplReq && crypto.SSEC.IsEncrypted(mi.UserDefined)) {
 			// Calculating object encryption key
 			key, err = decryptObjectMeta(key, bucket, object, mi.UserDefined)
@@ -847,6 +841,7 @@ func (api objectAPIHandlers) PutObjectPartHandler(w http.ResponseWriter, r *http
 	}
 	opts.IndexCB = idxCb
 
+	opts.ReplicationRequest = sourceReplReq
 	putObjectPart := objectAPI.PutObjectPart
 
 	partInfo, err := putObjectPart(ctx, bucket, object, uploadID, partID, pReader, opts)
@@ -1028,7 +1023,7 @@ func (api objectAPIHandlers) CompleteMultipartUploadHandler(w http.ResponseWrite
 		}
 	}
 
-	setPutObjHeaders(w, objInfo, false)
+	setPutObjHeaders(w, objInfo, false, r.Header)
 	if dsc := mustReplicate(ctx, bucket, object, objInfo.getMustReplicateOptions(replication.ObjectReplicationType, opts)); dsc.ReplicateAny() {
 		scheduleReplication(ctx, objInfo, objectAPI, dsc, replication.ObjectReplicationType)
 	}
@@ -1040,7 +1035,7 @@ func (api objectAPIHandlers) CompleteMultipartUploadHandler(w http.ResponseWrite
 	// Get object location.
 	location := getObjectLocation(r, globalDomainNames, bucket, object)
 	// Generate complete multipart response.
-	response := generateCompleteMultpartUploadResponse(bucket, object, location, objInfo)
+	response := generateCompleteMultipartUploadResponse(bucket, object, location, objInfo, r.Header)
 	encodedSuccessResponse := encodeResponse(response)
 
 	// Write success response.
@@ -1058,22 +1053,6 @@ func (api objectAPIHandlers) CompleteMultipartUploadHandler(w http.ResponseWrite
 	}
 	sendEvent(evt)
 
-	asize, err := objInfo.GetActualSize()
-	if err != nil {
-		asize = objInfo.Size
-	}
-
-	defer globalCacheConfig.Set(&cache.ObjectInfo{
-		Key:          objInfo.Name,
-		Bucket:       objInfo.Bucket,
-		ETag:         objInfo.ETag,
-		ModTime:      objInfo.ModTime,
-		Expires:      objInfo.ExpiresStr(),
-		CacheControl: objInfo.CacheControl,
-		Size:         asize,
-		Metadata:     cleanReservedKeys(objInfo.UserDefined),
-	})
-
 	if objInfo.NumVersions > int(scannerExcessObjectVersions.Load()) {
 		evt.EventName = event.ObjectManyVersions
 		sendEvent(evt)

cmd/object_api_suite_test.go

@@ -511,12 +511,12 @@ func testBucketRecreateFails(obj ObjectLayer, instanceType string, t TestErrHand
 	}
 }
 
-func enableCompression(t *testing.T, encrypt bool) {
+func enableCompression(t *testing.T, encrypt bool, mimeTypes []string, extensions []string) {
 	// Enable compression and exec...
 	globalCompressConfigMu.Lock()
 	globalCompressConfig.Enabled = true
-	globalCompressConfig.MimeTypes = nil
+	globalCompressConfig.MimeTypes = mimeTypes
-	globalCompressConfig.Extensions = nil
+	globalCompressConfig.Extensions = extensions
 	globalCompressConfig.AllowEncrypted = encrypt
 	globalCompressConfigMu.Unlock()
 	if encrypt {
@@ -553,11 +553,14 @@ func resetCompressEncryption() {
 	GlobalKMS = nil
 }
 
-func execExtended(t *testing.T, fn func(t *testing.T)) {
+func execExtended(t *testing.T, fn func(t *testing.T, init func(), bucketOptions MakeBucketOptions)) {
 	// Exec with default settings...
 	resetCompressEncryption()
 	t.Run("default", func(t *testing.T) {
-		fn(t)
+		fn(t, nil, MakeBucketOptions{})
+	})
+	t.Run("defaultVerioned", func(t *testing.T) {
+		fn(t, nil, MakeBucketOptions{VersioningEnabled: true})
 	})
 
 	if testing.Short() {
@@ -565,28 +568,55 @@ func execExtended(t *testing.T, fn func(t *testing.T)) {
 	}
 
 	t.Run("compressed", func(t *testing.T) {
+		fn(t, func() {
 			resetCompressEncryption()
-		enableCompression(t, false)
+			enableCompression(t, false, []string{"*"}, []string{"*"})
-		fn(t)
+		}, MakeBucketOptions{})
+	})
+	t.Run("compressedVerioned", func(t *testing.T) {
+		fn(t, func() {
+			resetCompressEncryption()
+			enableCompression(t, false, []string{"*"}, []string{"*"})
+		}, MakeBucketOptions{
+			VersioningEnabled: true,
+		})
 	})
 
 	t.Run("encrypted", func(t *testing.T) {
+		fn(t, func() {
 			resetCompressEncryption()
 			enableEncryption(t)
-		fn(t)
+		}, MakeBucketOptions{})
+	})
+	t.Run("encryptedVerioned", func(t *testing.T) {
+		fn(t, func() {
+			resetCompressEncryption()
+			enableEncryption(t)
+		}, MakeBucketOptions{
+			VersioningEnabled: true,
+		})
 	})
 
 	t.Run("compressed+encrypted", func(t *testing.T) {
+		fn(t, func() {
 			resetCompressEncryption()
-		enableCompression(t, true)
+			enableCompression(t, true, []string{"*"}, []string{"*"})
-		fn(t)
+		}, MakeBucketOptions{})
+	})
+	t.Run("compressed+encryptedVerioned", func(t *testing.T) {
+		fn(t, func() {
+			resetCompressEncryption()
+			enableCompression(t, true, []string{"*"}, []string{"*"})
+		}, MakeBucketOptions{
+			VersioningEnabled: true,
+		})
 	})
 }
 
 // ExecExtendedObjectLayerTest will execute the tests with combinations of encrypted & compressed.
 // This can be used to test functionality when reading and writing data.
 func ExecExtendedObjectLayerTest(t *testing.T, objTest objTestType) {
-	execExtended(t, func(t *testing.T) {
+	execExtended(t, func(t *testing.T, init func(), bucketOptions MakeBucketOptions) {
 		ExecObjectLayerTest(t, objTest)
 	})
 }

cmd/peer-rest-client.go

@@ -427,11 +427,14 @@ func (client *peerRESTClient) CommitBinary(ctx context.Context) error {
 }
 
 // SignalService - sends signal to peer nodes.
-func (client *peerRESTClient) SignalService(sig serviceSignal, subSys string, dryRun bool) error {
+func (client *peerRESTClient) SignalService(sig serviceSignal, subSys string, dryRun bool, execAt *time.Time) error {
 	values := grid.NewMSS()
 	values.Set(peerRESTSignal, strconv.Itoa(int(sig)))
 	values.Set(peerRESTDryRun, strconv.FormatBool(dryRun))
 	values.Set(peerRESTSubSys, subSys)
+	if execAt != nil {
+		values.Set(peerRESTExecAt, execAt.Format(time.RFC3339Nano))
+	}
 	_, err := signalServiceRPC.Call(context.Background(), client.gridConn(), values)
 	return err
 }

cmd/peer-rest-common.go

@@ -17,6 +17,8 @@
 
 package cmd
 
+import "time"
+
 const (
 	peerRESTVersion       = "v39" // add more flags to speedtest API
 	peerRESTVersionPrefix = SlashSeparator + peerRESTVersion
@@ -69,6 +71,7 @@ const (
 	peerRESTURL         = "url"
 	peerRESTSha256Sum   = "sha256sum"
 	peerRESTReleaseInfo = "releaseinfo"
+	peerRESTExecAt      = "exec-at"
 
 	peerRESTListenBucket = "bucket"
 	peerRESTListenPrefix = "prefix"
@@ -76,3 +79,5 @@ const (
 	peerRESTListenEvents = "events"
 	peerRESTLogMask      = "log-mask"
 )
+
+const restartUpdateDelay = 250 * time.Millisecond

cmd/peer-rest-server.go

@@ -693,6 +693,18 @@ func (s *peerRESTServer) SignalServiceHandler(vars *grid.MSS) (np grid.NoPayload
 	if err != nil {
 		return np, grid.NewRemoteErr(err)
 	}
+
+	// Wait until the specified time before executing the signal.
+	if t := vars.Get(peerRESTExecAt); t != "" {
+		execAt, err := time.Parse(time.RFC3339Nano, vars.Get(peerRESTExecAt))
+		if err != nil {
+			logger.LogIf(GlobalContext, "signalservice", err)
+			execAt = time.Now().Add(restartUpdateDelay)
+		}
+		if d := time.Until(execAt); d > 0 {
+			time.Sleep(d)
+		}
+	}
 	signal := serviceSignal(si)
 	switch signal {
 	case serviceRestart, serviceStop:

cmd/peer-s3-client.go

@@ -138,8 +138,12 @@ func (sys *S3PeerSys) HealBucket(ctx context.Context, bucket string, opts madmin
 		poolErrs = append(poolErrs, reduceWriteQuorumErrs(ctx, perPoolErrs, bucketOpIgnoredErrs, quorum))
 	}
 
+	if !opts.Recreate {
+		// when there is no force recreate look for pool
+		// errors to recreate the bucket on all pools.
 		opts.Remove = isAllBucketsNotFound(poolErrs)
 		opts.Recreate = !opts.Remove
+	}
 
 	g = errgroup.WithNErrs(len(sys.peerClients))
 	healBucketResults := make([]madmin.HealResultItem, len(sys.peerClients))

cmd/peer-s3-server.go

@@ -123,7 +123,7 @@ func healBucketLocal(ctx context.Context, bucket string, opts madmin.HealOpts) (
 		g.Wait()
 	}
 
-	// Create the quorum lost volume only if its nor makred for delete
+	// Create the lost volume only if its not marked for delete
 	if !opts.Remove {
 		// Initialize sync waitgroup.
 		g = errgroup.WithNErrs(len(localDrives))

cmd/s3-zip-handlers.go

@@ -142,7 +142,7 @@ func (api objectAPIHandlers) getObjectInArchiveFileHandler(ctx context.Context,
 		return
 	}
 
-	zipInfo := zipObjInfo.ArchiveInfo()
+	zipInfo := zipObjInfo.ArchiveInfo(r.Header)
 	if len(zipInfo) == 0 {
 		opts.EncryptFn, err = zipObjInfo.metadataEncryptFn(r.Header)
 		if err != nil {
@@ -233,7 +233,7 @@ func (api objectAPIHandlers) getObjectInArchiveFileHandler(ctx context.Context,
 }
 
 // listObjectsV2InArchive generates S3 listing result ListObjectsV2Info from zip file, all parameters are already validated by the caller.
-func listObjectsV2InArchive(ctx context.Context, objectAPI ObjectLayer, bucket, prefix, token, delimiter string, maxKeys int, fetchOwner bool, startAfter string) (ListObjectsV2Info, error) {
+func listObjectsV2InArchive(ctx context.Context, objectAPI ObjectLayer, bucket, prefix, token, delimiter string, maxKeys int, startAfter string, h http.Header) (ListObjectsV2Info, error) {
 	zipPath, _, err := splitZipExtensionPath(prefix)
 	if err != nil {
 		// Return empty listing
@@ -246,7 +246,7 @@ func listObjectsV2InArchive(ctx context.Context, objectAPI ObjectLayer, bucket,
 		return ListObjectsV2Info{}, nil
 	}
 
-	zipInfo := zipObjInfo.ArchiveInfo()
+	zipInfo := zipObjInfo.ArchiveInfo(h)
 	if len(zipInfo) == 0 {
 		// Always update the latest version
 		zipInfo, err = updateObjectMetadataWithZipInfo(ctx, objectAPI, bucket, zipPath, ObjectOptions{})
@@ -438,7 +438,7 @@ func (api objectAPIHandlers) headObjectInArchiveFileHandler(ctx context.Context,
 		return
 	}
 
-	zipInfo := zipObjInfo.ArchiveInfo()
+	zipInfo := zipObjInfo.ArchiveInfo(r.Header)
 	if len(zipInfo) == 0 {
 		opts.EncryptFn, err = zipObjInfo.metadataEncryptFn(r.Header)
 		if err != nil {

cmd/server-main.go

@@ -84,11 +84,12 @@ var ServerFlags = []cli.Flag{
 	},
 	cli.DurationFlag{
 		Name:   "shutdown-timeout",
-		Value:  xhttp.DefaultShutdownTimeout,
+		Value:  time.Second * 30,
-		Usage:  "shutdown timeout to gracefully shutdown server",
+		Usage:  "shutdown timeout to gracefully shutdown server (DEPRECATED)",
 		EnvVar: "MINIO_SHUTDOWN_TIMEOUT",
 		Hidden: true,
 	},
+
 	cli.DurationFlag{
 		Name:   "idle-timeout",
 		Value:  xhttp.DefaultIdleTimeout,
@@ -415,7 +416,8 @@ func serverHandleCmdArgs(ctxt serverCtxt) {
 
 	globalTCPOptions = xhttp.TCPOptions{
 		UserTimeout: int(ctxt.UserTimeout.Milliseconds()),
-		DriveOPTimeout: globalDriveConfig.GetOPTimeout,
+		// FIXME: Bring this back when we have valid way to handle deadlines
+		//		DriveOPTimeout: globalDriveConfig.GetOPTimeout,
 		Interface:   ctxt.Interface,
 		SendBufSize: ctxt.SendBufSize,
 		RecvBufSize: ctxt.RecvBufSize,
@@ -430,7 +432,7 @@ func serverHandleCmdArgs(ctxt serverCtxt) {
 		RoundTripper: globalRemoteTargetTransport,
 		Logger: func(err error) {
 			if err != nil && !errors.Is(err, context.Canceled) {
-				replLogIf(GlobalContext, err)
+				proxyLogIf(GlobalContext, err)
 			}
 		},
 	})
@@ -679,10 +681,8 @@ func getServerListenAddrs() []string {
 	// Use a string set to avoid duplication
 	addrs := set.NewStringSet()
 	// Listen on local interface to receive requests from Console
-	for _, ip := range mustGetLocalIPs() {
+	for _, ip := range localLoopbacks.ToSlice() {
-		if ip != nil && ip.IsLoopback() {
+		addrs.Add(net.JoinHostPort(ip, globalMinioPort))
-			addrs.Add(net.JoinHostPort(ip.String(), globalMinioPort))
-		}
 	}
 	host, _ := mustSplitHostPort(globalMinioAddr)
 	if host != "" {
@@ -739,6 +739,8 @@ func initializeLogRotate(ctx *cli.Context) (io.WriteCloser, error) {
 
 // serverMain handler called for 'minio server' command.
 func serverMain(ctx *cli.Context) {
+	r := rand.New(rand.NewSource(time.Now().UnixNano()))
+
 	var warnings []string
 
 	signal.Notify(globalOSSignalCh, os.Interrupt, syscall.SIGTERM, syscall.SIGQUIT)
@@ -834,7 +836,7 @@ func serverMain(ctx *cli.Context) {
 
 	// Set system resources to maximum.
 	bootstrapTrace("setMaxResources", func() {
-		_ = setMaxResources(ctx)
+		_ = setMaxResources(globalServerCtxt)
 	})
 
 	// Verify kernel release and version.
@@ -865,7 +867,6 @@ func serverMain(ctx *cli.Context) {
 		httpServer := xhttp.NewServer(getServerListenAddrs()).
 			UseHandler(setCriticalErrorHandler(corsHandler(handler))).
 			UseTLSConfig(newTLSConfig(getCert)).
-			UseShutdownTimeout(globalServerCtxt.ShutdownTimeout).
 			UseIdleTimeout(globalServerCtxt.IdleTimeout).
 			UseReadHeaderTimeout(globalServerCtxt.ReadHeaderTimeout).
 			UseBaseContext(GlobalContext).
@@ -919,6 +920,20 @@ func serverMain(ctx *cli.Context) {
 		globalNodeNamesHex[hex.EncodeToString(nodeNameSum[:])] = struct{}{}
 	}
 
+	bootstrapTrace("waitForQuorum", func() {
+		result := newObject.Health(context.Background(), HealthOptions{})
+		for !result.HealthyRead {
+			if debugNoExit {
+				logger.Info("Not waiting for quorum since we are debugging.. possible cause unhealthy sets (%s)", result)
+				break
+			}
+			d := time.Duration(r.Float64() * float64(time.Second))
+			logger.Info("Waiting for quorum READ healthcheck to succeed.. possible cause unhealthy sets (%s), retrying in %s", result, d)
+			time.Sleep(d)
+			result = newObject.Health(context.Background(), HealthOptions{})
+		}
+	})
+
 	var err error
 	bootstrapTrace("initServerConfig", func() {
 		if err = initServerConfig(GlobalContext, newObject); err != nil {
@@ -985,8 +1000,6 @@ func serverMain(ctx *cli.Context) {
 	}()
 
 	go func() {
-		r := rand.New(rand.NewSource(time.Now().UnixNano()))
-
 		if !globalDisableFreezeOnBoot {
 			defer bootstrapTrace("unfreezeServices", unfreezeServices)
 			t := time.AfterFunc(5*time.Minute, func() {
@@ -1040,11 +1053,11 @@ func serverMain(ctx *cli.Context) {
 			bootLogIf(GlobalContext, globalEventNotifier.InitBucketTargets(GlobalContext, newObject))
 		})
 
-		var buckets []BucketInfo
+		var buckets []string
 		// List buckets to initialize bucket metadata sub-sys.
 		bootstrapTrace("listBuckets", func() {
 			for {
-				buckets, err = newObject.ListBuckets(GlobalContext, BucketOptions{})
+				bucketsList, err := newObject.ListBuckets(GlobalContext, BucketOptions{NoMetadata: true})
 				if err != nil {
 					if configRetriableErrors(err) {
 						logger.Info("Waiting for list buckets to succeed to initialize buckets.. possible cause (%v)", err)
@@ -1054,6 +1067,10 @@ func serverMain(ctx *cli.Context) {
 					bootLogIf(GlobalContext, fmt.Errorf("Unable to list buckets to initialize bucket metadata sub-system: %w", err))
 				}
 
+				buckets = make([]string, len(bucketsList))
+				for i := range bucketsList {
+					buckets[i] = bucketsList[i].Name
+				}
 				break
 			}
 		})

cmd/server-rlimit.go

@@ -22,7 +22,6 @@ import (
 	"runtime/debug"
 
 	"github.com/dustin/go-humanize"
-	"github.com/minio/cli"
 	"github.com/minio/madmin-go/v3/kernel"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/v3/sys"
@@ -45,7 +44,7 @@ func oldLinux() bool {
 	return currentKernel < kernel.Version(4, 0, 0)
 }
 
-func setMaxResources(ctx *cli.Context) (err error) {
+func setMaxResources(ctx serverCtxt) (err error) {
 	// Set the Go runtime max threads threshold to 90% of kernel setting.
 	sysMaxThreads, err := sys.GetMaxThreads()
 	if err == nil {
@@ -72,32 +71,28 @@ func setMaxResources(ctx *cli.Context) (err error) {
 		return err
 	}
 
-	// Set max memory limit as current memory limit.
+	_, vssLimit, err := sys.GetMaxMemoryLimit()
-	if _, maxLimit, err = sys.GetMaxMemoryLimit(); err != nil {
-		return err
-	}
-
-	// set debug memory limit instead of GOMEMLIMIT env
-	_ = setDebugMemoryLimit(ctx)
-
-	err = sys.SetMaxMemoryLimit(maxLimit, maxLimit)
-	return err
-}
-
-func setDebugMemoryLimit(ctx *cli.Context) error {
-	if ctx == nil {
-		return nil
-	}
-	if ctx.IsSet("memlimit") {
-		memlimit := ctx.String("memlimit")
-		if memlimit == "" {
-			memlimit = ctx.GlobalString("memlimit")
-		}
-		mlimit, err := humanize.ParseBytes(memlimit)
 	if err != nil {
 		return err
 	}
-		debug.SetMemoryLimit(int64(mlimit))
+
+	if vssLimit > 0 && vssLimit < humanize.GiByte {
+		logger.Info("WARNING: maximum virtual memory limit (%s) is too small for 'go runtime', please consider setting `ulimit -v` to unlimited",
+			humanize.IBytes(vssLimit))
 	}
+
+	if ctx.MemLimit > 0 {
+		maxLimit = ctx.MemLimit
+	}
+
+	if maxLimit > 0 {
+		debug.SetMemoryLimit(int64(maxLimit))
+	}
+
+	// Do not use RLIMIT_AS as that is not useful and at times on systems < 4Gi
+	// this can crash the Go runtime if the value is smaller refer
+	// - https://github.com/golang/go/issues/38010
+	// - https://github.com/golang/go/issues/43699
+	// So do not add `sys.SetMaxMemoryLimit()` this is not useful for any practical purposes.
 	return nil
 }

cmd/server-startup-msg.go

@@ -158,7 +158,7 @@ func printEventNotifiers() {
 		return
 	}
 
-	arns := globalEventNotifier.GetARNList(true)
+	arns := globalEventNotifier.GetARNList()
 	if len(arns) == 0 {
 		return
 	}

cmd/sftp-server-driver.go

@@ -1,4 +1,4 @@
-// Copyright (c) 2015-2023 MinIO, Inc.
+// Copyright (c) 2015-2024 MinIO, Inc.
 //
 // This file is part of MinIO Object Storage stack
 //
@@ -32,13 +32,16 @@ import (
 	"github.com/minio/madmin-go/v3"
 	"github.com/minio/minio-go/v7"
 	"github.com/minio/minio-go/v7/pkg/credentials"
-	"github.com/minio/minio/internal/auth"
 	xioutil "github.com/minio/minio/internal/ioutil"
 	"github.com/minio/pkg/v3/mimedb"
 	"github.com/pkg/sftp"
 	"golang.org/x/crypto/ssh"
 )
 
+// Maximum write offset for incoming SFTP blocks.
+// Set to 100MiB to prevent hostile DOS attacks.
+const ftpMaxWriteOffset = 100 << 20
+
 type sftpDriver struct {
 	permissions *ssh.Permissions
 	endpoint    string
@@ -49,7 +52,7 @@ type sftpMetrics struct{}
 
 var globalSftpMetrics sftpMetrics
 
-func sftpTrace(s *sftp.Request, startTime time.Time, source string, user string, err error) madmin.TraceInfo {
+func sftpTrace(s *sftp.Request, startTime time.Time, source string, user string, err error, sz int64) madmin.TraceInfo {
 	var errStr string
 	if err != nil {
 		errStr = err.Error()
@@ -58,18 +61,25 @@ func sftpTrace(s *sftp.Request, startTime time.Time, source string, user string,
 		TraceType: madmin.TraceFTP,
 		Time:      startTime,
 		NodeName:  globalLocalNodeName,
-		FuncName:  fmt.Sprintf("sftp USER=%s COMMAND=%s PARAM=%s, Source=%s", user, s.Method, s.Filepath, source),
+		FuncName:  s.Method,
 		Duration:  time.Since(startTime),
 		Path:      s.Filepath,
 		Error:     errStr,
+		Bytes:     sz,
+		Custom: map[string]string{
+			"user":   user,
+			"cmd":    s.Method,
+			"param":  s.Filepath,
+			"source": source,
+		},
 	}
 }
 
-func (m *sftpMetrics) log(s *sftp.Request, user string) func(err error) {
+func (m *sftpMetrics) log(s *sftp.Request, user string) func(sz int64, err error) {
 	startTime := time.Now()
 	source := getSource(2)
-	return func(err error) {
+	return func(sz int64, err error) {
-		globalTrace.Publish(sftpTrace(s, startTime, source, user, err))
+		globalTrace.Publish(sftpTrace(s, startTime, source, user, err, sz))
 	}
 }
 
@@ -90,77 +100,11 @@ func NewSFTPDriver(perms *ssh.Permissions) sftp.Handlers {
 }
 
 func (f *sftpDriver) getMinIOClient() (*minio.Client, error) {
-	ui, ok := globalIAMSys.GetUser(context.Background(), f.AccessKey())
+	mcreds := credentials.NewStaticV4(
-	if !ok && !globalIAMSys.LDAPConfig.Enabled() {
+		f.permissions.CriticalOptions["AccessKey"],
-		return nil, errNoSuchUser
+		f.permissions.CriticalOptions["SecretKey"],
-	}
+		f.permissions.CriticalOptions["SessionToken"],
-	if !ok && globalIAMSys.LDAPConfig.Enabled() {
+	)
-		sa, _, err := globalIAMSys.getServiceAccount(context.Background(), f.AccessKey())
-		if err != nil && !errors.Is(err, errNoSuchServiceAccount) {
-			return nil, err
-		}
-		var mcreds *credentials.Credentials
-		if errors.Is(err, errNoSuchServiceAccount) {
-			lookupResult, targetGroups, err := globalIAMSys.LDAPConfig.LookupUserDN(f.AccessKey())
-			if err != nil {
-				return nil, err
-			}
-			expiryDur, err := globalIAMSys.LDAPConfig.GetExpiryDuration("")
-			if err != nil {
-				return nil, err
-			}
-			claims := make(map[string]interface{})
-			claims[expClaim] = UTCNow().Add(expiryDur).Unix()
-			for k, v := range f.permissions.CriticalOptions {
-				claims[k] = v
-			}
-
-			// Set LDAP claims.
-			claims[ldapUserN] = f.AccessKey()
-			claims[ldapUser] = lookupResult.NormDN
-			// Add LDAP attributes that were looked up into the claims.
-			for attribKey, attribValue := range lookupResult.Attributes {
-				claims[ldapAttribPrefix+attribKey] = attribValue
-			}
-
-			cred, err := auth.GetNewCredentialsWithMetadata(claims, globalActiveCred.SecretKey)
-			if err != nil {
-				return nil, err
-			}
-
-			// Set the parent of the temporary access key, this is useful
-			// in obtaining service accounts by this cred.
-			cred.ParentUser = lookupResult.NormDN
-
-			// Set this value to LDAP groups, LDAP user can be part
-			// of large number of groups
-			cred.Groups = targetGroups
-
-			// Set the newly generated credentials, policyName is empty on purpose
-			// LDAP policies are applied automatically using their ldapUser, ldapGroups
-			// mapping.
-			updatedAt, err := globalIAMSys.SetTempUser(context.Background(), cred.AccessKey, cred, "")
-			if err != nil {
-				return nil, err
-			}
-
-			// Call hook for site replication.
-			replLogIf(context.Background(), globalSiteReplicationSys.IAMChangeHook(context.Background(), madmin.SRIAMItem{
-				Type: madmin.SRIAMItemSTSAcc,
-				STSCredential: &madmin.SRSTSCredential{
-					AccessKey:    cred.AccessKey,
-					SecretKey:    cred.SecretKey,
-					SessionToken: cred.SessionToken,
-					ParentUser:   cred.ParentUser,
-				},
-				UpdatedAt: updatedAt,
-			}))
-
-			mcreds = credentials.NewStaticV4(cred.AccessKey, cred.SecretKey, cred.SessionToken)
-		} else {
-			mcreds = credentials.NewStaticV4(sa.Credentials.AccessKey, sa.Credentials.SecretKey, "")
-		}
-
 	return minio.New(f.endpoint, &minio.Options{
 		Creds:     mcreds,
 		Secure:    globalIsTLS,
@@ -168,30 +112,14 @@ func (f *sftpDriver) getMinIOClient() (*minio.Client, error) {
 	})
 }
 
-	// ok == true - at this point
-
-	if ui.Credentials.IsTemp() {
-		// Temporary credentials are not allowed.
-		return nil, errAuthentication
-	}
-
-	return minio.New(f.endpoint, &minio.Options{
-		Creds:     credentials.NewStaticV4(ui.Credentials.AccessKey, ui.Credentials.SecretKey, ""),
-		Secure:    globalIsTLS,
-		Transport: globalRemoteFTPClientTransport,
-	})
-}
-
 func (f *sftpDriver) AccessKey() string {
-	if _, ok := f.permissions.CriticalOptions["accessKey"]; !ok {
+	return f.permissions.CriticalOptions["AccessKey"]
-		return f.permissions.CriticalOptions[ldapUserN]
-	}
-	return f.permissions.CriticalOptions["accessKey"]
 }
 
 func (f *sftpDriver) Fileread(r *sftp.Request) (ra io.ReaderAt, err error) {
+	// This is not timing the actual read operation, but the time it takes to prepare the reader.
 	stopFn := globalSftpMetrics.log(r, f.AccessKey())
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	flags := r.Pflags()
 	if !flags.Read {
@@ -269,6 +197,9 @@ func (w *writerAt) WriteAt(b []byte, offset int64) (n int, err error) {
 		n, err = w.w.Write(b)
 		w.nextOffset += int64(n)
 	} else {
+		if offset > w.nextOffset+ftpMaxWriteOffset {
+			return 0, fmt.Errorf("write offset %d is too far ahead of next offset %d", offset, w.nextOffset)
+		}
 		w.buffer[offset] = make([]byte, len(b))
 		copy(w.buffer[offset], b)
 		n = len(b)
@@ -294,7 +225,12 @@ again:
 
 func (f *sftpDriver) Filewrite(r *sftp.Request) (w io.WriterAt, err error) {
 	stopFn := globalSftpMetrics.log(r, f.AccessKey())
-	defer stopFn(err)
+	defer func() {
+		if err != nil {
+			// If there is an error, we never started the goroutine.
+			stopFn(0, err)
+		}
+	}()
 
 	flags := r.Pflags()
 	if !flags.Write {
@@ -329,10 +265,11 @@ func (f *sftpDriver) Filewrite(r *sftp.Request) (w io.WriterAt, err error) {
 	}
 	wa.wg.Add(1)
 	go func() {
-		_, err := clnt.PutObject(r.Context(), bucket, object, pr, -1, minio.PutObjectOptions{
+		oi, err := clnt.PutObject(r.Context(), bucket, object, pr, -1, minio.PutObjectOptions{
 			ContentType:          mimedb.TypeByExtension(path.Ext(object)),
 			DisableContentSha256: true,
 		})
+		stopFn(oi.Size, err)
 		pr.CloseWithError(err)
 		wa.wg.Done()
 	}()
@@ -341,7 +278,7 @@ func (f *sftpDriver) Filewrite(r *sftp.Request) (w io.WriterAt, err error) {
 
 func (f *sftpDriver) Filecmd(r *sftp.Request) (err error) {
 	stopFn := globalSftpMetrics.log(r, f.AccessKey())
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	clnt, err := f.getMinIOClient()
 	if err != nil {
@@ -437,7 +374,7 @@ func (f listerAt) ListAt(ls []os.FileInfo, offset int64) (int, error) {
 
 func (f *sftpDriver) Filelist(r *sftp.Request) (la sftp.ListerAt, err error) {
 	stopFn := globalSftpMetrics.log(r, f.AccessKey())
-	defer stopFn(err)
+	defer stopFn(0, err)
 
 	clnt, err := f.getMinIOClient()
 	if err != nil {

cmd/sftp-server.go

@@ -1,4 +1,4 @@
-// Copyright (c) 2015-2023 MinIO, Inc.
+// Copyright (c) 2015-2024 MinIO, Inc.
 //
 // This file is part of MinIO Object Storage stack
 //
@@ -18,7 +18,6 @@
 package cmd
 
 import (
-	"bytes"
 	"context"
 	"crypto/subtle"
 	"errors"
@@ -29,18 +28,15 @@ import (
 	"strings"
 	"time"
 
+	"github.com/minio/madmin-go/v3"
+	"github.com/minio/minio/internal/auth"
 	"github.com/minio/minio/internal/logger"
+	xldap "github.com/minio/pkg/v3/ldap"
 	xsftp "github.com/minio/pkg/v3/sftp"
 	"github.com/pkg/sftp"
 	"golang.org/x/crypto/ssh"
 )
 
-type sftpLogger struct{}
-
-func (s *sftpLogger) Info(tag xsftp.LogType, msg string) {
-	logger.Info(msg)
-}
-
 const (
 	kexAlgoDH1SHA1                = "diffie-hellman-group1-sha1"
 	kexAlgoDH14SHA1               = "diffie-hellman-group14-sha1"
@@ -58,6 +54,16 @@ const (
 	tripledescbcID     = "3des-cbc"
 )
 
+var (
+	errSFTPPublicKeyBadFormat = errors.New("the public key provided could not be parsed")
+	errSFTPUserHasNoPolicies  = errors.New("no policies present on this account")
+	errSFTPLDAPNotEnabled     = errors.New("ldap authentication is not enabled")
+)
+
+// if the sftp parameter --trusted-user-ca-key is set, then
+// the final form of the key file will be set as this variable.
+var caPublicKey ssh.PublicKey
+
 // https://cs.opensource.google/go/x/crypto/+/refs/tags/v0.22.0:ssh/common.go;l=46
 // preferredKexAlgos specifies the default preference for key-exchange
 // algorithms in preference order. The diffie-hellman-group16-sha512 algorithm
@@ -120,6 +126,226 @@ var supportedMACs = []string{
 	"hmac-sha2-256-etm@openssh.com", "hmac-sha2-512-etm@openssh.com", "hmac-sha2-256", "hmac-sha2-512", "hmac-sha1", "hmac-sha1-96",
 }
 
+func sshPubKeyAuth(c ssh.ConnMetadata, key ssh.PublicKey) (*ssh.Permissions, error) {
+	return authenticateSSHConnection(c, key, nil)
+}
+
+func sshPasswordAuth(c ssh.ConnMetadata, pass []byte) (*ssh.Permissions, error) {
+	return authenticateSSHConnection(c, nil, pass)
+}
+
+func authenticateSSHConnection(c ssh.ConnMetadata, key ssh.PublicKey, pass []byte) (*ssh.Permissions, error) {
+	user, found := strings.CutSuffix(c.User(), "=ldap")
+	if found {
+		if !globalIAMSys.LDAPConfig.Enabled() {
+			return nil, errSFTPLDAPNotEnabled
+		}
+		return processLDAPAuthentication(key, pass, user)
+	}
+
+	user, found = strings.CutSuffix(c.User(), "=svc")
+	if found {
+		goto internalAuth
+	}
+
+	if globalIAMSys.LDAPConfig.Enabled() {
+		perms, _ := processLDAPAuthentication(key, pass, user)
+		if perms != nil {
+			return perms, nil
+		}
+	}
+
+internalAuth:
+	ui, ok := globalIAMSys.GetUser(context.Background(), user)
+	if !ok {
+		return nil, errNoSuchUser
+	}
+
+	if caPublicKey != nil {
+		err := validateKey(c, key)
+		if err != nil {
+			return nil, errAuthentication
+		}
+	} else {
+
+		// Temporary credentials are not allowed.
+		if ui.Credentials.IsTemp() {
+			return nil, errAuthentication
+		}
+
+		if subtle.ConstantTimeCompare([]byte(ui.Credentials.SecretKey), pass) != 1 {
+			return nil, errAuthentication
+		}
+	}
+
+	return &ssh.Permissions{
+		CriticalOptions: map[string]string{
+			"AccessKey":    ui.Credentials.AccessKey,
+			"SecretKey":    ui.Credentials.SecretKey,
+			"SessionToken": ui.Credentials.SessionToken,
+		},
+		Extensions: make(map[string]string),
+	}, nil
+}
+
+func processLDAPAuthentication(key ssh.PublicKey, pass []byte, user string) (perms *ssh.Permissions, err error) {
+	var lookupResult *xldap.DNSearchResult
+	var targetGroups []string
+
+	if pass == nil && key == nil {
+		return nil, errAuthentication
+	}
+
+	if pass != nil {
+		sa, _, err := globalIAMSys.getServiceAccount(context.Background(), user)
+		if err == nil {
+			if subtle.ConstantTimeCompare([]byte(sa.Credentials.SecretKey), pass) != 1 {
+				return nil, errAuthentication
+			}
+
+			return &ssh.Permissions{
+				CriticalOptions: map[string]string{
+					"AccessKey":    sa.Credentials.AccessKey,
+					"SecretKey":    sa.Credentials.SecretKey,
+					"SessionToken": sa.Credentials.SessionToken,
+				},
+				Extensions: make(map[string]string),
+			}, nil
+		}
+
+		if !errors.Is(err, errNoSuchServiceAccount) {
+			return nil, err
+		}
+
+		lookupResult, targetGroups, err = globalIAMSys.LDAPConfig.Bind(user, string(pass))
+		if err != nil {
+			return nil, err
+		}
+
+	} else if key != nil {
+
+		lookupResult, targetGroups, err = globalIAMSys.LDAPConfig.LookupUserDN(user)
+		if err != nil {
+			return nil, err
+		}
+
+	}
+
+	if lookupResult == nil {
+		return nil, errNoSuchUser
+	}
+
+	ldapPolicies, _ := globalIAMSys.PolicyDBGet(lookupResult.NormDN, targetGroups...)
+	if len(ldapPolicies) == 0 {
+		return nil, errSFTPUserHasNoPolicies
+	}
+
+	claims := make(map[string]interface{})
+	for attribKey, attribValue := range lookupResult.Attributes {
+		// we skip multi-value attributes here, as they cannot
+		// be stored in the critical options.
+		if len(attribValue) != 1 {
+			continue
+		}
+
+		if attribKey == "sshPublicKey" && key != nil {
+			key2, _, _, _, err := ssh.ParseAuthorizedKey([]byte(attribValue[0]))
+			if err != nil {
+				return nil, errSFTPPublicKeyBadFormat
+			}
+
+			if subtle.ConstantTimeCompare(key2.Marshal(), key.Marshal()) != 1 {
+				return nil, errAuthentication
+			}
+		}
+		claims[ldapAttribPrefix+attribKey] = attribValue[0]
+	}
+
+	expiryDur, err := globalIAMSys.LDAPConfig.GetExpiryDuration("")
+	if err != nil {
+		return nil, err
+	}
+
+	claims[expClaim] = UTCNow().Add(expiryDur).Unix()
+	claims[ldapUserN] = user
+	claims[ldapUser] = lookupResult.NormDN
+
+	cred, err := auth.GetNewCredentialsWithMetadata(claims, globalActiveCred.SecretKey)
+	if err != nil {
+		return nil, err
+	}
+
+	// Set the parent of the temporary access key, this is useful
+	// in obtaining service accounts by this cred.
+	cred.ParentUser = lookupResult.NormDN
+
+	// Set this value to LDAP groups, LDAP user can be part
+	// of large number of groups
+	cred.Groups = targetGroups
+
+	// Set the newly generated credentials, policyName is empty on purpose
+	// LDAP policies are applied automatically using their ldapUser, ldapGroups
+	// mapping.
+	updatedAt, err := globalIAMSys.SetTempUser(context.Background(), cred.AccessKey, cred, "")
+	if err != nil {
+		return nil, err
+	}
+
+	replLogIf(context.Background(), globalSiteReplicationSys.IAMChangeHook(context.Background(), madmin.SRIAMItem{
+		Type: madmin.SRIAMItemSTSAcc,
+		STSCredential: &madmin.SRSTSCredential{
+			AccessKey:    cred.AccessKey,
+			SecretKey:    cred.SecretKey,
+			SessionToken: cred.SessionToken,
+			ParentUser:   cred.ParentUser,
+		},
+		UpdatedAt: updatedAt,
+	}))
+
+	return &ssh.Permissions{
+		CriticalOptions: map[string]string{
+			"AccessKey":    cred.AccessKey,
+			"SecretKey":    cred.SecretKey,
+			"SessionToken": cred.SessionToken,
+		},
+		Extensions: make(map[string]string),
+	}, nil
+}
+
+func validateKey(c ssh.ConnMetadata, clientKey ssh.PublicKey) (err error) {
+	if caPublicKey == nil {
+		return errors.New("public key authority validation requested but no ca public key specified.")
+	}
+
+	cert, ok := clientKey.(*ssh.Certificate)
+	if !ok {
+		return errSftpPublicKeyWithoutCert
+	}
+
+	// ssh.CheckCert called by ssh.Authenticate accepts certificates
+	// with empty principles list so we block those in here.
+	if len(cert.ValidPrincipals) == 0 {
+		return errSftpCertWithoutPrincipals
+	}
+
+	// Verify that certificate provided by user is issued by trusted CA,
+	// username in authentication request matches to identities in certificate
+	// and that certificate type is correct.
+	checker := ssh.CertChecker{}
+	checker.IsUserAuthority = func(k ssh.PublicKey) bool {
+		return subtle.ConstantTimeCompare(caPublicKey.Marshal(), k.Marshal()) == 1
+	}
+
+	_, err = checker.Authenticate(c, clientKey)
+	return
+}
+
+type sftpLogger struct{}
+
+func (s *sftpLogger) Info(tag xsftp.LogType, msg string) {
+	logger.Info(msg)
+}
+
 func (s *sftpLogger) Error(tag xsftp.LogType, err error) {
 	switch tag {
 	case xsftp.AcceptNetworkError:
@@ -164,12 +390,15 @@ func startSFTPServer(args []string) {
 		publicIP        string
 		sshPrivateKey   string
 		userCaKeyFile   string
+		disablePassAuth bool
 	)
+
 	allowPubKeys := supportedPubKeyAuthAlgos
 	allowKexAlgos := preferredKexAlgos
 	allowCiphers := preferredCiphers
 	allowMACs := supportedMACs
 	var err error
+
 	for _, arg := range args {
 		tokens := strings.SplitN(arg, "=", 2)
 		if len(tokens) != 2 {
@@ -201,6 +430,8 @@ func startSFTPServer(args []string) {
 			allowMACs = filterAlgos(arg, strings.Split(tokens[1], ","), supportedMACs)
 		case "trusted-user-ca-key":
 			userCaKeyFile = tokens[1]
+		case "password-auth":
+			disablePassAuth, _ = strconv.ParseBool(tokens[1])
 		}
 	}
 
@@ -222,6 +453,18 @@ func startSFTPServer(args []string) {
 		logger.Fatal(fmt.Errorf("invalid arguments passed, private key file is not parseable: %v", err), "unable to start SFTP server")
 	}
 
+	if userCaKeyFile != "" {
+		keyBytes, err := os.ReadFile(userCaKeyFile)
+		if err != nil {
+			logger.Fatal(fmt.Errorf("invalid arguments passed, trusted user certificate authority public key file is not accessible: %v", err), "unable to start SFTP server")
+		}
+
+		caPublicKey, _, _, _, err = ssh.ParseAuthorizedKey(keyBytes)
+		if err != nil {
+			logger.Fatal(fmt.Errorf("invalid arguments passed, trusted user certificate authority public key file is not parseable: %v", err), "unable to start SFTP server")
+		}
+	}
+
 	// An SSH server is represented by a ServerConfig, which holds
 	// certificate details and handles authentication of ServerConns.
 	sshConfig := &ssh.ServerConfig{
@@ -231,116 +474,13 @@ func startSFTPServer(args []string) {
 			MACs:         allowMACs,
 		},
 		PublicKeyAuthAlgorithms: allowPubKeys,
-		PasswordCallback: func(c ssh.ConnMetadata, pass []byte) (*ssh.Permissions, error) {
+		PublicKeyCallback:       sshPubKeyAuth,
-			if globalIAMSys.LDAPConfig.Enabled() {
-				sa, _, err := globalIAMSys.getServiceAccount(context.Background(), c.User())
-				if err != nil && !errors.Is(err, errNoSuchServiceAccount) {
-					return nil, err
-				}
-				if errors.Is(err, errNoSuchServiceAccount) {
-					lookupResult, targetGroups, err := globalIAMSys.LDAPConfig.Bind(c.User(), string(pass))
-					if err != nil {
-						return nil, err
-					}
-					targetUser := lookupResult.NormDN
-					ldapPolicies, _ := globalIAMSys.PolicyDBGet(targetUser, targetGroups...)
-					if len(ldapPolicies) == 0 {
-						return nil, errAuthentication
-					}
-					criticalOptions := map[string]string{
-						ldapUser:       targetUser,
-						ldapActualUser: lookupResult.ActualDN,
-						ldapUserN:      c.User(),
-					}
-					for attribKey, attribValue := range lookupResult.Attributes {
-						// we skip multi-value attributes here, as they cannot
-						// be stored in the critical options.
-						if len(attribValue) == 1 {
-							criticalOptions[ldapAttribPrefix+attribKey] = attribValue[0]
-						}
 	}
 
-					return &ssh.Permissions{
+	if !disablePassAuth {
-						CriticalOptions: criticalOptions,
+		sshConfig.PasswordCallback = sshPasswordAuth
-						Extensions:      make(map[string]string),
+	} else {
-					}, nil
+		sshConfig.PasswordCallback = nil
-				}
-				if subtle.ConstantTimeCompare([]byte(sa.Credentials.SecretKey), pass) == 1 {
-					return &ssh.Permissions{
-						CriticalOptions: map[string]string{
-							"accessKey": c.User(),
-						},
-						Extensions: make(map[string]string),
-					}, nil
-				}
-				return nil, errAuthentication
-			}
-
-			ui, ok := globalIAMSys.GetUser(context.Background(), c.User())
-			if !ok {
-				return nil, errNoSuchUser
-			}
-
-			if subtle.ConstantTimeCompare([]byte(ui.Credentials.SecretKey), pass) == 1 {
-				return &ssh.Permissions{
-					CriticalOptions: map[string]string{
-						"accessKey": c.User(),
-					},
-					Extensions: make(map[string]string),
-				}, nil
-			}
-			return nil, errAuthentication
-		},
-	}
-
-	if userCaKeyFile != "" {
-		keyBytes, err := os.ReadFile(userCaKeyFile)
-		if err != nil {
-			logger.Fatal(fmt.Errorf("invalid arguments passed, trusted user certificate authority public key file is not accessible: %v", err), "unable to start SFTP server")
-		}
-
-		caPublicKey, _, _, _, err := ssh.ParseAuthorizedKey(keyBytes)
-		if err != nil {
-			logger.Fatal(fmt.Errorf("invalid arguments passed, trusted user certificate authority public key file is not parseable: %v", err), "unable to start SFTP server")
-		}
-
-		sshConfig.PublicKeyCallback = func(c ssh.ConnMetadata, key ssh.PublicKey) (*ssh.Permissions, error) {
-			_, ok := globalIAMSys.GetUser(context.Background(), c.User())
-			if !ok {
-				return nil, errNoSuchUser
-			}
-
-			// Verify that client provided certificate, not only public key.
-			cert, ok := key.(*ssh.Certificate)
-			if !ok {
-				return nil, errSftpPublicKeyWithoutCert
-			}
-
-			// ssh.CheckCert called by ssh.Authenticate accepts certificates
-			// with empty principles list so we block those in here.
-			if len(cert.ValidPrincipals) == 0 {
-				return nil, errSftpCertWithoutPrincipals
-			}
-
-			// Verify that certificate provided by user is issued by trusted CA,
-			// username in authentication request matches to identities in certificate
-			// and that certificate type is correct.
-			checker := ssh.CertChecker{}
-			checker.IsUserAuthority = func(k ssh.PublicKey) bool {
-				return bytes.Equal(k.Marshal(), caPublicKey.Marshal())
-			}
-			_, err = checker.Authenticate(c, key)
-			if err != nil {
-				return nil, err
-			}
-
-			return &ssh.Permissions{
-				CriticalOptions: map[string]string{
-					"accessKey": c.User(),
-				},
-				Extensions: make(map[string]string),
-			}, nil
-		}
 	}
 
 	sshConfig.AddHostKey(private)

cmd/sftp-server_test.go

@@ -0,0 +1,289 @@
+// Copyright (c) 2015-2024 MinIO, Inc.
+//
+// This file is part of MinIO Object Storage stack
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package cmd
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"net"
+	"os"
+	"testing"
+
+	"github.com/minio/madmin-go/v3"
+	"golang.org/x/crypto/ssh"
+)
+
+type MockConnMeta struct {
+	username string
+}
+
+func (m *MockConnMeta) User() string {
+	return m.username
+}
+
+func (m *MockConnMeta) SessionID() []byte {
+	return []byte{}
+}
+
+func (m *MockConnMeta) ClientVersion() []byte {
+	return []byte{}
+}
+
+func (m *MockConnMeta) ServerVersion() []byte {
+	return []byte{}
+}
+
+func (m *MockConnMeta) RemoteAddr() net.Addr {
+	return nil
+}
+
+func (m *MockConnMeta) LocalAddr() net.Addr {
+	return nil
+}
+
+func newSSHConnMock(username string) ssh.ConnMetadata {
+	return &MockConnMeta{username: username}
+}
+
+func TestSFTPAuthentication(t *testing.T) {
+	for i, testCase := range iamTestSuites {
+		t.Run(
+			fmt.Sprintf("Test: %d, ServerType: %s", i+1, testCase.ServerTypeDescription),
+			func(t *testing.T) {
+				c := &check{t, testCase.serverType}
+				suite := testCase
+
+				suite.SetUpSuite(c)
+
+				suite.SFTPServiceAccountLogin(c)
+				suite.SFTPInvalidServiceAccountPassword(c)
+
+				// LDAP tests
+				ldapServer := os.Getenv(EnvTestLDAPServer)
+				if ldapServer == "" {
+					c.Skipf("Skipping LDAP test as no LDAP server is provided via %s", EnvTestLDAPServer)
+				}
+
+				suite.SetUpLDAP(c, ldapServer)
+
+				suite.SFTPFailedAuthDueToMissingPolicy(c)
+				suite.SFTPFailedAuthDueToInvalidUser(c)
+				suite.SFTPFailedForcedServiceAccountAuthOnLDAPUser(c)
+				suite.SFTPFailedAuthDueToInvalidPassword(c)
+
+				suite.SFTPValidLDAPLoginWithPassword(c)
+
+				suite.SFTPPublicKeyAuthentication(c)
+				suite.SFTPFailedPublicKeyAuthenticationInvalidKey(c)
+
+				suite.TearDownSuite(c)
+			},
+		)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPFailedPublicKeyAuthenticationInvalidKey(c *check) {
+	keyBytes, err := os.ReadFile("./testdata/invalid_test_key.pub")
+	if err != nil {
+		c.Fatalf("could not read test key file: %s", err)
+	}
+
+	testKey, _, _, _, err := ssh.ParseAuthorizedKey(keyBytes)
+	if err != nil {
+		c.Fatalf("could not parse test key file: %s", err)
+	}
+
+	newSSHCon := newSSHConnMock("dillon=ldap")
+	_, err = sshPubKeyAuth(newSSHCon, testKey)
+	if err == nil || !errors.Is(err, errAuthentication) {
+		c.Fatalf("expected err(%s) but got (%s)", errAuthentication, err)
+	}
+
+	newSSHCon = newSSHConnMock("dillon")
+	_, err = sshPubKeyAuth(newSSHCon, testKey)
+	if err == nil || !errors.Is(err, errNoSuchUser) {
+		c.Fatalf("expected err(%s) but got (%s)", errNoSuchUser, err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPPublicKeyAuthentication(c *check) {
+	keyBytes, err := os.ReadFile("./testdata/dillon_test_key.pub")
+	if err != nil {
+		c.Fatalf("could not read test key file: %s", err)
+	}
+
+	testKey, _, _, _, err := ssh.ParseAuthorizedKey(keyBytes)
+	if err != nil {
+		c.Fatalf("could not parse test key file: %s", err)
+	}
+
+	newSSHCon := newSSHConnMock("dillon=ldap")
+	_, err = sshPubKeyAuth(newSSHCon, testKey)
+	if err != nil {
+		c.Fatalf("expected no error but got(%s)", err)
+	}
+
+	newSSHCon = newSSHConnMock("dillon")
+	_, err = sshPubKeyAuth(newSSHCon, testKey)
+	if err != nil {
+		c.Fatalf("expected no error but got(%s)", err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPFailedAuthDueToMissingPolicy(c *check) {
+	newSSHCon := newSSHConnMock("dillon=ldap")
+	_, err := sshPasswordAuth(newSSHCon, []byte("dillon"))
+	if err == nil || !errors.Is(err, errSFTPUserHasNoPolicies) {
+		c.Fatalf("expected err(%s) but got (%s)", errSFTPUserHasNoPolicies, err)
+	}
+
+	newSSHCon = newSSHConnMock("dillon")
+	_, err = sshPasswordAuth(newSSHCon, []byte("dillon"))
+	if err == nil || !errors.Is(err, errNoSuchUser) {
+		c.Fatalf("expected err(%s) but got (%s)", errNoSuchUser, err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPFailedAuthDueToInvalidUser(c *check) {
+	newSSHCon := newSSHConnMock("dillon_error")
+	_, err := sshPasswordAuth(newSSHCon, []byte("dillon_error"))
+	if err == nil || !errors.Is(err, errNoSuchUser) {
+		c.Fatalf("expected err(%s) but got (%s)", errNoSuchUser, err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPFailedForcedServiceAccountAuthOnLDAPUser(c *check) {
+	newSSHCon := newSSHConnMock("dillon=svc")
+	_, err := sshPasswordAuth(newSSHCon, []byte("dillon"))
+	if err == nil || !errors.Is(err, errNoSuchUser) {
+		c.Fatalf("expected err(%s) but got (%s)", errNoSuchUser, err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPFailedAuthDueToInvalidPassword(c *check) {
+	newSSHCon := newSSHConnMock("dillon")
+	_, err := sshPasswordAuth(newSSHCon, []byte("dillon_error"))
+	if err == nil || !errors.Is(err, errNoSuchUser) {
+		c.Fatalf("expected err(%s) but got (%s)", errNoSuchUser, err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPInvalidServiceAccountPassword(c *check) {
+	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
+	defer cancel()
+
+	accessKey, secretKey := mustGenerateCredentials(c)
+	err := s.adm.SetUser(ctx, accessKey, secretKey, madmin.AccountEnabled)
+	if err != nil {
+		c.Fatalf("Unable to set user: %v", err)
+	}
+
+	err = s.adm.SetPolicy(ctx, "readwrite", accessKey, false)
+	if err != nil {
+		c.Fatalf("unable to set policy: %v", err)
+	}
+
+	newSSHCon := newSSHConnMock(accessKey + "=svc")
+	_, err = sshPasswordAuth(newSSHCon, []byte("invalid"))
+	if err == nil || !errors.Is(err, errAuthentication) {
+		c.Fatalf("expected err(%s) but got (%s)", errAuthentication, err)
+	}
+
+	newSSHCon = newSSHConnMock(accessKey)
+	_, err = sshPasswordAuth(newSSHCon, []byte("invalid"))
+	if err == nil || !errors.Is(err, errAuthentication) {
+		c.Fatalf("expected err(%s) but got (%s)", errAuthentication, err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPServiceAccountLogin(c *check) {
+	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
+	defer cancel()
+
+	accessKey, secretKey := mustGenerateCredentials(c)
+	err := s.adm.SetUser(ctx, accessKey, secretKey, madmin.AccountEnabled)
+	if err != nil {
+		c.Fatalf("Unable to set user: %v", err)
+	}
+
+	err = s.adm.SetPolicy(ctx, "readwrite", accessKey, false)
+	if err != nil {
+		c.Fatalf("unable to set policy: %v", err)
+	}
+
+	newSSHCon := newSSHConnMock(accessKey + "=svc")
+	_, err = sshPasswordAuth(newSSHCon, []byte(secretKey))
+	if err != nil {
+		c.Fatalf("expected no error but got (%s)", err)
+	}
+
+	newSSHCon = newSSHConnMock(accessKey)
+	_, err = sshPasswordAuth(newSSHCon, []byte(secretKey))
+	if err != nil {
+		c.Fatalf("expected no error but got (%s)", err)
+	}
+}
+
+func (s *TestSuiteIAM) SFTPValidLDAPLoginWithPassword(c *check) {
+	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
+	defer cancel()
+
+	// we need to do this so that the user has a policy before authentication.
+	// ldap user accounts without policies are denied access in sftp.
+	policy := "mypolicy"
+	policyBytes := []byte(`{
+ "Version": "2012-10-17",
+ "Statement": [
+  {
+   "Effect": "Allow",
+   "Action": [
+    "s3:PutObject",
+    "s3:GetObject",
+    "s3:ListBucket"
+   ],
+   "Resource": [
+    "arn:aws:s3:::BUCKET/*"
+   ]
+  }
+ ]
+}`)
+
+	err := s.adm.AddCannedPolicy(ctx, policy, policyBytes)
+	if err != nil {
+		c.Fatalf("policy add error: %v", err)
+	}
+
+	userDN := "uid=dillon,ou=people,ou=swengg,dc=min,dc=io"
+	err = s.adm.SetPolicy(ctx, policy, userDN, false)
+	if err != nil {
+		c.Fatalf("Unable to set policy: %v", err)
+	}
+
+	newSSHCon := newSSHConnMock("dillon=ldap")
+	_, err = sshPasswordAuth(newSSHCon, []byte("dillon"))
+	if err != nil {
+		c.Fatal("Password authentication failed for user (dillon):", err)
+	}
+
+	newSSHCon = newSSHConnMock("dillon")
+	_, err = sshPasswordAuth(newSSHCon, []byte("dillon"))
+	if err != nil {
+		c.Fatal("Password authentication failed for user (dillon):", err)
+	}
+}

cmd/signature-v4-utils.go

@@ -152,11 +152,14 @@ func checkKeyValid(r *http.Request, accessKey string) (auth.Credentials, bool, A
 			// Check if server has initialized, then only proceed
 			// to check for IAM users otherwise its okay for clients
 			// to retry with 503 errors when server is coming up.
-			return auth.Credentials{}, false, ErrServerNotInitialized
+			return auth.Credentials{}, false, ErrIAMNotInitialized
 		}
 
 		// Check if the access key is part of users credentials.
-		u, ok := globalIAMSys.GetUser(r.Context(), accessKey)
+		u, ok, err := globalIAMSys.CheckKey(r.Context(), accessKey)
+		if err != nil {
+			return auth.Credentials{}, false, ErrIAMNotInitialized
+		}
 		if !ok {
 			// Credentials could be valid but disabled - return a different
 			// error in such a scenario.

cmd/signature-v4_test.go

@@ -37,6 +37,12 @@ func niceError(code APIErrorCode) string {
 }
 
 func TestDoesPolicySignatureMatch(t *testing.T) {
+	_, fsDir, err := prepareFS(context.Background())
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer removeRoots([]string{fsDir})
+
 	credentialTemplate := "%s/%s/%s/s3/aws4_request"
 	now := UTCNow()
 	accessKey := globalActiveCred.AccessKey