Add periodic callhome functionality (#14918)

* Add periodic callhome functionality

Periodically (every 24hrs by default), fetch callhome information and
upload it to SUBNET.

New config keys under the `callhome` subsystem:

enable - Set to `on` for enabling callhome. Default `off`
frequency - Interval between callhome cycles. Default `24h`

* Improvements based on review comments

- Update `enableCallhome` safely
- Rename pctx to ctx
- Block during execution of callhome
- Store parsed proxy URL in global subnet config
- Store callhome URL(s) in constants
- Use existing global transport
- Pass auth token to subnetPostReq
- Use `config.EnableOn` instead of `"on"`

* Use atomic package instead of lock

* Use uber atomic package

* Use `Cancel` instead of `cancel`

Co-authored-by: Harshavardhana <harsha@minio.io>

Co-authored-by: Harshavardhana <harsha@minio.io>
Co-authored-by: Aditya Manthramurthy <donatello@users.noreply.github.com>

.github/workflows/go-fips.yml

@@ -0,0 +1,51 @@
+name: FIPS Build Test
+
+on:
+  pull_request:
+    branches:
+    - master
+
+# This ensures that previous jobs for the PR are canceled when the PR is
+# updated.
+concurrency:
+  group: ${{ github.workflow }}-${{ github.head_ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+jobs:
+  build:
+    name: Go BoringCrypto ${{ matrix.go-version }} on ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        go-version: [1.17.11b7, 1.18.3b7]
+        os: [ubuntu-latest]
+    steps:
+      - uses: actions/checkout@v2
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+
+      - name: Setup dockerfile for build test
+        run: |
+          echo "FROM us-docker.pkg.dev/google.com/api-project-999119582588/go-boringcrypto/golang:${{ matrix.go-version }}" > Dockerfile.fips.test
+          echo "COPY . /minio" >> Dockerfile.fips.test
+          echo "WORKDIR /minio" >> Dockerfile.fips.test
+          echo "RUN make" >> Dockerfile.fips.test
+
+      - name: Build
+        uses: docker/build-push-action@v3
+        with:
+          context: .
+          file: Dockerfile.fips.test
+          push: false
+          load: true
+          tags: minio/fips-test:latest
+
+      # This should fail if grep returns non-zero exit
+      - name: Test binary
+        run: |
+          docker run --rm minio/fips-test:latest ./minio --version
+          docker run --rm -i minio/fips-test:latest /bin/bash -c 'go tool nm ./minio' | grep -q FIPS

.github/workflows/iam-integrations.yaml

@@ -108,6 +108,17 @@ jobs:
           sudo sysctl net.ipv6.conf.all.disable_ipv6=0
           sudo sysctl net.ipv6.conf.default.disable_ipv6=0
           make test-iam
+      - name: Test with Access Management Plugin enabled
+        env:
+          LDAP_TEST_SERVER: ${{ matrix.ldap }}
+          ETCD_SERVER: ${{ matrix.etcd }}
+          OPENID_TEST_SERVER: ${{ matrix.openid }}
+          POLICY_PLUGIN_ENDPOINT: "http://127.0.0.1:8080"
+        run: |
+          sudo sysctl net.ipv6.conf.all.disable_ipv6=0
+          sudo sysctl net.ipv6.conf.default.disable_ipv6=0
+          go run docs/iam/access-manager-plugin.go &
+          make test-iam
       - name: Test LDAP for automatic site replication
         if: matrix.ldap == 'localhost:389'
         run: |

Makefile

@@ -53,11 +53,12 @@ test-iam: build ## verify IAM (external IDP, etcd backends)
 	@echo "Running tests for IAM (external IDP, etcd backends)"
 	@CGO_ENABLED=0 go test -tags kqueue -v -run TestIAM* ./cmd
 	@echo "Running tests for IAM (external IDP, etcd backends) with -race"
-	@CGO_ENABLED=1 go test -race -tags kqueue -v -run TestIAM* ./cmd
+	@GORACE=history_size=7 CGO_ENABLED=1 go test -race -tags kqueue -v -run TestIAM* ./cmd
 
 test-replication: install ## verify multi site replication
 	@echo "Running tests for replicating three sites"
 	@(env bash $(PWD)/docs/bucket/replication/setup_3site_replication.sh)
+	@(env bash $(PWD)/docs/bucket/replication/setup_2site_existing_replication.sh)
 
 test-site-replication-ldap: install ## verify automatic site replication
 	@echo "Running tests for automatic site replication of IAM (with LDAP)"
@@ -73,18 +74,18 @@ test-site-replication-minio: install ## verify automatic site replication
 
 verify: ## verify minio various setups
 	@echo "Verifying build with race"
-	@CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@GORACE=history_size=7 CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
 	@(env bash $(PWD)/buildscripts/verify-build.sh)
 
 verify-healing: ## verify healing and replacing disks with minio binary
 	@echo "Verify healing build with race"
-	@CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@GORACE=history_size=7 CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
 	@(env bash $(PWD)/buildscripts/verify-healing.sh)
 	@(env bash $(PWD)/buildscripts/unaligned-healing.sh)
 
 verify-healing-inconsistent-versions: ## verify resolving inconsistent versions
 	@echo "Verify resolving inconsistent versions build with race"
-	@CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@GORACE=history_size=7 CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
 	@(env bash $(PWD)/buildscripts/resolve-right-versions.sh)
 
 build: checks ## builds minio to $(PWD)

README.fips.md

@@ -0,0 +1,7 @@
+# MinIO FIPS Builds
+
+MinIO creates FIPS builds using a patched version of the Go compiler (that uses BoringCrypto, from BoringSSL, which is [FIPS 140-2 validated](https://csrc.nist.gov/csrc/media/projects/cryptographic-module-validation-program/documents/security-policies/140sp2964.pdf)) published by the Golang Team [here](https://github.com/golang/go/tree/dev.boringcrypto/misc/boring).
+
+MinIO FIPS executables are available at http://dl.min.io - they are only published for `linux-amd64` architecture as binary files with the suffix `.fips`. We also publish corresponding container images to our official image repositories.
+
+We are not making any statements or representations about the suitability of this code or build in relation to the FIPS 140-2 standard. Interested users will have to evaluate for themselves whether this is useful for their own purposes.

README.md

@@ -196,12 +196,6 @@ iptables -A INPUT -p tcp --dport 9000:9010 -j ACCEPT
 service iptables restart
 ```
 
-## Pre-existing data
-
-When deployed on a single drive, MinIO server lets clients access any pre-existing data in the data directory. For example, if MinIO is started with the command  `minio server /mnt/data`, any pre-existing data in the `/mnt/data` directory would be accessible to the clients.
-
-The above statement is also valid for all gateway backends.
-
 ## Test MinIO Connectivity
 
 ### Test using MinIO Console
@@ -244,13 +238,13 @@ mc admin update <minio alias, e.g., myminio>
 
 - For deployments without external internet access (e.g. airgapped environments), download the binary from <https://dl.min.io> and replace the existing MinIO binary let's say for example `/opt/bin/minio`, apply executable permissions `chmod +x /opt/bin/minio` and do `mc admin service restart alias/`.
 
-- For RPM/DEB installations, upgrade packages **parallelly** on all servers. Once upgraded, perform `systemctl restart minio` across all nodes in **parallel**. RPM/DEB based installations are usually automated using [`ansible`](https://github.com/minio/ansible-minio).
+- For installations using Systemd MinIO service, upgrade via RPM/DEB packages **parallelly** on all servers or replace the binary lets say `/opt/bin/minio` on all nodes, apply executable permissions `chmod +x /opt/bin/minio`. Proceed to perform `systemctl restart minio` across all nodes in **parallel**.
 
 ### Upgrade Checklist
 
 - Test all upgrades in a lower environment (DEV, QA, UAT) before applying to production. Performing blind upgrades in production environments carries significant risk.
 - Read the release notes for the targeted MinIO release *before* performing any installation, there is no forced requirement to upgrade to latest releases every week. If it has a bug fix you are looking for then yes, else avoid actively upgrading a running production system.
-- Make sure MinIO process has write access to `/opt/bin` if you plan to use `mc admin update`. This is needed for MinIO to download the latest binary from <https://dl.min.io> and save it locally for upgrades.
+- If you plan to use `mc admin update`, MinIO process must have write access to the parent directory to provide in-place upgrades.
 - `mc admin update` is not supported in kubernetes/container environments, container environments provide their own mechanisms for container updates.
 - **We do not recommend upgrading one MinIO server at a time, the product is designed to support parallel upgrades please follow our recommended guidelines.**
 

buildscripts/race.sh

@@ -2,6 +2,7 @@
 
 set -e
 
+export GORACE="history_size=7"
 ## TODO remove `dsync` from race detector once this is merged and released https://go-review.googlesource.com/c/go/+/333529/
 for d in $(go list ./... | grep -v dsync); do
     CGO_ENABLED=1 go test -v -race --timeout 100m "$d"

cmd/admin-bucket-handlers.go

@@ -145,7 +145,7 @@ func (a adminAPIHandlers) SetRemoteTargetHandler(w http.ResponseWriter, r *http.
 	bucket := pathClean(vars["bucket"])
 	update := r.Form.Get("update") == "true"
 
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
 		return
 	}
@@ -274,7 +274,8 @@ func (a adminAPIHandlers) ListRemoteTargetsHandler(w http.ResponseWriter, r *htt
 	vars := mux.Vars(r)
 	bucket := pathClean(vars["bucket"])
 	arnType := vars["type"]
-	if !globalIsErasure {
+
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
 		return
 	}
@@ -314,7 +315,7 @@ func (a adminAPIHandlers) RemoveRemoteTargetHandler(w http.ResponseWriter, r *ht
 	bucket := pathClean(vars["bucket"])
 	arn := vars["arn"]
 
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
 		return
 	}

cmd/admin-handler-utils.go

@@ -180,7 +180,7 @@ func toAdminAPIErr(ctx context.Context, err error) APIError {
 			apiErr = APIError{
 				Code:           "XMinioAdminTierBackendInUse",
 				Description:    err.Error(),
-				HTTPStatusCode: http.StatusConflict,
+				HTTPStatusCode: http.StatusBadRequest,
 			}
 		case errors.Is(err, errTierBackendNotEmpty):
 			apiErr = APIError{

cmd/admin-handlers-config-kv.go

@@ -33,7 +33,8 @@ import (
 	"github.com/minio/minio/internal/config/etcd"
 	xldap "github.com/minio/minio/internal/config/identity/ldap"
 	"github.com/minio/minio/internal/config/identity/openid"
-	"github.com/minio/minio/internal/config/policy/opa"
+	idplugin "github.com/minio/minio/internal/config/identity/plugin"
+	polplugin "github.com/minio/minio/internal/config/policy/plugin"
 	"github.com/minio/minio/internal/config/storageclass"
 	"github.com/minio/minio/internal/logger"
 	iampolicy "github.com/minio/pkg/iam/policy"
@@ -104,7 +105,7 @@ func applyDynamic(ctx context.Context, objectAPI ObjectLayer, cfg config.Config,
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
-	globalNotificationSys.SignalService(serviceReloadDynamic)
+	globalNotificationSys.SignalConfigReload(subSys)
 	// Tell the client that dynamic config was applied.
 	w.Header().Set(madmin.ConfigAppliedHeader, madmin.ConfigAppliedTrue)
 }
@@ -436,14 +437,16 @@ func (a adminAPIHandlers) GetConfigHandler(w http.ResponseWriter, r *http.Reques
 				off = !cache.Enabled(kv)
 			case config.StorageClassSubSys:
 				off = !storageclass.Enabled(kv)
-			case config.PolicyOPASubSys:
-				off = !opa.Enabled(kv)
+			case config.PolicyPluginSubSys:
+				off = !polplugin.Enabled(kv)
 			case config.IdentityOpenIDSubSys:
 				off = !openid.Enabled(kv)
 			case config.IdentityLDAPSubSys:
 				off = !xldap.Enabled(kv)
 			case config.IdentityTLSSubSys:
 				off = !globalSTSTLSConfig.Enabled
+			case config.IdentityPluginSubSys:
+				off = !idplugin.Enabled(kv)
 			}
 			if off {
 				s.WriteString(config.KvComment)

cmd/admin-handlers-site-replication.go

@@ -241,6 +241,8 @@ func (a adminAPIHandlers) SRPeerReplicateBucketItem(w http.ResponseWriter, r *ht
 				return
 			}
 		}
+	case madmin.SRBucketMetaTypeVersionConfig:
+		err = globalSiteReplicationSys.PeerBucketVersioningHandler(ctx, item.Bucket, item.Versioning)
 	case madmin.SRBucketMetaTypeTags:
 		err = globalSiteReplicationSys.PeerBucketTaggingHandler(ctx, item.Bucket, item.Tags)
 	case madmin.SRBucketMetaTypeObjectLockConfig:

cmd/admin-handlers-users-race_test.go

@@ -26,6 +26,7 @@ package cmd
 import (
 	"context"
 	"fmt"
+	"runtime"
 	"sync"
 	"testing"
 	"time"
@@ -41,11 +42,15 @@ func runAllIAMConcurrencyTests(suite *TestSuiteIAM, c *check) {
 }
 
 func TestIAMInternalIDPConcurrencyServerSuite(t *testing.T) {
+	if runtime.GOOS == globalWindowsOSName {
+		t.Skip("windows is clunky")
+	}
+
 	baseTestCases := []TestSuiteCommon{
-		// Init and run test on FS backend with signature v4.
-		{serverType: "FS", signer: signerV4},
-		// Init and run test on FS backend, with tls enabled.
-		{serverType: "FS", signer: signerV4, secure: true},
+		// Init and run test on ErasureSD backend with signature v4.
+		{serverType: "ErasureSD", signer: signerV4},
+		// Init and run test on ErasureSD backend, with tls enabled.
+		{serverType: "ErasureSD", signer: signerV4, secure: true},
 		// Init and run test on Erasure backend.
 		{serverType: "Erasure", signer: signerV4},
 		// Init and run test on ErasureSet backend.

cmd/admin-handlers-users.go

@@ -241,6 +241,15 @@ func (a adminAPIHandlers) UpdateGroupMembers(w http.ResponseWriter, r *http.Requ
 	if updReq.IsRemove {
 		err = globalIAMSys.RemoveUsersFromGroup(ctx, updReq.Group, updReq.Members)
 	} else {
+		// Check if group already exists
+		if _, gerr := globalIAMSys.GetGroupDescription(updReq.Group); gerr != nil {
+			// If group does not exist, then check if the group has beginning and end space characters
+			// we will reject such group names.
+			if errors.Is(gerr, errNoSuchGroup) && hasSpaceBE(updReq.Group) {
+				writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminResourceInvalidArgument), r.URL)
+				return
+			}
+		}
 		err = globalIAMSys.AddUsersToGroup(ctx, updReq.Group, updReq.Members)
 	}
 	if err != nil {
@@ -442,6 +451,12 @@ func (a adminAPIHandlers) AddUser(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	// Check if accessKey has beginning and end space characters, this only applies to new users.
+	if !exists && hasSpaceBE(accessKey) {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminResourceInvalidArgument), r.URL)
+		return
+	}
+
 	checkDenyOnly := false
 	if accessKey == cred.AccessKey {
 		// Check that there is no explicit deny - otherwise it's allowed
@@ -533,6 +548,12 @@ func (a adminAPIHandlers) AddServiceAccount(w http.ResponseWriter, r *http.Reque
 		return
 	}
 
+	// service account access key cannot have space characters beginning and end of the string.
+	if hasSpaceBE(createReq.AccessKey) {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminResourceInvalidArgument), r.URL)
+		return
+	}
+
 	var (
 		targetUser   string
 		targetGroups []string
@@ -656,29 +677,6 @@ func (a adminAPIHandlers) AddServiceAccount(w http.ResponseWriter, r *http.Reque
 		return
 	}
 
-	// Call hook for cluster-replication if the service account is not for a
-	// root user.
-	if newCred.ParentUser != globalActiveCred.AccessKey {
-		err = globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
-			Type: madmin.SRIAMItemSvcAcc,
-			SvcAccChange: &madmin.SRSvcAccChange{
-				Create: &madmin.SRSvcAccCreate{
-					Parent:        newCred.ParentUser,
-					AccessKey:     newCred.AccessKey,
-					SecretKey:     newCred.SecretKey,
-					Groups:        newCred.Groups,
-					Claims:        opts.claims,
-					SessionPolicy: createReq.Policy,
-					Status:        auth.AccountOn,
-				},
-			},
-		})
-		if err != nil {
-			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-			return
-		}
-	}
-
 	createResp := madmin.AddServiceAccountResp{
 		Credentials: madmin.Credentials{
 			AccessKey: newCred.AccessKey,
@@ -699,6 +697,29 @@ func (a adminAPIHandlers) AddServiceAccount(w http.ResponseWriter, r *http.Reque
 	}
 
 	writeSuccessResponseJSON(w, encryptedData)
+
+	// Call hook for cluster-replication if the service account is not for a
+	// root user.
+	if newCred.ParentUser != globalActiveCred.AccessKey {
+		err = globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+			Type: madmin.SRIAMItemSvcAcc,
+			SvcAccChange: &madmin.SRSvcAccChange{
+				Create: &madmin.SRSvcAccCreate{
+					Parent:        newCred.ParentUser,
+					AccessKey:     newCred.AccessKey,
+					SecretKey:     newCred.SecretKey,
+					Groups:        newCred.Groups,
+					Claims:        opts.claims,
+					SessionPolicy: createReq.Policy,
+					Status:        auth.AccountOn,
+				},
+			},
+		})
+		if err != nil {
+			logger.LogIf(ctx, err)
+			return
+		}
+	}
 }
 
 // UpdateServiceAccount - POST /minio/admin/v3/update-service-account
@@ -993,11 +1014,9 @@ func (a adminAPIHandlers) DeleteServiceAccount(w http.ResponseWriter, r *http.Re
 		return
 	}
 
-	svcAccount, _, err := globalIAMSys.GetServiceAccount(ctx, serviceAccount)
-	if err != nil {
-		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-		return
-	}
+	// We do not care if service account is readable or not at this point,
+	// since this is a delete call we shall allow it to be deleted if possible.
+	svcAccount, _, _ := globalIAMSys.GetServiceAccount(ctx, serviceAccount)
 
 	adminPrivilege := globalIAMSys.IsAllowed(iampolicy.Args{
 		AccountName:     cred.AccessKey,
@@ -1012,7 +1031,7 @@ func (a adminAPIHandlers) DeleteServiceAccount(w http.ResponseWriter, r *http.Re
 		if cred.ParentUser != "" {
 			parentUser = cred.ParentUser
 		}
-		if parentUser != svcAccount.ParentUser {
+		if svcAccount.ParentUser != "" && parentUser != svcAccount.ParentUser {
 			// The service account belongs to another user but return not
 			// found error to mitigate brute force attacks. or the
 			// serviceAccount doesn't exist.
@@ -1021,23 +1040,21 @@ func (a adminAPIHandlers) DeleteServiceAccount(w http.ResponseWriter, r *http.Re
 		}
 	}
 
-	err = globalIAMSys.DeleteServiceAccount(ctx, serviceAccount, true)
-	if err != nil {
+	if err := globalIAMSys.DeleteServiceAccount(ctx, serviceAccount, true); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
 
 	// Call site replication hook - non-root user accounts are replicated.
-	if svcAccount.ParentUser != globalActiveCred.AccessKey {
-		err = globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+	if svcAccount.ParentUser != "" && svcAccount.ParentUser != globalActiveCred.AccessKey {
+		if err := globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
 			Type: madmin.SRIAMItemSvcAcc,
 			SvcAccChange: &madmin.SRSvcAccChange{
 				Delete: &madmin.SRSvcAccDelete{
 					AccessKey: serviceAccount,
 				},
 			},
-		})
-		if err != nil {
+		}); err != nil {
 			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 			return
 		}
@@ -1088,6 +1105,19 @@ func (a adminAPIHandlers) AccountInfoHandler(w http.ResponseWriter, r *http.Requ
 			rd = true
 		}
 
+		if globalIAMSys.IsAllowed(iampolicy.Args{
+			AccountName:     cred.AccessKey,
+			Groups:          cred.Groups,
+			Action:          iampolicy.GetBucketLocationAction,
+			BucketName:      bucketName,
+			ConditionValues: getConditionValues(r, "", cred.AccessKey, claims),
+			IsOwner:         owner,
+			ObjectName:      "",
+			Claims:          claims,
+		}) {
+			rd = true
+		}
+
 		if globalIAMSys.IsAllowed(iampolicy.Args{
 			AccountName:     cred.AccessKey,
 			Groups:          cred.Groups,
@@ -1384,6 +1414,12 @@ func (a adminAPIHandlers) AddCannedPolicy(w http.ResponseWriter, r *http.Request
 	vars := mux.Vars(r)
 	policyName := vars["name"]
 
+	// Policy has space characters in begin and end reject such inputs.
+	if hasSpaceBE(policyName) {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminResourceInvalidArgument), r.URL)
+		return
+	}
+
 	// Error out if Content-Length is missing.
 	if r.ContentLength <= 0 {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrMissingContentLength), r.URL)

cmd/admin-handlers-users_test.go

@@ -28,6 +28,7 @@ import (
 	"net/http"
 	"net/url"
 	"os"
+	"runtime"
 	"strings"
 	"testing"
 	"time"
@@ -97,6 +98,29 @@ func (s *TestSuiteIAM) iamSetup(c *check) {
 	}
 }
 
+// List of all IAM test suites (i.e. test server configuration combinations)
+// common to tests.
+var iamTestSuites = func() []*TestSuiteIAM {
+	baseTestCases := []TestSuiteCommon{
+		// Init and run test on ErasureSD backend with signature v4.
+		{serverType: "ErasureSD", signer: signerV4},
+		// Init and run test on ErasureSD backend, with tls enabled.
+		{serverType: "ErasureSD", signer: signerV4, secure: true},
+		// Init and run test on Erasure backend.
+		{serverType: "Erasure", signer: signerV4},
+		// Init and run test on ErasureSet backend.
+		{serverType: "ErasureSet", signer: signerV4},
+	}
+	testCases := []*TestSuiteIAM{}
+	for _, bt := range baseTestCases {
+		testCases = append(testCases,
+			newTestSuiteIAM(bt, false),
+			newTestSuiteIAM(bt, true),
+		)
+	}
+	return testCases
+}()
+
 const (
 	EnvTestEtcdBackend = "ETCD_SERVER"
 )
@@ -166,30 +190,12 @@ func (s *TestSuiteIAM) getUserClient(c *check, accessKey, secretKey, sessionToke
 }
 
 func TestIAMInternalIDPServerSuite(t *testing.T) {
-	baseTestCases := []TestSuiteCommon{
-		// Init and run test on FS backend with signature v4.
-		{serverType: "FS", signer: signerV4},
-		// Init and run test on FS backend, with tls enabled.
-		{serverType: "FS", signer: signerV4, secure: true},
-		// Init and run test on Erasure backend.
-		{serverType: "Erasure", signer: signerV4},
-		// Init and run test on ErasureSet backend.
-		{serverType: "ErasureSet", signer: signerV4},
+	if runtime.GOOS == globalWindowsOSName {
+		t.Skip("windows is clunky disable these tests")
 	}
-	testCases := []*TestSuiteIAM{}
-	for _, bt := range baseTestCases {
-		testCases = append(testCases,
-			newTestSuiteIAM(bt, false),
-			newTestSuiteIAM(bt, true),
-		)
-	}
-	for i, testCase := range testCases {
-		etcdStr := ""
-		if testCase.withEtcdBackend {
-			etcdStr = " (with etcd backend)"
-		}
+	for i, testCase := range iamTestSuites {
 		t.Run(
-			fmt.Sprintf("Test: %d, ServerType: %s%s", i+1, testCase.serverType, etcdStr),
+			fmt.Sprintf("Test: %d, ServerType: %s", i+1, testCase.ServerTypeDescription),
 			func(t *testing.T) {
 				suite := testCase
 				c := &check{t, testCase.serverType}
@@ -236,6 +242,7 @@ func (s *TestSuiteIAM) TestUserCreate(c *check) {
 	if err != nil {
 		c.Fatalf("unable to set policy: %v", err)
 	}
+
 	client := s.getUserClient(c, accessKey, secretKey, "")
 	err = client.MakeBucket(ctx, getRandomBucketName(), minio.MakeBucketOptions{})
 	if err != nil {
@@ -973,6 +980,166 @@ func (s *TestSuiteIAM) TestServiceAccountOpsByAdmin(c *check) {
 	c.assertSvcAccDeletion(ctx, s, s.adm, accessKey, bucket)
 }
 
+func (s *TestSuiteIAM) SetUpAccMgmtPlugin(c *check) {
+	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
+	defer cancel()
+
+	pluginEndpoint := os.Getenv("POLICY_PLUGIN_ENDPOINT")
+	if pluginEndpoint == "" {
+		c.Skip("POLICY_PLUGIN_ENDPOINT not given - skipping.")
+	}
+
+	configCmds := []string{
+		"policy_plugin",
+		"url=" + pluginEndpoint,
+	}
+
+	_, err := s.adm.SetConfigKV(ctx, strings.Join(configCmds, " "))
+	if err != nil {
+		c.Fatalf("unable to setup access management plugin for tests: %v", err)
+	}
+
+	s.RestartIAMSuite(c)
+}
+
+// TestIAM_AMPInternalIDPServerSuite - tests for access management plugin
+func TestIAM_AMPInternalIDPServerSuite(t *testing.T) {
+	for i, testCase := range iamTestSuites {
+		t.Run(
+			fmt.Sprintf("Test: %d, ServerType: %s", i+1, testCase.ServerTypeDescription),
+			func(t *testing.T) {
+				suite := testCase
+				c := &check{t, testCase.serverType}
+
+				suite.SetUpSuite(c)
+				defer suite.TearDownSuite(c)
+
+				suite.SetUpAccMgmtPlugin(c)
+
+				suite.TestAccMgmtPlugin(c)
+			},
+		)
+	}
+}
+
+// TestAccMgmtPlugin - this test assumes that the access-management-plugin is
+// the same as the example in `docs/iam/access-manager-plugin.go` -
+// specifically, it denies only `s3:Put*` operations on non-root accounts.
+func (s *TestSuiteIAM) TestAccMgmtPlugin(c *check) {
+	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
+	defer cancel()
+
+	// 0. Check that owner is able to make-bucket.
+	bucket := getRandomBucketName()
+	err := s.client.MakeBucket(ctx, bucket, minio.MakeBucketOptions{})
+	if err != nil {
+		c.Fatalf("bucket creat error: %v", err)
+	}
+
+	// 1. Create a user.
+	accessKey, secretKey := mustGenerateCredentials(c)
+	err = s.adm.SetUser(ctx, accessKey, secretKey, madmin.AccountEnabled)
+	if err != nil {
+		c.Fatalf("Unable to set user: %v", err)
+	}
+
+	// 2. Check new user appears in listing
+	usersMap, err := s.adm.ListUsers(ctx)
+	if err != nil {
+		c.Fatalf("error listing: %v", err)
+	}
+	v, ok := usersMap[accessKey]
+	if !ok {
+		c.Fatalf("user not listed: %s", accessKey)
+	}
+	c.Assert(v.Status, madmin.AccountEnabled)
+
+	// 3. Check that user is able to make a bucket.
+	client := s.getUserClient(c, accessKey, secretKey, "")
+	err = client.MakeBucket(ctx, getRandomBucketName(), minio.MakeBucketOptions{})
+	if err != nil {
+		c.Fatalf("user not create bucket: %v", err)
+	}
+
+	// 3.1 check user has access to bucket
+	c.mustListObjects(ctx, client, bucket)
+
+	// 3.2 check that user cannot upload an object.
+	_, err = client.PutObject(ctx, bucket, "objectName", bytes.NewBuffer([]byte("some content")), 12, minio.PutObjectOptions{})
+	if err == nil {
+		c.Fatalf("user was able to upload unexpectedly")
+	}
+
+	// Create an madmin client with user creds
+	userAdmClient, err := madmin.NewWithOptions(s.endpoint, &madmin.Options{
+		Creds:  cr.NewStaticV4(accessKey, secretKey, ""),
+		Secure: s.secure,
+	})
+	if err != nil {
+		c.Fatalf("Err creating user admin client: %v", err)
+	}
+	userAdmClient.SetCustomTransport(s.TestSuiteCommon.client.Transport)
+
+	// Create svc acc
+	cr := c.mustCreateSvcAccount(ctx, accessKey, userAdmClient)
+
+	// 1. Check that svc account appears in listing
+	c.assertSvcAccAppearsInListing(ctx, userAdmClient, accessKey, cr.AccessKey)
+
+	// 2. Check that svc account info can be queried
+	c.assertSvcAccInfoQueryable(ctx, userAdmClient, accessKey, cr.AccessKey, false)
+
+	// 3. Check S3 access
+	c.assertSvcAccS3Access(ctx, s, cr, bucket)
+
+	// Check that session policies do not apply - as policy enforcement is
+	// delegated to plugin.
+	{
+		svcAK, svcSK := mustGenerateCredentials(c)
+
+		// This policy does not allow listing objects.
+		policyBytes := []byte(fmt.Sprintf(`{
+ "Version": "2012-10-17",
+ "Statement": [
+  {
+   "Effect": "Allow",
+   "Action": [
+    "s3:PutObject",
+    "s3:GetObject"
+   ],
+   "Resource": [
+    "arn:aws:s3:::%s/*"
+   ]
+  }
+ ]
+}`, bucket))
+		cr, err := userAdmClient.AddServiceAccount(ctx, madmin.AddServiceAccountReq{
+			Policy:     policyBytes,
+			TargetUser: accessKey,
+			AccessKey:  svcAK,
+			SecretKey:  svcSK,
+		})
+		if err != nil {
+			c.Fatalf("Unable to create svc acc: %v", err)
+		}
+		svcClient := s.getUserClient(c, cr.AccessKey, cr.SecretKey, "")
+		// Though the attached policy does not allow listing, it will be
+		// ignored because the plugin allows it.
+		c.mustListObjects(ctx, svcClient, bucket)
+	}
+
+	// 4. Check that service account's secret key and account status can be
+	// updated.
+	c.assertSvcAccSecretKeyAndStatusUpdate(ctx, s, userAdmClient, accessKey, bucket)
+
+	// 5. Check that service account can be deleted.
+	c.assertSvcAccDeletion(ctx, s, userAdmClient, accessKey, bucket)
+
+	// 6. Check that service account **can** be created for some other user.
+	// This is possible because of the policy enforced in the plugin.
+	c.mustCreateSvcAccount(ctx, globalActiveCred.AccessKey, userAdmClient)
+}
+
 func (c *check) mustCreateIAMUser(ctx context.Context, admClnt *madmin.AdminClient) madmin.Credentials {
 	randUser := mustGetUUID()
 	randPass := mustGetUUID()
@@ -1026,7 +1193,7 @@ func (c *check) mustNotListObjects(ctx context.Context, client *minio.Client, bu
 	res := client.ListObjects(ctx, bucket, minio.ListObjectsOptions{})
 	v, ok := <-res
 	if !ok || v.Err == nil {
-		c.Fatalf("user was able to list unexpectedly!")
+		c.Fatalf("user was able to list unexpectedly! on %s", bucket)
 	}
 }
 
@@ -1034,11 +1201,20 @@ func (c *check) mustListObjects(ctx context.Context, client *minio.Client, bucke
 	res := client.ListObjects(ctx, bucket, minio.ListObjectsOptions{})
 	v, ok := <-res
 	if ok && v.Err != nil {
-		msg := fmt.Sprintf("user was unable to list: %v", v.Err)
-		c.Fatalf(msg)
+		c.Fatalf("user was unable to list: %v", v.Err)
 	}
 }
 
+func (c *check) mustNotUpload(ctx context.Context, client *minio.Client, bucket string) {
+	_, err := client.PutObject(ctx, bucket, "some-object", bytes.NewBuffer([]byte("stuff")), 5, minio.PutObjectOptions{})
+	if e, ok := err.(minio.ErrorResponse); ok {
+		if e.Code == "AccessDenied" {
+			return
+		}
+	}
+	c.Fatalf("upload did not get an AccessDenied error - got %#v instead", err)
+}
+
 func (c *check) assertSvcAccS3Access(ctx context.Context, s *TestSuiteIAM, cr madmin.Credentials, bucket string) {
 	svcClient := s.getUserClient(c, cr.AccessKey, cr.SecretKey, "")
 	c.mustListObjects(ctx, svcClient, bucket)

cmd/admin-handlers.go

@@ -57,7 +57,8 @@ import (
 )
 
 const (
-	maxEConfigJSONSize = 262272
+	maxEConfigJSONSize        = 262272
+	kubernetesVersionEndpoint = "https://kubernetes.default.svc/version"
 )
 
 // Only valid query params for mgmt admin APIs.
@@ -653,12 +654,10 @@ func (a adminAPIHandlers) ProfileHandler(w http.ResponseWriter, r *http.Request)
 	globalProfilerMu.Unlock()
 
 	timer := time.NewTimer(duration)
+	defer timer.Stop()
 	for {
 		select {
 		case <-ctx.Done():
-			if !timer.Stop() {
-				<-timer.C
-			}
 			for k, v := range globalProfiler {
 				v.Stop()
 				delete(globalProfiler, k)
@@ -806,8 +805,7 @@ func (a adminAPIHandlers) HealHandler(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	// Check if this setup has an erasure coded backend.
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrHealNotImplemented), r.URL)
 		return
 	}
@@ -999,7 +997,7 @@ func (a adminAPIHandlers) BackgroundHealStatusHandler(w http.ResponseWriter, r *
 	}
 
 	// Check if this setup has an erasure coded backend.
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrHealNotImplemented), r.URL)
 		return
 	}
@@ -1079,7 +1077,7 @@ func (a adminAPIHandlers) ObjectSpeedtestHandler(w http.ResponseWriter, r *http.
 		return
 	}
 
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
 		return
 	}
@@ -1229,7 +1227,7 @@ func (a adminAPIHandlers) DriveSpeedtestHandler(w http.ResponseWriter, r *http.R
 		return
 	}
 
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
 		return
 	}
@@ -1411,9 +1409,15 @@ func (a adminAPIHandlers) TraceHandler(w http.ResponseWriter, r *http.Request) {
 
 	peers, _ := newPeerRestClients(globalEndpoints)
 
-	globalTrace.Subscribe(traceCh, ctx.Done(), func(entry interface{}) bool {
+	traceFn := func(entry interface{}) bool {
 		return mustTrace(entry, traceOpts)
-	})
+	}
+
+	err = globalTrace.Subscribe(traceCh, ctx.Done(), traceFn)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrSlowDown), r.URL)
+		return
+	}
 
 	for _, peer := range peers {
 		if peer == nil {
@@ -1485,7 +1489,11 @@ func (a adminAPIHandlers) ConsoleLogHandler(w http.ResponseWriter, r *http.Reque
 
 	peers, _ := newPeerRestClients(globalEndpoints)
 
-	globalConsoleSys.Subscribe(logCh, ctx.Done(), node, limitLines, logKind, nil)
+	err = globalConsoleSys.Subscribe(logCh, ctx.Done(), node, limitLines, logKind, nil)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrSlowDown), r.URL)
+		return
+	}
 
 	for _, peer := range peers {
 		if peer == nil {
@@ -1768,6 +1776,36 @@ func getServerInfo(ctx context.Context, r *http.Request) madmin.InfoMessage {
 	}
 }
 
+func getKubernetesInfo(dctx context.Context) madmin.KubernetesInfo {
+	ctx, cancel := context.WithCancel(dctx)
+	defer cancel()
+
+	ki := madmin.KubernetesInfo{}
+
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, kubernetesVersionEndpoint, nil)
+	if err != nil {
+		ki.Error = err.Error()
+		return ki
+	}
+
+	client := &http.Client{
+		Transport: NewGatewayHTTPTransport(),
+		Timeout:   10 * time.Second,
+	}
+
+	resp, err := client.Do(req)
+	if err != nil {
+		ki.Error = err.Error()
+		return ki
+	}
+
+	decoder := json.NewDecoder(resp.Body)
+	if err := decoder.Decode(&ki); err != nil {
+		ki.Error = err.Error()
+	}
+	return ki
+}
+
 // HealthInfoHandler - GET /minio/admin/v3/healthinfo
 // ----------
 // Get server health info
@@ -1793,9 +1831,6 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 	healthInfoCh := make(chan madmin.HealthInfo)
 
 	enc := json.NewEncoder(w)
-	partialWrite := func(oinfo madmin.HealthInfo) {
-		healthInfoCh <- oinfo
-	}
 
 	setCommonHeaders(w)
 
@@ -1826,9 +1861,6 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 		}
 	}
 
-	deadlinedCtx, deadlineCancel := context.WithTimeout(ctx, deadline)
-	defer deadlineCancel()
-
 	nsLock := objectAPI.NewNSLock(minioMetaBucket, "health-check-in-progress")
 	lkctx, err := nsLock.GetLock(ctx, newDynamicTimeout(deadline, deadline))
 	if err != nil { // returns a locked lock
@@ -1837,6 +1869,9 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 	}
 	defer nsLock.Unlock(lkctx.Cancel)
 
+	healthCtx, healthCancel := context.WithTimeout(lkctx.Context(), deadline)
+	defer healthCancel()
+
 	hostAnonymizer := createHostAnonymizer()
 	// anonAddr - Anonymizes hosts in given input string.
 	anonAddr := func(addr string) string {
@@ -1855,13 +1890,27 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 		info.SetAddr(anonAddr(info.GetAddr()))
 	}
 
+	partialWrite := func(oinfo madmin.HealthInfo) {
+		select {
+		case healthInfoCh <- oinfo:
+		case <-healthCtx.Done():
+		}
+	}
+
+	getAndWritePlatformInfo := func() {
+		if IsKubernetes() {
+			healthInfo.Sys.KubernetesInfo = getKubernetesInfo(healthCtx)
+			partialWrite(healthInfo)
+		}
+	}
+
 	getAndWriteCPUs := func() {
 		if query.Get("syscpu") == "true" {
-			localCPUInfo := madmin.GetCPUs(deadlinedCtx, globalLocalNodeName)
+			localCPUInfo := madmin.GetCPUs(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localCPUInfo)
 			healthInfo.Sys.CPUInfo = append(healthInfo.Sys.CPUInfo, localCPUInfo)
 
-			peerCPUInfo := globalNotificationSys.GetCPUs(deadlinedCtx)
+			peerCPUInfo := globalNotificationSys.GetCPUs(healthCtx)
 			for _, cpuInfo := range peerCPUInfo {
 				anonymizeAddr(&cpuInfo)
 				healthInfo.Sys.CPUInfo = append(healthInfo.Sys.CPUInfo, cpuInfo)
@@ -1873,11 +1922,11 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWritePartitions := func() {
 		if query.Get("sysdrivehw") == "true" {
-			localPartitions := madmin.GetPartitions(deadlinedCtx, globalLocalNodeName)
+			localPartitions := madmin.GetPartitions(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localPartitions)
 			healthInfo.Sys.Partitions = append(healthInfo.Sys.Partitions, localPartitions)
 
-			peerPartitions := globalNotificationSys.GetPartitions(deadlinedCtx)
+			peerPartitions := globalNotificationSys.GetPartitions(healthCtx)
 			for _, p := range peerPartitions {
 				anonymizeAddr(&p)
 				healthInfo.Sys.Partitions = append(healthInfo.Sys.Partitions, p)
@@ -1888,11 +1937,11 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWriteOSInfo := func() {
 		if query.Get("sysosinfo") == "true" {
-			localOSInfo := madmin.GetOSInfo(deadlinedCtx, globalLocalNodeName)
+			localOSInfo := madmin.GetOSInfo(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localOSInfo)
 			healthInfo.Sys.OSInfo = append(healthInfo.Sys.OSInfo, localOSInfo)
 
-			peerOSInfos := globalNotificationSys.GetOSInfo(deadlinedCtx)
+			peerOSInfos := globalNotificationSys.GetOSInfo(healthCtx)
 			for _, o := range peerOSInfos {
 				anonymizeAddr(&o)
 				healthInfo.Sys.OSInfo = append(healthInfo.Sys.OSInfo, o)
@@ -1903,11 +1952,11 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWriteMemInfo := func() {
 		if query.Get("sysmem") == "true" {
-			localMemInfo := madmin.GetMemInfo(deadlinedCtx, globalLocalNodeName)
+			localMemInfo := madmin.GetMemInfo(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localMemInfo)
 			healthInfo.Sys.MemInfo = append(healthInfo.Sys.MemInfo, localMemInfo)
 
-			peerMemInfos := globalNotificationSys.GetMemInfo(deadlinedCtx)
+			peerMemInfos := globalNotificationSys.GetMemInfo(healthCtx)
 			for _, m := range peerMemInfos {
 				anonymizeAddr(&m)
 				healthInfo.Sys.MemInfo = append(healthInfo.Sys.MemInfo, m)
@@ -1918,12 +1967,12 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWriteSysErrors := func() {
 		if query.Get(string(madmin.HealthDataTypeSysErrors)) == "true" {
-			localSysErrors := madmin.GetSysErrors(deadlinedCtx, globalLocalNodeName)
+			localSysErrors := madmin.GetSysErrors(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localSysErrors)
 			healthInfo.Sys.SysErrs = append(healthInfo.Sys.SysErrs, localSysErrors)
 			partialWrite(healthInfo)
 
-			peerSysErrs := globalNotificationSys.GetSysErrors(deadlinedCtx)
+			peerSysErrs := globalNotificationSys.GetSysErrors(healthCtx)
 			for _, se := range peerSysErrs {
 				anonymizeAddr(&se)
 				healthInfo.Sys.SysErrs = append(healthInfo.Sys.SysErrs, se)
@@ -1934,12 +1983,12 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWriteSysConfig := func() {
 		if query.Get(string(madmin.HealthDataTypeSysConfig)) == "true" {
-			localSysConfig := madmin.GetSysConfig(deadlinedCtx, globalLocalNodeName)
+			localSysConfig := madmin.GetSysConfig(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localSysConfig)
 			healthInfo.Sys.SysConfig = append(healthInfo.Sys.SysConfig, localSysConfig)
 			partialWrite(healthInfo)
 
-			peerSysConfig := globalNotificationSys.GetSysConfig(deadlinedCtx)
+			peerSysConfig := globalNotificationSys.GetSysConfig(healthCtx)
 			for _, sc := range peerSysConfig {
 				anonymizeAddr(&sc)
 				healthInfo.Sys.SysConfig = append(healthInfo.Sys.SysConfig, sc)
@@ -1950,12 +1999,12 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWriteSysServices := func() {
 		if query.Get(string(madmin.HealthDataTypeSysServices)) == "true" {
-			localSysServices := madmin.GetSysServices(deadlinedCtx, globalLocalNodeName)
+			localSysServices := madmin.GetSysServices(healthCtx, globalLocalNodeName)
 			anonymizeAddr(&localSysServices)
 			healthInfo.Sys.SysServices = append(healthInfo.Sys.SysServices, localSysServices)
 			partialWrite(healthInfo)
 
-			peerSysServices := globalNotificationSys.GetSysServices(deadlinedCtx)
+			peerSysServices := globalNotificationSys.GetSysServices(healthCtx)
 			for _, ss := range peerSysServices {
 				anonymizeAddr(&ss)
 				healthInfo.Sys.SysServices = append(healthInfo.Sys.SysServices, ss)
@@ -2030,10 +2079,10 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 
 	getAndWriteProcInfo := func() {
 		if query.Get("sysprocess") == "true" {
-			localProcInfo := madmin.GetProcInfo(deadlinedCtx, globalLocalNodeName)
+			localProcInfo := madmin.GetProcInfo(healthCtx, globalLocalNodeName)
 			anonymizeProcInfo(&localProcInfo)
 			healthInfo.Sys.ProcInfo = append(healthInfo.Sys.ProcInfo, localProcInfo)
-			peerProcInfos := globalNotificationSys.GetProcInfo(deadlinedCtx)
+			peerProcInfos := globalNotificationSys.GetProcInfo(healthCtx)
 			for _, p := range peerProcInfos {
 				anonymizeProcInfo(&p)
 				healthInfo.Sys.ProcInfo = append(healthInfo.Sys.ProcInfo, p)
@@ -2177,6 +2226,7 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 		defer close(healthInfoCh)
 
 		partialWrite(healthInfo) // Write first message with only version and deployment id populated
+		getAndWritePlatformInfo()
 		getAndWriteCPUs()
 		getAndWritePartitions()
 		getAndWriteOSInfo()
@@ -2258,7 +2308,7 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 				return
 			}
 			w.(http.Flusher).Flush()
-		case <-deadlinedCtx.Done():
+		case <-healthCtx.Done():
 			return
 		}
 	}

cmd/admin-heal-ops.go

@@ -194,7 +194,6 @@ func (ahs *allHealState) periodicHealSeqsClean(ctx context.Context) {
 	for {
 		select {
 		case <-periodicTimer.C:
-			periodicTimer.Reset(time.Minute * 5)
 			now := UTCNow()
 			ahs.Lock()
 			for path, h := range ahs.healSeqMap {
@@ -203,6 +202,8 @@ func (ahs *allHealState) periodicHealSeqsClean(ctx context.Context) {
 				}
 			}
 			ahs.Unlock()
+
+			periodicTimer.Reset(time.Minute * 5)
 		case <-ctx.Done():
 			// server could be restarting - need
 			// to exit immediately
@@ -581,12 +582,7 @@ func (h *healSequence) pushHealResultItem(r madmin.HealResultItem) error {
 	// heal-results in memory and the client has not consumed it
 	// for too long.
 	unconsumedTimer := time.NewTimer(healUnconsumedTimeout)
-	defer func() {
-		// stop the timeout timer so it is garbage collected.
-		if !unconsumedTimer.Stop() {
-			<-unconsumedTimer.C
-		}
-	}()
+	defer unconsumedTimer.Stop()
 
 	var itemsLen int
 	for {

cmd/admin-router.go

@@ -170,50 +170,48 @@ func registerAdminRouter(router *mux.Router, enableConfigOps bool) {
 		// Set Group Status
 		adminRouter.Methods(http.MethodPut).Path(adminVersion+"/set-group-status").HandlerFunc(gz(httpTraceHdrs(adminAPI.SetGroupStatus))).Queries("group", "{group:.*}").Queries("status", "{status:.*}")
 
-		if globalIsDistErasure || globalIsErasure {
-			// GetBucketQuotaConfig
-			adminRouter.Methods(http.MethodGet).Path(adminVersion+"/get-bucket-quota").HandlerFunc(
-				gz(httpTraceHdrs(adminAPI.GetBucketQuotaConfigHandler))).Queries("bucket", "{bucket:.*}")
-			// PutBucketQuotaConfig
-			adminRouter.Methods(http.MethodPut).Path(adminVersion+"/set-bucket-quota").HandlerFunc(
-				gz(httpTraceHdrs(adminAPI.PutBucketQuotaConfigHandler))).Queries("bucket", "{bucket:.*}")
+		// GetBucketQuotaConfig
+		adminRouter.Methods(http.MethodGet).Path(adminVersion+"/get-bucket-quota").HandlerFunc(
+			gz(httpTraceHdrs(adminAPI.GetBucketQuotaConfigHandler))).Queries("bucket", "{bucket:.*}")
+		// PutBucketQuotaConfig
+		adminRouter.Methods(http.MethodPut).Path(adminVersion+"/set-bucket-quota").HandlerFunc(
+			gz(httpTraceHdrs(adminAPI.PutBucketQuotaConfigHandler))).Queries("bucket", "{bucket:.*}")
 
-			// Bucket replication operations
-			// GetBucketTargetHandler
-			adminRouter.Methods(http.MethodGet).Path(adminVersion+"/list-remote-targets").HandlerFunc(
-				gz(httpTraceHdrs(adminAPI.ListRemoteTargetsHandler))).Queries("bucket", "{bucket:.*}", "type", "{type:.*}")
-			// SetRemoteTargetHandler
-			adminRouter.Methods(http.MethodPut).Path(adminVersion+"/set-remote-target").HandlerFunc(
-				gz(httpTraceHdrs(adminAPI.SetRemoteTargetHandler))).Queries("bucket", "{bucket:.*}")
-			// RemoveRemoteTargetHandler
-			adminRouter.Methods(http.MethodDelete).Path(adminVersion+"/remove-remote-target").HandlerFunc(
-				gz(httpTraceHdrs(adminAPI.RemoveRemoteTargetHandler))).Queries("bucket", "{bucket:.*}", "arn", "{arn:.*}")
+		// Bucket replication operations
+		// GetBucketTargetHandler
+		adminRouter.Methods(http.MethodGet).Path(adminVersion+"/list-remote-targets").HandlerFunc(
+			gz(httpTraceHdrs(adminAPI.ListRemoteTargetsHandler))).Queries("bucket", "{bucket:.*}", "type", "{type:.*}")
+		// SetRemoteTargetHandler
+		adminRouter.Methods(http.MethodPut).Path(adminVersion+"/set-remote-target").HandlerFunc(
+			gz(httpTraceHdrs(adminAPI.SetRemoteTargetHandler))).Queries("bucket", "{bucket:.*}")
+		// RemoveRemoteTargetHandler
+		adminRouter.Methods(http.MethodDelete).Path(adminVersion+"/remove-remote-target").HandlerFunc(
+			gz(httpTraceHdrs(adminAPI.RemoveRemoteTargetHandler))).Queries("bucket", "{bucket:.*}", "arn", "{arn:.*}")
 
-			// Remote Tier management operations
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/tier").HandlerFunc(gz(httpTraceHdrs(adminAPI.AddTierHandler)))
-			adminRouter.Methods(http.MethodPost).Path(adminVersion + "/tier/{tier}").HandlerFunc(gz(httpTraceHdrs(adminAPI.EditTierHandler)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/tier").HandlerFunc(gz(httpTraceHdrs(adminAPI.ListTierHandler)))
-			adminRouter.Methods(http.MethodDelete).Path(adminVersion + "/tier/{tier}").HandlerFunc(gz(httpTraceHdrs(adminAPI.RemoveTierHandler)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/tier/{tier}").HandlerFunc(gz(httpTraceHdrs(adminAPI.VerifyTierHandler)))
-			// Tier stats
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/tier-stats").HandlerFunc(gz(httpTraceHdrs(adminAPI.TierStatsHandler)))
+		// Remote Tier management operations
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/tier").HandlerFunc(gz(httpTraceHdrs(adminAPI.AddTierHandler)))
+		adminRouter.Methods(http.MethodPost).Path(adminVersion + "/tier/{tier}").HandlerFunc(gz(httpTraceHdrs(adminAPI.EditTierHandler)))
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/tier").HandlerFunc(gz(httpTraceHdrs(adminAPI.ListTierHandler)))
+		adminRouter.Methods(http.MethodDelete).Path(adminVersion + "/tier/{tier}").HandlerFunc(gz(httpTraceHdrs(adminAPI.RemoveTierHandler)))
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/tier/{tier}").HandlerFunc(gz(httpTraceHdrs(adminAPI.VerifyTierHandler)))
+		// Tier stats
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/tier-stats").HandlerFunc(gz(httpTraceHdrs(adminAPI.TierStatsHandler)))
 
-			// Cluster Replication APIs
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/add").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationAdd)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/remove").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationRemove)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/info").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationInfo)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/metainfo").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationMetaInfo)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/status").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationStatus)))
+		// Cluster Replication APIs
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/add").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationAdd)))
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/remove").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationRemove)))
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/info").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationInfo)))
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/metainfo").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationMetaInfo)))
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/status").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationStatus)))
 
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/join").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerJoin)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion+"/site-replication/peer/bucket-ops").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerBucketOps))).Queries("bucket", "{bucket:.*}").Queries("operation", "{operation:.*}")
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/iam-item").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerReplicateIAMItem)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/bucket-meta").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerReplicateBucketItem)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/peer/idp-settings").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerGetIDPSettings)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/edit").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationEdit)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/edit").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerEdit)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/remove").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerRemove)))
-		}
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/join").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerJoin)))
+		adminRouter.Methods(http.MethodPut).Path(adminVersion+"/site-replication/peer/bucket-ops").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerBucketOps))).Queries("bucket", "{bucket:.*}").Queries("operation", "{operation:.*}")
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/iam-item").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerReplicateIAMItem)))
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/bucket-meta").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerReplicateBucketItem)))
+		adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/peer/idp-settings").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerGetIDPSettings)))
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/edit").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationEdit)))
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/edit").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerEdit)))
+		adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/remove").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRPeerRemove)))
 
 		if globalIsDistErasure {
 			// Top locks

cmd/admin-server-info.go

@@ -31,7 +31,10 @@ import (
 // local endpoints from given list of endpoints
 func getLocalServerProperty(endpointServerPools EndpointServerPools, r *http.Request) madmin.ServerProperties {
 	var localEndpoints Endpoints
-	addr := r.Host
+	addr := globalLocalNodeName
+	if r != nil {
+		addr = r.Host
+	}
 	if globalIsDistErasure {
 		addr = globalLocalNodeName
 	}
@@ -40,7 +43,7 @@ func getLocalServerProperty(endpointServerPools EndpointServerPools, r *http.Req
 		for _, endpoint := range ep.Endpoints {
 			nodeName := endpoint.Host
 			if nodeName == "" {
-				nodeName = r.Host
+				nodeName = addr
 			}
 			if endpoint.IsLocal {
 				// Only proceed for local endpoints

cmd/api-errors.go

@@ -131,7 +131,7 @@ const (
 	ErrReplicationNeedsVersioningError
 	ErrReplicationBucketNeedsVersioningError
 	ErrReplicationDenyEditError
-	ErrReplicationNoMatchingRuleError
+	ErrReplicationNoExistingObjects
 	ErrObjectRestoreAlreadyInProgress
 	ErrNoSuchKey
 	ErrNoSuchUpload
@@ -280,6 +280,7 @@ const (
 	ErrSiteReplicationBucketConfigError
 	ErrSiteReplicationBucketMetaError
 	ErrSiteReplicationIAMError
+	ErrSiteReplicationConfigMissing
 
 	// Bucket Quota error codes
 	ErrAdminBucketQuotaExceeded
@@ -383,6 +384,7 @@ const (
 	ErrAdminProfilerNotEnabled
 	ErrInvalidDecompressedSize
 	ErrAddUserInvalidArgument
+	ErrAdminResourceInvalidArgument
 	ErrAdminAccountNotEligible
 	ErrAccountNotEligible
 	ErrAdminServiceAccountNotFound
@@ -891,15 +893,15 @@ var errorCodes = errorCodeMap{
 		Description:    "Bandwidth limit for remote target must be atleast 100MBps",
 		HTTPStatusCode: http.StatusBadRequest,
 	},
-	ErrReplicationNoMatchingRuleError: {
-		Code:           "XMinioReplicationNoMatchingRule",
-		Description:    "No matching replication rule found for this object prefix",
+	ErrReplicationNoExistingObjects: {
+		Code:           "XMinioReplicationNoExistingObjects",
+		Description:    "No matching ExistingsObjects rule enabled",
 		HTTPStatusCode: http.StatusBadRequest,
 	},
 	ErrReplicationDenyEditError: {
 		Code:           "XMinioReplicationDenyEdit",
 		Description:    "Cannot alter local replication config since this server is in a cluster replication setup",
-		HTTPStatusCode: http.StatusConflict,
+		HTTPStatusCode: http.StatusBadRequest,
 	},
 	ErrBucketRemoteIdenticalToSource: {
 		Code:           "XMinioAdminRemoteIdenticalToSource",
@@ -1165,7 +1167,7 @@ var errorCodes = errorCodeMap{
 	ErrObjectExistsAsDirectory: {
 		Code:           "XMinioObjectExistsAsDirectory",
 		Description:    "Object name already exists as a directory.",
-		HTTPStatusCode: http.StatusConflict,
+		HTTPStatusCode: http.StatusBadRequest,
 	},
 	ErrInvalidObjectName: {
 		Code:           "XMinioInvalidObjectName",
@@ -1339,7 +1341,11 @@ var errorCodes = errorCodeMap{
 		Description:    "Error while replicating an IAM item",
 		HTTPStatusCode: http.StatusServiceUnavailable,
 	},
-
+	ErrSiteReplicationConfigMissing: {
+		Code:           "XMinioSiteReplicationConfigMissingError",
+		Description:    "Site not found in site replication configuration",
+		HTTPStatusCode: http.StatusBadRequest,
+	},
 	ErrMaximumExpires: {
 		Code:           "AuthorizationQueryParametersError",
 		Description:    "X-Amz-Expires must be less than a week (in seconds); that is, the given X-Amz-Expires must be less than 604800 seconds",
@@ -1825,6 +1831,11 @@ var errorCodes = errorCodeMap{
 		Description:    "User is not allowed to be same as admin access key",
 		HTTPStatusCode: http.StatusForbidden,
 	},
+	ErrAdminResourceInvalidArgument: {
+		Code:           "XMinioInvalidResource",
+		Description:    "Policy, user or group names are not allowed to begin or end with space characters",
+		HTTPStatusCode: http.StatusBadRequest,
+	},
 	ErrAdminAccountNotEligible: {
 		Code:           "XMinioInvalidIAMCredentials",
 		Description:    "The administrator key is not eligible for this operation",

cmd/api-response.go

@@ -728,6 +728,14 @@ func generateMultiDeleteResponse(quiet bool, deletedObjects []DeletedObject, err
 }
 
 func writeResponse(w http.ResponseWriter, statusCode int, response []byte, mType mimeType) {
+	if statusCode == 0 {
+		statusCode = 200
+	}
+	// Similar check to http.checkWriteHeaderCode
+	if statusCode < 100 || statusCode > 999 {
+		logger.Error(fmt.Sprintf("invalid WriteHeader code %v", statusCode))
+		statusCode = http.StatusInternalServerError
+	}
 	setCommonHeaders(w)
 	if mType != mimeNone {
 		w.Header().Set(xhttp.ContentType, string(mType))
@@ -791,6 +799,12 @@ func writeErrorResponse(ctx context.Context, w http.ResponseWriter, err APIError
 		err.Description = fmt.Sprintf("The authorization header is malformed; the region is wrong; expecting '%s'.", globalSite.Region)
 	}
 
+	// Similar check to http.checkWriteHeaderCode
+	if err.HTTPStatusCode < 100 || err.HTTPStatusCode > 999 {
+		logger.Error(fmt.Sprintf("invalid WriteHeader code %v from %v", err.HTTPStatusCode, err.Code))
+		err.HTTPStatusCode = http.StatusInternalServerError
+	}
+
 	// Generate error response.
 	errorResponse := getAPIErrorResponse(ctx, err, reqURL.Path,
 		w.Header().Get(xhttp.AmzRequestID), globalDeploymentID)

cmd/api-router.go

@@ -342,7 +342,7 @@ func registerAPIRouter(router *mux.Router) {
 			collectAPIStats("getbucketnotification", maxClients(gz(httpTraceAll(api.GetBucketNotificationHandler))))).Queries("notification", "")
 		// ListenNotification
 		router.Methods(http.MethodGet).HandlerFunc(
-			collectAPIStats("listennotification", maxClients(gz(httpTraceAll(api.ListenNotificationHandler))))).Queries("events", "{events:.*}")
+			collectAPIStats("listennotification", gz(httpTraceAll(api.ListenNotificationHandler)))).Queries("events", "{events:.*}")
 		// ResetBucketReplicationStatus - MinIO extension API
 		router.Methods(http.MethodGet).HandlerFunc(
 			collectAPIStats("resetbucketreplicationstatus", maxClients(gz(httpTraceAll(api.ResetBucketReplicationStatusHandler))))).Queries("replication-reset-status", "")
@@ -474,7 +474,7 @@ func registerAPIRouter(router *mux.Router) {
 
 	// ListenNotification
 	apiRouter.Methods(http.MethodGet).Path(SlashSeparator).HandlerFunc(
-		collectAPIStats("listennotification", maxClients(gz(httpTraceAll(api.ListenNotificationHandler))))).Queries("events", "{events:.*}")
+		collectAPIStats("listennotification", gz(httpTraceAll(api.ListenNotificationHandler)))).Queries("events", "{events:.*}")
 
 	// ListBuckets
 	apiRouter.Methods(http.MethodGet).Path(SlashSeparator).HandlerFunc(

cmd/auth-handler.go

@@ -217,8 +217,8 @@ func getClaimsFromTokenWithSecret(token, secret string) (map[string]interface{},
 		}
 	}
 
-	// If OPA is set, return without any further checks.
-	if globalPolicyOPA != nil {
+	// If AuthZPlugin is set, return without any further checks.
+	if newGlobalAuthZPluginFn() != nil {
 		return claims.Map(), nil
 	}
 
@@ -235,7 +235,7 @@ func getClaimsFromTokenWithSecret(token, secret string) (map[string]interface{},
 			logger.LogIf(GlobalContext, err, logger.Application)
 			return nil, errAuthentication
 		}
-		claims.MapClaims[iampolicy.SessionPolicyName] = string(spBytes)
+		claims.MapClaims[sessionPolicyNameExtracted] = string(spBytes)
 	}
 
 	return claims.Map(), nil

cmd/background-newdisks-heal-ops.go

@@ -312,9 +312,6 @@ func monitorLocalDisksAndHeal(ctx context.Context, z *erasureServerPools, bgSeq
 		case <-ctx.Done():
 			return
 		case <-diskCheckTimer.C:
-			// Reset to next interval.
-			diskCheckTimer.Reset(defaultMonitorNewDiskInterval)
-
 			var erasureSetInPoolDisksToHeal []map[int][]StorageAPI
 
 			healDisks := globalBackgroundHealState.getHealLocalDiskEndpoints()
@@ -448,6 +445,9 @@ func monitorLocalDisksAndHeal(ctx context.Context, z *erasureServerPools, bgSeq
 				}
 			}
 			wg.Wait()
+
+			// Reset for next interval.
+			diskCheckTimer.Reset(defaultMonitorNewDiskInterval)
 		}
 	}
 }

cmd/bitrot.go

@@ -19,7 +19,6 @@ package cmd
 
 import (
 	"bytes"
-	"crypto/sha256"
 	"encoding/hex"
 	"errors"
 	"fmt"
@@ -27,6 +26,7 @@ import (
 	"io"
 
 	"github.com/minio/highwayhash"
+	"github.com/minio/minio/internal/hash/sha256"
 	"golang.org/x/crypto/blake2b"
 
 	xioutil "github.com/minio/minio/internal/ioutil"

cmd/bucket-handlers.go

@@ -21,7 +21,6 @@ import (
 	"bytes"
 	"context"
 	"encoding/base64"
-	"encoding/json"
 	"encoding/xml"
 	"fmt"
 	"io"
@@ -33,7 +32,6 @@ import (
 	"strconv"
 	"strings"
 	"sync"
-	"time"
 
 	"github.com/google/uuid"
 	"github.com/gorilla/mux"
@@ -364,6 +362,18 @@ func (api objectAPIHandlers) ListBucketsHandler(w http.ResponseWriter, r *http.R
 			}) {
 				bucketsInfo[n] = bucketInfo
 				n++
+			} else if globalIAMSys.IsAllowed(iampolicy.Args{
+				AccountName:     cred.AccessKey,
+				Groups:          cred.Groups,
+				Action:          iampolicy.GetBucketLocationAction,
+				BucketName:      bucketInfo.Name,
+				ConditionValues: getConditionValues(r, "", cred.AccessKey, cred.Claims),
+				IsOwner:         owner,
+				ObjectName:      "",
+				Claims:          cred.Claims,
+			}) {
+				bucketsInfo[n] = bucketInfo
+				n++
 			}
 		}
 		bucketsInfo = bucketsInfo[:n]
@@ -471,9 +481,6 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 		hasLockEnabled = true
 	}
 
-	versioned := globalBucketVersioningSys.Enabled(bucket)
-	suspended := globalBucketVersioningSys.Suspended(bucket)
-
 	type deleteResult struct {
 		delInfo DeletedObject
 		errInfo DeleteError
@@ -481,8 +488,8 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 
 	deleteResults := make([]deleteResult, len(deleteObjectsReq.Objects))
 
+	vc, _ := globalBucketVersioningSys.Get(bucket)
 	oss := make([]*objSweeper, len(deleteObjectsReq.Objects))
-
 	for index, object := range deleteObjectsReq.Objects {
 		if apiErrCode := checkRequestAuthType(ctx, r, policy.DeleteObjectAction, bucket, object.ObjectName); apiErrCode != ErrNone {
 			if apiErrCode == ErrSignatureDoesNotMatch || apiErrCode == ErrInvalidAccessKeyID {
@@ -514,11 +521,11 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 
 		opts := ObjectOptions{
 			VersionID:        object.VersionID,
-			Versioned:        versioned,
-			VersionSuspended: suspended,
+			Versioned:        vc.PrefixEnabled(object.ObjectName),
+			VersionSuspended: vc.Suspended(),
 		}
 
-		if replicateDeletes || object.VersionID != "" && hasLockEnabled || !globalTierConfigMgr.Empty() {
+		if replicateDeletes || hasLockEnabled || !globalTierConfigMgr.Empty() {
 			if !globalTierConfigMgr.Empty() && object.VersionID == "" && opts.VersionSuspended {
 				opts.VersionID = nullVersionID
 			}
@@ -526,7 +533,7 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 		}
 
 		if !globalTierConfigMgr.Empty() {
-			oss[index] = newObjSweeper(bucket, object.ObjectName).WithVersion(opts.VersionID).WithVersioning(versioned, suspended)
+			oss[index] = newObjSweeper(bucket, object.ObjectName).WithVersion(opts.VersionID).WithVersioning(opts.Versioned, opts.VersionSuspended)
 			oss[index].SetTransitionState(goi.TransitionedObject)
 		}
 
@@ -547,7 +554,7 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 				object.ReplicateDecisionStr = dsc.String()
 			}
 		}
-		if object.VersionID != "" && hasLockEnabled {
+		if hasLockEnabled {
 			if apiErrCode := enforceRetentionBypassForDelete(ctx, r, bucket, object, goi, gerr); apiErrCode != ErrNone {
 				apiErr := errorCodes.ToAPIErr(apiErrCode)
 				deleteResults[index].errInfo = DeleteError{
@@ -581,8 +588,8 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 
 	deleteList := toNames(objectsToDelete)
 	dObjects, errs := deleteObjectsFn(ctx, bucket, deleteList, ObjectOptions{
-		Versioned:        versioned,
-		VersionSuspended: suspended,
+		PrefixEnabledFn:  vc.PrefixEnabled,
+		VersionSuspended: vc.Suspended(),
 	})
 
 	for i := range errs {
@@ -638,22 +645,12 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 			continue
 		}
 
-		if replicateDeletes {
-			if dobj.DeleteMarkerReplicationStatus() == replication.Pending || dobj.VersionPurgeStatus() == Pending {
-				dv := DeletedObjectReplicationInfo{
-					DeletedObject: dobj,
-					Bucket:        bucket,
-				}
-				scheduleReplicationDelete(ctx, dv, objectAPI)
+		if replicateDeletes && (dobj.DeleteMarkerReplicationStatus() == replication.Pending || dobj.VersionPurgeStatus() == Pending) {
+			dv := DeletedObjectReplicationInfo{
+				DeletedObject: dobj,
+				Bucket:        bucket,
 			}
-		}
-
-	}
-
-	// Notify deleted event for objects.
-	for _, dobj := range deletedObjects {
-		if dobj.ObjectName == "" {
-			continue
+			scheduleReplicationDelete(ctx, dv, objectAPI)
 		}
 
 		eventName := event.ObjectRemovedDelete
@@ -1365,7 +1362,7 @@ func (api objectAPIHandlers) PutBucketObjectLockConfigHandler(w http.ResponseWri
 		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
 		return
 	}
-	if !globalIsErasure {
+	if globalIsGateway {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
 		return
 	}
@@ -1597,375 +1594,3 @@ func (api objectAPIHandlers) DeleteBucketTaggingHandler(w http.ResponseWriter, r
 	// Write success response.
 	writeSuccessResponseHeadersOnly(w)
 }
-
-// PutBucketReplicationConfigHandler - PUT Bucket replication configuration.
-// ----------
-// Add a replication configuration on the specified bucket as specified in https://docs.aws.amazon.com/AmazonS3/latest/API/API_PutBucketReplication.html
-func (api objectAPIHandlers) PutBucketReplicationConfigHandler(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "PutBucketReplicationConfig")
-	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
-
-	vars := mux.Vars(r)
-	bucket := vars["bucket"]
-	objectAPI := api.ObjectAPI()
-	if objectAPI == nil {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
-		return
-	}
-	if !globalIsErasure {
-		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
-		return
-	}
-	if s3Error := checkRequestAuthType(ctx, r, policy.PutReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
-		return
-	}
-	// Check if bucket exists.
-	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	if globalSiteReplicationSys.isEnabled() {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationDenyEditError), r.URL)
-		return
-	}
-	if versioned := globalBucketVersioningSys.Enabled(bucket); !versioned {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationNeedsVersioningError), r.URL)
-		return
-	}
-	replicationConfig, err := replication.ParseConfig(io.LimitReader(r.Body, r.ContentLength))
-	if err != nil {
-		apiErr := errorCodes.ToAPIErr(ErrMalformedXML)
-		apiErr.Description = err.Error()
-		writeErrorResponse(ctx, w, apiErr, r.URL)
-		return
-	}
-	sameTarget, apiErr := validateReplicationDestination(ctx, bucket, replicationConfig)
-	if apiErr != noError {
-		writeErrorResponse(ctx, w, apiErr, r.URL)
-		return
-	}
-	// Validate the received bucket replication config
-	if err = replicationConfig.Validate(bucket, sameTarget); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	configData, err := xml.Marshal(replicationConfig)
-	if err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	if err = globalBucketMetadataSys.Update(ctx, bucket, bucketReplicationConfig, configData); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	// Write success response.
-	writeSuccessResponseHeadersOnly(w)
-}
-
-// GetBucketReplicationConfigHandler - GET Bucket replication configuration.
-// ----------
-// Gets the replication configuration for a bucket.
-func (api objectAPIHandlers) GetBucketReplicationConfigHandler(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "GetBucketReplicationConfig")
-
-	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
-
-	vars := mux.Vars(r)
-	bucket := vars["bucket"]
-
-	objectAPI := api.ObjectAPI()
-	if objectAPI == nil {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
-		return
-	}
-
-	// check if user has permissions to perform this operation
-	if s3Error := checkRequestAuthType(ctx, r, policy.GetReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
-		return
-	}
-	// Check if bucket exists.
-	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	config, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket)
-	if err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	configData, err := xml.Marshal(config)
-	if err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	// Write success response.
-	writeSuccessResponseXML(w, configData)
-}
-
-// DeleteBucketReplicationConfigHandler - DELETE Bucket replication config.
-// ----------
-func (api objectAPIHandlers) DeleteBucketReplicationConfigHandler(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "DeleteBucketReplicationConfig")
-	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
-	vars := mux.Vars(r)
-	bucket := vars["bucket"]
-
-	objectAPI := api.ObjectAPI()
-	if objectAPI == nil {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
-		return
-	}
-
-	if s3Error := checkRequestAuthType(ctx, r, policy.PutReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
-		return
-	}
-	// Check if bucket exists.
-	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	if globalSiteReplicationSys.isEnabled() {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationDenyEditError), r.URL)
-		return
-	}
-	if err := globalBucketMetadataSys.Update(ctx, bucket, bucketReplicationConfig, nil); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	// Write success response.
-	writeSuccessResponseHeadersOnly(w)
-}
-
-// GetBucketReplicationMetricsHandler - GET Bucket replication metrics.
-// ----------
-// Gets the replication metrics for a bucket.
-func (api objectAPIHandlers) GetBucketReplicationMetricsHandler(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "GetBucketReplicationMetrics")
-
-	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
-
-	vars := mux.Vars(r)
-	bucket := vars["bucket"]
-
-	objectAPI := api.ObjectAPI()
-	if objectAPI == nil {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
-		return
-	}
-
-	// check if user has permissions to perform this operation
-	if s3Error := checkRequestAuthType(ctx, r, policy.GetReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
-		return
-	}
-
-	// Check if bucket exists.
-	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	var usageInfo BucketUsageInfo
-	dataUsageInfo, err := loadDataUsageFromBackend(ctx, objectAPI)
-	if err == nil && !dataUsageInfo.LastUpdate.IsZero() {
-		usageInfo = dataUsageInfo.BucketsUsage[bucket]
-	}
-
-	w.Header().Set(xhttp.ContentType, string(mimeJSON))
-
-	enc := json.NewEncoder(w)
-	if err = enc.Encode(getLatestReplicationStats(bucket, usageInfo)); err != nil {
-		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-		return
-	}
-}
-
-// ResetBucketReplicationStartHandler - starts a replication reset for all objects in a bucket which
-// qualify for replication and re-sync the object(s) to target, provided ExistingObjectReplication is
-// enabled for the qualifying rule. This API is a MinIO only extension provided for situations where
-// remote target is entirely lost,and previously replicated objects need to be re-synced. If resync is
-// already in progress it returns an error
-func (api objectAPIHandlers) ResetBucketReplicationStartHandler(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "ResetBucketReplicationStart")
-	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
-
-	vars := mux.Vars(r)
-	bucket := vars["bucket"]
-	durationStr := r.URL.Query().Get("older-than")
-	arn := r.URL.Query().Get("arn")
-	resetID := r.URL.Query().Get("reset-id")
-	if resetID == "" {
-		resetID = mustGetUUID()
-	}
-	var (
-		days time.Duration
-		err  error
-	)
-	if durationStr != "" {
-		days, err = time.ParseDuration(durationStr)
-		if err != nil {
-			writeErrorResponse(ctx, w, toAPIError(ctx, InvalidArgument{
-				Bucket: bucket,
-				Err:    fmt.Errorf("invalid query parameter older-than %s for %s : %w", durationStr, bucket, err),
-			}), r.URL)
-		}
-	}
-	resetBeforeDate := UTCNow().AddDate(0, 0, -1*int(days/24))
-
-	objectAPI := api.ObjectAPI()
-	if objectAPI == nil {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
-		return
-	}
-
-	if s3Error := checkRequestAuthType(ctx, r, policy.ResetBucketReplicationStateAction, bucket, ""); s3Error != ErrNone {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
-		return
-	}
-
-	// Check if bucket exists.
-	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	config, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket)
-	if err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-	if !config.HasActiveRules("", true) {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationNoMatchingRuleError), r.URL)
-		return
-	}
-	tgtArns := config.FilterTargetArns(
-		replication.ObjectOpts{
-			OpType:    replication.ResyncReplicationType,
-			TargetArn: arn,
-		})
-
-	if len(tgtArns) == 0 {
-		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
-			Bucket: bucket,
-			Err:    fmt.Errorf("Remote target ARN %s missing or ineligible for replication resync", arn),
-		}), r.URL)
-		return
-	}
-
-	if len(tgtArns) > 1 && arn == "" {
-		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
-			Bucket: bucket,
-			Err:    fmt.Errorf("ARN should be specified for replication reset"),
-		}), r.URL)
-		return
-	}
-	var rinfo ResyncTargetsInfo
-	target := globalBucketTargetSys.GetRemoteBucketTargetByArn(ctx, bucket, tgtArns[0])
-	target.ResetBeforeDate = UTCNow().AddDate(0, 0, -1*int(days/24))
-	target.ResetID = resetID
-	rinfo.Targets = append(rinfo.Targets, ResyncTarget{Arn: tgtArns[0], ResetID: target.ResetID})
-	if err = globalBucketTargetSys.SetTarget(ctx, bucket, &target, true); err != nil {
-		switch err.(type) {
-		case BucketRemoteConnectionErr:
-			writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrReplicationRemoteConnectionError, err), r.URL)
-		default:
-			writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
-		}
-	}
-	if err := startReplicationResync(ctx, bucket, arn, resetID, resetBeforeDate, objectAPI); err != nil {
-		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
-			Bucket: bucket,
-			Err:    err,
-		}), r.URL)
-		return
-	}
-
-	data, err := json.Marshal(rinfo)
-	if err != nil {
-		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-		return
-	}
-	// Write success response.
-	writeSuccessResponseJSON(w, data)
-}
-
-// ResetBucketReplicationStatusHandler - returns the status of replication reset.
-// This API is a MinIO only extension
-func (api objectAPIHandlers) ResetBucketReplicationStatusHandler(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "ResetBucketReplicationStatus")
-	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
-
-	vars := mux.Vars(r)
-	bucket := vars["bucket"]
-	arn := r.URL.Query().Get("arn")
-	var err error
-
-	objectAPI := api.ObjectAPI()
-	if objectAPI == nil {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
-		return
-	}
-
-	if s3Error := checkRequestAuthType(ctx, r, policy.ResetBucketReplicationStateAction, bucket, ""); s3Error != ErrNone {
-		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
-		return
-	}
-
-	// Check if bucket exists.
-	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	if _, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket); err != nil {
-		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
-		return
-	}
-
-	globalReplicationPool.resyncState.RLock()
-	brs, ok := globalReplicationPool.resyncState.statusMap[bucket]
-	if !ok {
-		brs, err = loadBucketResyncMetadata(ctx, bucket, objectAPI)
-		if err != nil {
-			writeErrorResponse(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
-				Bucket: bucket,
-				Err:    fmt.Errorf("No replication resync status available for %s", arn),
-			}), r.URL)
-		}
-		return
-	}
-
-	var rinfo ResyncTargetsInfo
-	for tarn, st := range brs.TargetsMap {
-		if arn != "" && tarn != arn {
-			continue
-		}
-		rinfo.Targets = append(rinfo.Targets, ResyncTarget{
-			Arn:             tarn,
-			ResetID:         st.ResyncID,
-			StartTime:       st.StartTime,
-			EndTime:         st.EndTime,
-			ResyncStatus:    st.ResyncStatus.String(),
-			ReplicatedSize:  st.ReplicatedSize,
-			ReplicatedCount: st.ReplicatedCount,
-			FailedSize:      st.FailedSize,
-			FailedCount:     st.FailedCount,
-			Bucket:          st.Bucket,
-			Object:          st.Object,
-		})
-	}
-	globalReplicationPool.resyncState.RUnlock()
-	data, err := json.Marshal(rinfo)
-	if err != nil {
-		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-		return
-	}
-	// Write success response.
-	writeSuccessResponseJSON(w, data)
-}

cmd/bucket-lifecycle.go

@@ -330,7 +330,7 @@ const (
 // 2. when a transitioned object expires (based on an ILM rule).
 func expireTransitionedObject(ctx context.Context, objectAPI ObjectLayer, oi *ObjectInfo, lcOpts lifecycle.ObjectOpts, action expireAction) error {
 	var opts ObjectOptions
-	opts.Versioned = globalBucketVersioningSys.Enabled(oi.Bucket)
+	opts.Versioned = globalBucketVersioningSys.PrefixEnabled(oi.Bucket, oi.Name)
 	opts.VersionID = lcOpts.VersionID
 	opts.Expiration = ExpirationOptions{Expire: true}
 	switch action {
@@ -415,8 +415,8 @@ func transitionObject(ctx context.Context, objectAPI ObjectLayer, oi ObjectInfo)
 			ETag:   oi.ETag,
 		},
 		VersionID:        oi.VersionID,
-		Versioned:        globalBucketVersioningSys.Enabled(oi.Bucket),
-		VersionSuspended: globalBucketVersioningSys.Suspended(oi.Bucket),
+		Versioned:        globalBucketVersioningSys.PrefixEnabled(oi.Bucket, oi.Name),
+		VersionSuspended: globalBucketVersioningSys.PrefixSuspended(oi.Bucket, oi.Name),
 		MTime:            oi.ModTime,
 	}
 	return tier, objectAPI.TransitionObject(ctx, oi.Bucket, oi.Name, opts)
@@ -575,8 +575,8 @@ func (r *RestoreObjectRequest) validate(ctx context.Context, objAPI ObjectLayer)
 
 // postRestoreOpts returns ObjectOptions with version-id from the POST restore object request for a given bucket and object.
 func postRestoreOpts(ctx context.Context, r *http.Request, bucket, object string) (opts ObjectOptions, err error) {
-	versioned := globalBucketVersioningSys.Enabled(bucket)
-	versionSuspended := globalBucketVersioningSys.Suspended(bucket)
+	versioned := globalBucketVersioningSys.PrefixEnabled(bucket, object)
+	versionSuspended := globalBucketVersioningSys.PrefixSuspended(bucket, object)
 	vid := strings.TrimSpace(r.Form.Get(xhttp.VersionID))
 	if vid != "" && vid != nullVersionID {
 		_, err := uuid.Parse(vid)
@@ -627,8 +627,8 @@ func putRestoreOpts(bucket, object string, rreq *RestoreObjectRequest, objInfo O
 			meta[xhttp.AmzServerSideEncryption] = xhttp.AmzEncryptionAES
 		}
 		return ObjectOptions{
-			Versioned:        globalBucketVersioningSys.Enabled(bucket),
-			VersionSuspended: globalBucketVersioningSys.Suspended(bucket),
+			Versioned:        globalBucketVersioningSys.PrefixEnabled(bucket, object),
+			VersionSuspended: globalBucketVersioningSys.PrefixSuspended(bucket, object),
 			UserDefined:      meta,
 		}
 	}
@@ -640,8 +640,8 @@ func putRestoreOpts(bucket, object string, rreq *RestoreObjectRequest, objInfo O
 	}
 
 	return ObjectOptions{
-		Versioned:        globalBucketVersioningSys.Enabled(bucket),
-		VersionSuspended: globalBucketVersioningSys.Suspended(bucket),
+		Versioned:        globalBucketVersioningSys.PrefixEnabled(bucket, object),
+		VersionSuspended: globalBucketVersioningSys.PrefixSuspended(bucket, object),
 		UserDefined:      meta,
 		VersionID:        objInfo.VersionID,
 		MTime:            objInfo.ModTime,

cmd/bucket-lifecycle_test.go

@@ -203,7 +203,7 @@ func TestObjectIsRemote(t *testing.T) {
 		if got := fi.IsRemote(); got != tc.remote {
 			t.Fatalf("Test %d.a: expected %v got %v", i+1, tc.remote, got)
 		}
-		oi := fi.ToObjectInfo("bucket", "object")
+		oi := fi.ToObjectInfo("bucket", "object", false)
 		if got := oi.IsRemote(); got != tc.remote {
 			t.Fatalf("Test %d.b: expected %v got %v", i+1, tc.remote, got)
 		}

cmd/bucket-metadata-sys.go

@@ -127,20 +127,12 @@ func (sys *BucketMetadataSys) Update(ctx context.Context, bucket string, configF
 		meta.QuotaConfigJSON = configData
 		meta.QuotaConfigUpdatedAt = UTCNow()
 	case objectLockConfig:
-		if !globalIsErasure && !globalIsDistErasure {
-			return NotImplemented{}
-		}
 		meta.ObjectLockConfigXML = configData
 		meta.ObjectLockConfigUpdatedAt = UTCNow()
 	case bucketVersioningConfig:
-		if !globalIsErasure && !globalIsDistErasure {
-			return NotImplemented{}
-		}
 		meta.VersioningConfigXML = configData
+		meta.VersioningConfigUpdatedAt = UTCNow()
 	case bucketReplicationConfig:
-		if !globalIsErasure && !globalIsDistErasure {
-			return NotImplemented{}
-		}
 		meta.ReplicationConfigXML = configData
 		meta.ReplicationConfigUpdatedAt = UTCNow()
 	case bucketTargetsFile:
@@ -193,12 +185,15 @@ func (sys *BucketMetadataSys) Get(bucket string) (BucketMetadata, error) {
 
 // GetVersioningConfig returns configured versioning config
 // The returned object may not be modified.
-func (sys *BucketMetadataSys) GetVersioningConfig(bucket string) (*versioning.Versioning, error) {
+func (sys *BucketMetadataSys) GetVersioningConfig(bucket string) (*versioning.Versioning, time.Time, error) {
 	meta, err := sys.GetConfig(GlobalContext, bucket)
 	if err != nil {
-		return nil, err
+		if errors.Is(err, errConfigNotFound) {
+			return &versioning.Versioning{XMLNS: "http://s3.amazonaws.com/doc/2006-03-01/"}, meta.Created, nil
+		}
+		return &versioning.Versioning{XMLNS: "http://s3.amazonaws.com/doc/2006-03-01/"}, time.Time{}, err
 	}
-	return meta.versioningConfig, nil
+	return meta.versioningConfig, meta.VersioningConfigUpdatedAt, nil
 }
 
 // GetTaggingConfig returns configured tagging config
@@ -315,26 +310,27 @@ func (sys *BucketMetadataSys) CreatedAt(bucket string) (time.Time, error) {
 
 // GetPolicyConfig returns configured bucket policy
 // The returned object may not be modified.
-func (sys *BucketMetadataSys) GetPolicyConfig(bucket string) (*policy.Policy, error) {
+func (sys *BucketMetadataSys) GetPolicyConfig(bucket string) (*policy.Policy, time.Time, error) {
 	if globalIsGateway {
 		objAPI := newObjectLayerFn()
 		if objAPI == nil {
-			return nil, errServerNotInitialized
+			return nil, time.Time{}, errServerNotInitialized
 		}
-		return objAPI.GetBucketPolicy(GlobalContext, bucket)
+		p, err := objAPI.GetBucketPolicy(GlobalContext, bucket)
+		return p, UTCNow(), err
 	}
 
 	meta, err := sys.GetConfig(GlobalContext, bucket)
 	if err != nil {
 		if errors.Is(err, errConfigNotFound) {
-			return nil, BucketPolicyNotFound{Bucket: bucket}
+			return nil, time.Time{}, BucketPolicyNotFound{Bucket: bucket}
 		}
-		return nil, err
+		return nil, time.Time{}, err
 	}
 	if meta.policyConfig == nil {
-		return nil, BucketPolicyNotFound{Bucket: bucket}
+		return nil, time.Time{}, BucketPolicyNotFound{Bucket: bucket}
 	}
-	return meta.policyConfig, nil
+	return meta.policyConfig, meta.PolicyConfigUpdatedAt, nil
 }
 
 // GetQuotaConfig returns configured bucket quota
@@ -369,6 +365,9 @@ func (sys *BucketMetadataSys) GetReplicationConfig(ctx context.Context, bucket s
 func (sys *BucketMetadataSys) GetBucketTargetsConfig(bucket string) (*madmin.BucketTargets, error) {
 	meta, err := sys.GetConfig(GlobalContext, bucket)
 	if err != nil {
+		if errors.Is(err, errConfigNotFound) {
+			return nil, BucketRemoteTargetNotFound{Bucket: bucket}
+		}
 		return nil, err
 	}
 	if meta.bucketTargetConfig == nil {
@@ -377,20 +376,6 @@ func (sys *BucketMetadataSys) GetBucketTargetsConfig(bucket string) (*madmin.Buc
 	return meta.bucketTargetConfig, nil
 }
 
-// GetBucketTarget returns the target for the bucket and arn.
-func (sys *BucketMetadataSys) GetBucketTarget(bucket string, arn string) (madmin.BucketTarget, error) {
-	targets, err := sys.GetBucketTargetsConfig(bucket)
-	if err != nil {
-		return madmin.BucketTarget{}, err
-	}
-	for _, t := range targets.Targets {
-		if t.Arn == arn {
-			return t, nil
-		}
-	}
-	return madmin.BucketTarget{}, errConfigNotFound
-}
-
 // GetConfig returns a specific configuration from the bucket metadata.
 // The returned object may not be modified.
 func (sys *BucketMetadataSys) GetConfig(ctx context.Context, bucket string) (BucketMetadata, error) {

cmd/bucket-metadata.go

@@ -87,6 +87,7 @@ type BucketMetadata struct {
 	TaggingConfigUpdatedAt      time.Time
 	QuotaConfigUpdatedAt        time.Time
 	ReplicationConfigUpdatedAt  time.Time
+	VersioningConfigUpdatedAt   time.Time
 
 	// Unexported fields. Must be updated atomically.
 	policyConfig           *policy.Policy
@@ -364,6 +365,7 @@ func (b *BucketMetadata) defaultTimestamps() {
 	if b.PolicyConfigUpdatedAt.IsZero() {
 		b.PolicyConfigUpdatedAt = b.Created
 	}
+
 	if b.EncryptionConfigUpdatedAt.IsZero() {
 		b.EncryptionConfigUpdatedAt = b.Created
 	}
@@ -383,6 +385,10 @@ func (b *BucketMetadata) defaultTimestamps() {
 	if b.ReplicationConfigUpdatedAt.IsZero() {
 		b.ReplicationConfigUpdatedAt = b.Created
 	}
+
+	if b.VersioningConfigUpdatedAt.IsZero() {
+		b.VersioningConfigUpdatedAt = b.Created
+	}
 }
 
 // Save config to supplied ObjectLayer api.

cmd/bucket-metadata_gen.go

@@ -144,6 +144,12 @@ func (z *BucketMetadata) DecodeMsg(dc *msgp.Reader) (err error) {
 				err = msgp.WrapError(err, "ReplicationConfigUpdatedAt")
 				return
 			}
+		case "VersioningConfigUpdatedAt":
+			z.VersioningConfigUpdatedAt, err = dc.ReadTime()
+			if err != nil {
+				err = msgp.WrapError(err, "VersioningConfigUpdatedAt")
+				return
+			}
 		default:
 			err = dc.Skip()
 			if err != nil {
@@ -157,9 +163,9 @@ func (z *BucketMetadata) DecodeMsg(dc *msgp.Reader) (err error) {
 
 // EncodeMsg implements msgp.Encodable
 func (z *BucketMetadata) EncodeMsg(en *msgp.Writer) (err error) {
-	// map header, size 20
+	// map header, size 21
 	// write "Name"
-	err = en.Append(0xde, 0x0, 0x14, 0xa4, 0x4e, 0x61, 0x6d, 0x65)
+	err = en.Append(0xde, 0x0, 0x15, 0xa4, 0x4e, 0x61, 0x6d, 0x65)
 	if err != nil {
 		return
 	}
@@ -358,15 +364,25 @@ func (z *BucketMetadata) EncodeMsg(en *msgp.Writer) (err error) {
 		err = msgp.WrapError(err, "ReplicationConfigUpdatedAt")
 		return
 	}
+	// write "VersioningConfigUpdatedAt"
+	err = en.Append(0xb9, 0x56, 0x65, 0x72, 0x73, 0x69, 0x6f, 0x6e, 0x69, 0x6e, 0x67, 0x43, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x55, 0x70, 0x64, 0x61, 0x74, 0x65, 0x64, 0x41, 0x74)
+	if err != nil {
+		return
+	}
+	err = en.WriteTime(z.VersioningConfigUpdatedAt)
+	if err != nil {
+		err = msgp.WrapError(err, "VersioningConfigUpdatedAt")
+		return
+	}
 	return
 }
 
 // MarshalMsg implements msgp.Marshaler
 func (z *BucketMetadata) MarshalMsg(b []byte) (o []byte, err error) {
 	o = msgp.Require(b, z.Msgsize())
-	// map header, size 20
+	// map header, size 21
 	// string "Name"
-	o = append(o, 0xde, 0x0, 0x14, 0xa4, 0x4e, 0x61, 0x6d, 0x65)
+	o = append(o, 0xde, 0x0, 0x15, 0xa4, 0x4e, 0x61, 0x6d, 0x65)
 	o = msgp.AppendString(o, z.Name)
 	// string "Created"
 	o = append(o, 0xa7, 0x43, 0x72, 0x65, 0x61, 0x74, 0x65, 0x64)
@@ -425,6 +441,9 @@ func (z *BucketMetadata) MarshalMsg(b []byte) (o []byte, err error) {
 	// string "ReplicationConfigUpdatedAt"
 	o = append(o, 0xba, 0x52, 0x65, 0x70, 0x6c, 0x69, 0x63, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x43, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x55, 0x70, 0x64, 0x61, 0x74, 0x65, 0x64, 0x41, 0x74)
 	o = msgp.AppendTime(o, z.ReplicationConfigUpdatedAt)
+	// string "VersioningConfigUpdatedAt"
+	o = append(o, 0xb9, 0x56, 0x65, 0x72, 0x73, 0x69, 0x6f, 0x6e, 0x69, 0x6e, 0x67, 0x43, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x55, 0x70, 0x64, 0x61, 0x74, 0x65, 0x64, 0x41, 0x74)
+	o = msgp.AppendTime(o, z.VersioningConfigUpdatedAt)
 	return
 }
 
@@ -566,6 +585,12 @@ func (z *BucketMetadata) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "ReplicationConfigUpdatedAt")
 				return
 			}
+		case "VersioningConfigUpdatedAt":
+			z.VersioningConfigUpdatedAt, bts, err = msgp.ReadTimeBytes(bts)
+			if err != nil {
+				err = msgp.WrapError(err, "VersioningConfigUpdatedAt")
+				return
+			}
 		default:
 			bts, err = msgp.Skip(bts)
 			if err != nil {
@@ -580,6 +605,6 @@ func (z *BucketMetadata) 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 *BucketMetadata) Msgsize() (s int) {
-	s = 3 + 5 + msgp.StringPrefixSize + len(z.Name) + 8 + msgp.TimeSize + 12 + msgp.BoolSize + 17 + msgp.BytesPrefixSize + len(z.PolicyConfigJSON) + 22 + msgp.BytesPrefixSize + len(z.NotificationConfigXML) + 19 + msgp.BytesPrefixSize + len(z.LifecycleConfigXML) + 20 + msgp.BytesPrefixSize + len(z.ObjectLockConfigXML) + 20 + msgp.BytesPrefixSize + len(z.VersioningConfigXML) + 20 + msgp.BytesPrefixSize + len(z.EncryptionConfigXML) + 17 + msgp.BytesPrefixSize + len(z.TaggingConfigXML) + 16 + msgp.BytesPrefixSize + len(z.QuotaConfigJSON) + 21 + msgp.BytesPrefixSize + len(z.ReplicationConfigXML) + 24 + msgp.BytesPrefixSize + len(z.BucketTargetsConfigJSON) + 28 + msgp.BytesPrefixSize + len(z.BucketTargetsConfigMetaJSON) + 22 + msgp.TimeSize + 26 + msgp.TimeSize + 26 + msgp.TimeSize + 23 + msgp.TimeSize + 21 + msgp.TimeSize + 27 + msgp.TimeSize
+	s = 3 + 5 + msgp.StringPrefixSize + len(z.Name) + 8 + msgp.TimeSize + 12 + msgp.BoolSize + 17 + msgp.BytesPrefixSize + len(z.PolicyConfigJSON) + 22 + msgp.BytesPrefixSize + len(z.NotificationConfigXML) + 19 + msgp.BytesPrefixSize + len(z.LifecycleConfigXML) + 20 + msgp.BytesPrefixSize + len(z.ObjectLockConfigXML) + 20 + msgp.BytesPrefixSize + len(z.VersioningConfigXML) + 20 + msgp.BytesPrefixSize + len(z.EncryptionConfigXML) + 17 + msgp.BytesPrefixSize + len(z.TaggingConfigXML) + 16 + msgp.BytesPrefixSize + len(z.QuotaConfigJSON) + 21 + msgp.BytesPrefixSize + len(z.ReplicationConfigXML) + 24 + msgp.BytesPrefixSize + len(z.BucketTargetsConfigJSON) + 28 + msgp.BytesPrefixSize + len(z.BucketTargetsConfigMetaJSON) + 22 + msgp.TimeSize + 26 + msgp.TimeSize + 26 + msgp.TimeSize + 23 + msgp.TimeSize + 21 + msgp.TimeSize + 27 + msgp.TimeSize + 26 + msgp.TimeSize
 	return
 }

cmd/bucket-policy.go

@@ -38,7 +38,8 @@ type PolicySys struct{}
 
 // Get returns stored bucket policy
 func (sys *PolicySys) Get(bucket string) (*policy.Policy, error) {
-	return globalBucketMetadataSys.GetPolicyConfig(bucket)
+	policy, _, err := globalBucketMetadataSys.GetPolicyConfig(bucket)
+	return policy, err
 }
 
 // IsAllowed - checks given policy args is allowed to continue the Rest API.

cmd/bucket-replication-handlers.go

@@ -0,0 +1,404 @@
+// Copyright (c) 2015-2022 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 (
+	"encoding/json"
+	"encoding/xml"
+	"fmt"
+	"io"
+	"net/http"
+	"time"
+
+	"github.com/gorilla/mux"
+	"github.com/minio/minio/internal/bucket/replication"
+	xhttp "github.com/minio/minio/internal/http"
+	"github.com/minio/minio/internal/logger"
+	"github.com/minio/pkg/bucket/policy"
+)
+
+// PutBucketReplicationConfigHandler - PUT Bucket replication configuration.
+// ----------
+// Add a replication configuration on the specified bucket as specified in https://docs.aws.amazon.com/AmazonS3/latest/API/API_PutBucketReplication.html
+func (api objectAPIHandlers) PutBucketReplicationConfigHandler(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "PutBucketReplicationConfig")
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	vars := mux.Vars(r)
+	bucket := vars["bucket"]
+	objectAPI := api.ObjectAPI()
+	if objectAPI == nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
+		return
+	}
+	if globalIsGateway {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+	if s3Error := checkRequestAuthType(ctx, r, policy.PutReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
+		return
+	}
+	// Check if bucket exists.
+	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	if versioned := globalBucketVersioningSys.Enabled(bucket); !versioned {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationNeedsVersioningError), r.URL)
+		return
+	}
+	replicationConfig, err := replication.ParseConfig(io.LimitReader(r.Body, r.ContentLength))
+	if err != nil {
+		apiErr := errorCodes.ToAPIErr(ErrMalformedXML)
+		apiErr.Description = err.Error()
+		writeErrorResponse(ctx, w, apiErr, r.URL)
+		return
+	}
+	sameTarget, apiErr := validateReplicationDestination(ctx, bucket, replicationConfig, true)
+	if apiErr != noError {
+		writeErrorResponse(ctx, w, apiErr, r.URL)
+		return
+	}
+	// Validate the received bucket replication config
+	if err = replicationConfig.Validate(bucket, sameTarget); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+	configData, err := xml.Marshal(replicationConfig)
+	if err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+	if err = globalBucketMetadataSys.Update(ctx, bucket, bucketReplicationConfig, configData); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	// Write success response.
+	writeSuccessResponseHeadersOnly(w)
+}
+
+// GetBucketReplicationConfigHandler - GET Bucket replication configuration.
+// ----------
+// Gets the replication configuration for a bucket.
+func (api objectAPIHandlers) GetBucketReplicationConfigHandler(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "GetBucketReplicationConfig")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	vars := mux.Vars(r)
+	bucket := vars["bucket"]
+
+	objectAPI := api.ObjectAPI()
+	if objectAPI == nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
+		return
+	}
+
+	// check if user has permissions to perform this operation
+	if s3Error := checkRequestAuthType(ctx, r, policy.GetReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
+		return
+	}
+	// Check if bucket exists.
+	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	config, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket)
+	if err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+	configData, err := xml.Marshal(config)
+	if err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	// Write success response.
+	writeSuccessResponseXML(w, configData)
+}
+
+// DeleteBucketReplicationConfigHandler - DELETE Bucket replication config.
+// ----------
+func (api objectAPIHandlers) DeleteBucketReplicationConfigHandler(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "DeleteBucketReplicationConfig")
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+	vars := mux.Vars(r)
+	bucket := vars["bucket"]
+
+	objectAPI := api.ObjectAPI()
+	if objectAPI == nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
+		return
+	}
+
+	if s3Error := checkRequestAuthType(ctx, r, policy.PutReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
+		return
+	}
+	// Check if bucket exists.
+	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+	if globalSiteReplicationSys.isEnabled() {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationDenyEditError), r.URL)
+		return
+	}
+	if err := globalBucketMetadataSys.Update(ctx, bucket, bucketReplicationConfig, nil); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	// Write success response.
+	writeSuccessResponseHeadersOnly(w)
+}
+
+// GetBucketReplicationMetricsHandler - GET Bucket replication metrics.
+// ----------
+// Gets the replication metrics for a bucket.
+func (api objectAPIHandlers) GetBucketReplicationMetricsHandler(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "GetBucketReplicationMetrics")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	vars := mux.Vars(r)
+	bucket := vars["bucket"]
+
+	objectAPI := api.ObjectAPI()
+	if objectAPI == nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
+		return
+	}
+
+	// check if user has permissions to perform this operation
+	if s3Error := checkRequestAuthType(ctx, r, policy.GetReplicationConfigurationAction, bucket, ""); s3Error != ErrNone {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
+		return
+	}
+
+	// Check if bucket exists.
+	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+	var usageInfo BucketUsageInfo
+	dataUsageInfo, err := loadDataUsageFromBackend(ctx, objectAPI)
+	if err == nil && !dataUsageInfo.LastUpdate.IsZero() {
+		usageInfo = dataUsageInfo.BucketsUsage[bucket]
+	}
+
+	w.Header().Set(xhttp.ContentType, string(mimeJSON))
+
+	enc := json.NewEncoder(w)
+	if err = enc.Encode(getLatestReplicationStats(bucket, usageInfo)); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+}
+
+// ResetBucketReplicationStartHandler - starts a replication reset for all objects in a bucket which
+// qualify for replication and re-sync the object(s) to target, provided ExistingObjectReplication is
+// enabled for the qualifying rule. This API is a MinIO only extension provided for situations where
+// remote target is entirely lost,and previously replicated objects need to be re-synced. If resync is
+// already in progress it returns an error
+func (api objectAPIHandlers) ResetBucketReplicationStartHandler(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "ResetBucketReplicationStart")
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	vars := mux.Vars(r)
+	bucket := vars["bucket"]
+	durationStr := r.URL.Query().Get("older-than")
+	arn := r.URL.Query().Get("arn")
+	resetID := r.URL.Query().Get("reset-id")
+	if resetID == "" {
+		resetID = mustGetUUID()
+	}
+	var (
+		days time.Duration
+		err  error
+	)
+	if durationStr != "" {
+		days, err = time.ParseDuration(durationStr)
+		if err != nil {
+			writeErrorResponse(ctx, w, toAPIError(ctx, InvalidArgument{
+				Bucket: bucket,
+				Err:    fmt.Errorf("invalid query parameter older-than %s for %s : %w", durationStr, bucket, err),
+			}), r.URL)
+		}
+	}
+	resetBeforeDate := UTCNow().AddDate(0, 0, -1*int(days/24))
+
+	objectAPI := api.ObjectAPI()
+	if objectAPI == nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
+		return
+	}
+
+	if s3Error := checkRequestAuthType(ctx, r, policy.ResetBucketReplicationStateAction, bucket, ""); s3Error != ErrNone {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
+		return
+	}
+
+	// Check if bucket exists.
+	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	config, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket)
+	if err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+	if !config.HasExistingObjectReplication(arn) {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrReplicationNoExistingObjects), r.URL)
+		return
+	}
+
+	tgtArns := config.FilterTargetArns(
+		replication.ObjectOpts{
+			OpType:    replication.ResyncReplicationType,
+			TargetArn: arn,
+		})
+
+	if len(tgtArns) == 0 {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
+			Bucket: bucket,
+			Err:    fmt.Errorf("Remote target ARN %s missing or ineligible for replication resync", arn),
+		}), r.URL)
+		return
+	}
+
+	if len(tgtArns) > 1 && arn == "" {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
+			Bucket: bucket,
+			Err:    fmt.Errorf("ARN should be specified for replication reset"),
+		}), r.URL)
+		return
+	}
+	var rinfo ResyncTargetsInfo
+	target := globalBucketTargetSys.GetRemoteBucketTargetByArn(ctx, bucket, tgtArns[0])
+	target.ResetBeforeDate = UTCNow().AddDate(0, 0, -1*int(days/24))
+	target.ResetID = resetID
+	rinfo.Targets = append(rinfo.Targets, ResyncTarget{Arn: tgtArns[0], ResetID: target.ResetID})
+	if err = globalBucketTargetSys.SetTarget(ctx, bucket, &target, true); err != nil {
+		switch err.(type) {
+		case BucketRemoteConnectionErr:
+			writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrReplicationRemoteConnectionError, err), r.URL)
+		default:
+			writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
+		}
+	}
+	if err := startReplicationResync(ctx, bucket, arn, resetID, resetBeforeDate, objectAPI); err != nil {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
+			Bucket: bucket,
+			Err:    err,
+		}), r.URL)
+		return
+	}
+
+	data, err := json.Marshal(rinfo)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+	// Write success response.
+	writeSuccessResponseJSON(w, data)
+}
+
+// ResetBucketReplicationStatusHandler - returns the status of replication reset.
+// This API is a MinIO only extension
+func (api objectAPIHandlers) ResetBucketReplicationStatusHandler(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "ResetBucketReplicationStatus")
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	vars := mux.Vars(r)
+	bucket := vars["bucket"]
+	arn := r.URL.Query().Get("arn")
+	var err error
+
+	objectAPI := api.ObjectAPI()
+	if objectAPI == nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
+		return
+	}
+
+	if s3Error := checkRequestAuthType(ctx, r, policy.ResetBucketReplicationStateAction, bucket, ""); s3Error != ErrNone {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(s3Error), r.URL)
+		return
+	}
+
+	// Check if bucket exists.
+	if _, err := objectAPI.GetBucketInfo(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	if _, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
+	globalReplicationPool.resyncState.RLock()
+	brs, ok := globalReplicationPool.resyncState.statusMap[bucket]
+	globalReplicationPool.resyncState.RUnlock()
+	if !ok {
+		brs, err = loadBucketResyncMetadata(ctx, bucket, objectAPI)
+		if err != nil {
+			writeErrorResponse(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{
+				Bucket: bucket,
+				Err:    fmt.Errorf("No replication resync status available for %s", arn),
+			}), r.URL)
+			return
+		}
+	}
+
+	var rinfo ResyncTargetsInfo
+	for tarn, st := range brs.TargetsMap {
+		if arn != "" && tarn != arn {
+			continue
+		}
+		rinfo.Targets = append(rinfo.Targets, ResyncTarget{
+			Arn:             tarn,
+			ResetID:         st.ResyncID,
+			StartTime:       st.StartTime,
+			EndTime:         st.EndTime,
+			ResyncStatus:    st.ResyncStatus.String(),
+			ReplicatedSize:  st.ReplicatedSize,
+			ReplicatedCount: st.ReplicatedCount,
+			FailedSize:      st.FailedSize,
+			FailedCount:     st.FailedCount,
+			Bucket:          st.Bucket,
+			Object:          st.Object,
+		})
+	}
+	data, err := json.Marshal(rinfo)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	// Write success response.
+	writeSuccessResponseJSON(w, data)
+}

cmd/bucket-replication-stats.go

@@ -135,6 +135,23 @@ func (r *ReplicationStats) GetInitialUsage(bucket string) BucketReplicationStats
 	return st.Clone()
 }
 
+// GetAll returns replication metrics for all buckets at once.
+func (r *ReplicationStats) GetAll() map[string]BucketReplicationStats {
+	if r == nil {
+		return map[string]BucketReplicationStats{}
+	}
+
+	r.RLock()
+	defer r.RUnlock()
+
+	bucketReplicationStats := make(map[string]BucketReplicationStats, len(r.Cache))
+	for k, v := range r.Cache {
+		bucketReplicationStats[k] = v.Clone()
+	}
+
+	return bucketReplicationStats
+}
+
 // Get replication metrics for a bucket from this node since this node came up.
 func (r *ReplicationStats) Get(bucket string) BucketReplicationStats {
 	if r == nil {
@@ -161,7 +178,7 @@ func NewReplicationStats(ctx context.Context, objectAPI ObjectLayer) *Replicatio
 
 // load replication metrics at cluster start from initial data usage
 func (r *ReplicationStats) loadInitialReplicationMetrics(ctx context.Context) {
-	rTimer := time.NewTimer(time.Minute * 1)
+	rTimer := time.NewTimer(time.Minute)
 	defer rTimer.Stop()
 	var (
 		dui DataUsageInfo
@@ -174,13 +191,12 @@ outer:
 			return
 		case <-rTimer.C:
 			dui, err = loadDataUsageFromBackend(GlobalContext, newObjectLayerFn())
-			if err != nil {
-				continue
-			}
 			// If LastUpdate is set, data usage is available.
-			if !dui.LastUpdate.IsZero() {
+			if err == nil && !dui.LastUpdate.IsZero() {
 				break outer
 			}
+
+			rTimer.Reset(time.Minute)
 		}
 	}
 

cmd/bucket-replication-utils.go

@@ -509,7 +509,7 @@ func getHealReplicateObjectInfo(objInfo ObjectInfo, rcfg replicationConfig) Repl
 				ObjectName: oi.Name,
 				VersionID:  oi.VersionID,
 			},
-		}, oi, ObjectOptions{}, nil)
+		}, oi, ObjectOptions{VersionSuspended: globalBucketVersioningSys.PrefixSuspended(oi.Bucket, oi.Name)}, nil)
 	} else {
 		dsc = mustReplicate(GlobalContext, oi.Bucket, oi.Name, getMustReplicateOptions(ObjectInfo{
 			UserDefined: oi.UserDefined,

cmd/bucket-replication.go

@@ -87,7 +87,7 @@ func getReplicationConfig(ctx context.Context, bucketName string) (rc *replicati
 
 // validateReplicationDestination returns error if replication destination bucket missing or not configured
 // It also returns true if replication destination is same as this server.
-func validateReplicationDestination(ctx context.Context, bucket string, rCfg *replication.Config) (bool, APIError) {
+func validateReplicationDestination(ctx context.Context, bucket string, rCfg *replication.Config, checkRemote bool) (bool, APIError) {
 	var arns []string
 	if rCfg.RoleArn != "" {
 		arns = append(arns, rCfg.RoleArn)
@@ -96,26 +96,29 @@ func validateReplicationDestination(ctx context.Context, bucket string, rCfg *re
 			arns = append(arns, rule.Destination.String())
 		}
 	}
+	var sameTarget bool
 	for _, arnStr := range arns {
 		arn, err := madmin.ParseARN(arnStr)
 		if err != nil {
-			return false, errorCodes.ToAPIErrWithErr(ErrBucketRemoteArnInvalid, err)
+			return sameTarget, errorCodes.ToAPIErrWithErr(ErrBucketRemoteArnInvalid, err)
 		}
 		if arn.Type != madmin.ReplicationService {
-			return false, toAPIError(ctx, BucketRemoteArnTypeInvalid{Bucket: bucket})
+			return sameTarget, toAPIError(ctx, BucketRemoteArnTypeInvalid{Bucket: bucket})
 		}
 		clnt := globalBucketTargetSys.GetRemoteTargetClient(ctx, arnStr)
 		if clnt == nil {
-			return false, toAPIError(ctx, BucketRemoteTargetNotFound{Bucket: bucket})
+			return sameTarget, toAPIError(ctx, BucketRemoteTargetNotFound{Bucket: bucket})
 		}
-		if found, err := clnt.BucketExists(ctx, arn.Bucket); !found {
-			return false, errorCodes.ToAPIErrWithErr(ErrRemoteDestinationNotFoundError, err)
-		}
-		if ret, err := globalBucketObjectLockSys.Get(bucket); err == nil {
-			if ret.LockEnabled {
-				lock, _, _, _, err := clnt.GetObjectLockConfig(ctx, arn.Bucket)
-				if err != nil || lock != "Enabled" {
-					return false, errorCodes.ToAPIErrWithErr(ErrReplicationDestinationMissingLock, err)
+		if checkRemote { // validate remote bucket
+			if found, err := clnt.BucketExists(ctx, arn.Bucket); !found {
+				return sameTarget, errorCodes.ToAPIErrWithErr(ErrRemoteDestinationNotFoundError, err)
+			}
+			if ret, err := globalBucketObjectLockSys.Get(bucket); err == nil {
+				if ret.LockEnabled {
+					lock, _, _, _, err := clnt.GetObjectLockConfig(ctx, arn.Bucket)
+					if err != nil || lock != "Enabled" {
+						return sameTarget, errorCodes.ToAPIErrWithErr(ErrReplicationDestinationMissingLock, err)
+					}
 				}
 			}
 		}
@@ -123,12 +126,19 @@ func validateReplicationDestination(ctx context.Context, bucket string, rCfg *re
 		c, ok := globalBucketTargetSys.arnRemotesMap[arnStr]
 		if ok {
 			if c.EndpointURL().String() == clnt.EndpointURL().String() {
-				sameTarget, _ := isLocalHost(clnt.EndpointURL().Hostname(), clnt.EndpointURL().Port(), globalMinioPort)
-				return sameTarget, toAPIError(ctx, nil)
+				selfTarget, _ := isLocalHost(clnt.EndpointURL().Hostname(), clnt.EndpointURL().Port(), globalMinioPort)
+				if !sameTarget {
+					sameTarget = selfTarget
+				}
+				continue
 			}
 		}
 	}
-	return false, toAPIError(ctx, BucketRemoteTargetNotFound{Bucket: bucket})
+
+	if len(arns) == 0 {
+		return false, toAPIError(ctx, BucketRemoteTargetNotFound{Bucket: bucket})
+	}
+	return sameTarget, toAPIError(ctx, nil)
 }
 
 type mustReplicateOptions struct {
@@ -180,6 +190,17 @@ func mustReplicate(ctx context.Context, bucket, object string, mopts mustReplica
 		return
 	}
 
+	// object layer not initialized we return with no decision.
+	if newObjectLayerFn() == nil {
+		return
+	}
+
+	// Disable server-side replication on object prefixes which are excluded
+	// from versioning via the MinIO bucket versioning extension.
+	if globalBucketVersioningSys.PrefixSuspended(bucket, object) {
+		return
+	}
+
 	replStatus := mopts.ReplicationStatus()
 	if replStatus == replication.Replica && !mopts.isMetadataReplication() {
 		return
@@ -264,6 +285,11 @@ func checkReplicateDelete(ctx context.Context, bucket string, dobj ObjectToDelet
 	if delOpts.ReplicationRequest {
 		return
 	}
+	// Skip replication if this object's prefix is excluded from being
+	// versioned.
+	if !delOpts.Versioned {
+		return
+	}
 	opts := replication.ObjectOpts{
 		Name:         dobj.ObjectName,
 		SSEC:         crypto.SSEC.IsEncrypted(oi.UserDefined),
@@ -457,8 +483,8 @@ func replicateDelete(ctx context.Context, dobj DeletedObjectReplicationInfo, obj
 		VersionID:         versionID,
 		MTime:             dobj.DeleteMarkerMTime.Time,
 		DeleteReplication: drs,
-		Versioned:         globalBucketVersioningSys.Enabled(bucket),
-		VersionSuspended:  globalBucketVersioningSys.Suspended(bucket),
+		Versioned:         globalBucketVersioningSys.PrefixEnabled(bucket, dobj.ObjectName),
+		VersionSuspended:  globalBucketVersioningSys.PrefixSuspended(bucket, dobj.ObjectName),
 	})
 	if err != nil && !isErrVersionNotFound(err) { // VersionNotFound would be reported by pool that object version is missing on.
 		logger.LogIf(ctx, fmt.Errorf("Unable to update replication metadata for %s/%s(%s): %s", bucket, dobj.ObjectName, versionID, err))
@@ -1019,8 +1045,13 @@ func replicateObjectToTarget(ctx context.Context, ri ReplicateObjectInfo, object
 		return
 	}
 
+	versioned := globalBucketVersioningSys.PrefixEnabled(bucket, object)
+	versionSuspended := globalBucketVersioningSys.PrefixSuspended(bucket, object)
+
 	gr, err = objectAPI.GetObjectNInfo(ctx, bucket, object, nil, http.Header{}, readLock, ObjectOptions{
-		VersionID: objInfo.VersionID,
+		VersionID:        objInfo.VersionID,
+		Versioned:        versioned,
+		VersionSuspended: versionSuspended,
 	})
 	if err != nil {
 		sendEvent(eventArgs{
@@ -1590,7 +1621,11 @@ func proxyGetToReplicationTarget(ctx context.Context, bucket, object string, rs
 	return reader, proxyResult{Proxy: true}, nil
 }
 
-func getproxyTargets(ctx context.Context, bucket, object string, opts ObjectOptions) (tgts *madmin.BucketTargets) {
+func getProxyTargets(ctx context.Context, bucket, object string, opts ObjectOptions) (tgts *madmin.BucketTargets) {
+	if opts.VersionSuspended {
+		return &madmin.BucketTargets{}
+	}
+
 	cfg, err := getReplicationConfig(ctx, bucket)
 	if err != nil || cfg == nil {
 		return &madmin.BucketTargets{}
@@ -1727,10 +1762,6 @@ func (c replicationConfig) Resync(ctx context.Context, oi ObjectInfo, dsc *Repli
 	if c.Empty() {
 		return
 	}
-	// existing object replication does not apply to un-versioned objects
-	if oi.VersionID == "" {
-		return
-	}
 
 	// Now overlay existing object replication choices for target
 	if oi.DeleteMarker {
@@ -1831,9 +1862,26 @@ func resyncTarget(oi ObjectInfo, arn string, resetID string, resetBeforeDate tim
 	return
 }
 
-// get the most current of in-memory replication stats  and data usage info from crawler.
-func getLatestReplicationStats(bucket string, u BucketUsageInfo) (s BucketReplicationStats) {
-	bucketStats := globalNotificationSys.GetClusterBucketStats(GlobalContext, bucket)
+func getAllLatestReplicationStats(bucketsUsage map[string]BucketUsageInfo) (bucketsReplicationStats map[string]BucketReplicationStats) {
+	peerBucketStatsList := globalNotificationSys.GetClusterAllBucketStats(GlobalContext)
+	bucketsReplicationStats = make(map[string]BucketReplicationStats, len(bucketsUsage))
+
+	for bucket, u := range bucketsUsage {
+		bucketStats := make([]BucketStats, len(peerBucketStatsList))
+		for i, peerBucketStats := range peerBucketStatsList {
+			bucketStat, ok := peerBucketStats[bucket]
+			if !ok {
+				continue
+			}
+			bucketStats[i] = bucketStat
+		}
+		bucketsReplicationStats[bucket] = calculateBucketReplicationStats(bucket, u, bucketStats)
+	}
+
+	return bucketsReplicationStats
+}
+
+func calculateBucketReplicationStats(bucket string, u BucketUsageInfo, bucketStats []BucketStats) (s BucketReplicationStats) {
 	// accumulate cluster bucket stats
 	stats := make(map[string]*BucketReplicationStat)
 	var totReplicaSize int64
@@ -1902,6 +1950,12 @@ func getLatestReplicationStats(bucket string, u BucketUsageInfo) (s BucketReplic
 	return s
 }
 
+// get the most current of in-memory replication stats  and data usage info from crawler.
+func getLatestReplicationStats(bucket string, u BucketUsageInfo) (s BucketReplicationStats) {
+	bucketStats := globalNotificationSys.GetClusterBucketStats(GlobalContext, bucket)
+	return calculateBucketReplicationStats(bucket, u, bucketStats)
+}
+
 const resyncTimeInterval = time.Minute * 10
 
 // periodicResyncMetaSave saves in-memory resync meta stats to disk in periodic intervals
@@ -1912,7 +1966,6 @@ func (p *ReplicationPool) periodicResyncMetaSave(ctx context.Context, objectAPI
 	for {
 		select {
 		case <-resyncTimer.C:
-			resyncTimer.Reset(resyncTimeInterval)
 			now := UTCNow()
 			p.resyncState.RLock()
 			for bucket, brs := range p.resyncState.statusMap {
@@ -1933,6 +1986,8 @@ func (p *ReplicationPool) periodicResyncMetaSave(ctx context.Context, objectAPI
 				}
 			}
 			p.resyncState.RUnlock()
+
+			resyncTimer.Reset(resyncTimeInterval)
 		case <-ctx.Done():
 			// server could be restarting - need
 			// to exit immediately
@@ -1952,6 +2007,7 @@ func resyncBucket(ctx context.Context, bucket, arn string, heal bool, objectAPI
 		st.EndTime = UTCNow()
 		st.ResyncStatus = resyncStatus
 		m.TargetsMap[arn] = st
+		globalReplicationPool.resyncState.statusMap[bucket] = m
 		globalReplicationPool.resyncState.Unlock()
 	}()
 	// Allocate new results channel to receive ObjectInfo.
@@ -2012,7 +2068,6 @@ func resyncBucket(ctx context.Context, bucket, arn string, heal bool, objectAPI
 		}
 
 		if roi.DeleteMarker || !roi.VersionPurgeStatus.Empty() {
-
 			versionID := ""
 			dmVersionID := ""
 			if roi.VersionPurgeStatus.Empty() {
@@ -2059,6 +2114,7 @@ func resyncBucket(ctx context.Context, bucket, arn string, heal bool, objectAPI
 			st.ReplicatedSize += roi.Size
 		}
 		m.TargetsMap[arn] = st
+		globalReplicationPool.resyncState.statusMap[bucket] = m
 		globalReplicationPool.resyncState.Unlock()
 	}
 	resyncStatus = ResyncCompleted
@@ -2139,10 +2195,6 @@ func (p *ReplicationPool) initResync(ctx context.Context, buckets []BucketInfo,
 	if objAPI == nil {
 		return errServerNotInitialized
 	}
-	// replication applies only to erasure coded setups
-	if !globalIsErasure {
-		return nil
-	}
 	// Load bucket metadata sys in background
 	go p.loadResync(ctx, buckets, objAPI)
 	return nil

cmd/bucket-stats.go

@@ -51,6 +51,9 @@ func (rl *ReplicationLatency) update(size int64, duration time.Duration) {
 	rl.UploadHistogram.Add(size, duration)
 }
 
+// BucketStatsMap captures bucket statistics for all buckets
+type BucketStatsMap map[string]BucketStats
+
 // BucketStats bucket statistics
 type BucketStats struct {
 	ReplicationStats BucketReplicationStats

cmd/bucket-stats_gen.go

@@ -792,6 +792,183 @@ func (z *BucketStats) Msgsize() (s int) {
 	return
 }
 
+// DecodeMsg implements msgp.Decodable
+func (z *BucketStatsMap) DecodeMsg(dc *msgp.Reader) (err error) {
+	var zb0003 uint32
+	zb0003, err = dc.ReadMapHeader()
+	if err != nil {
+		err = msgp.WrapError(err)
+		return
+	}
+	if (*z) == nil {
+		(*z) = make(BucketStatsMap, zb0003)
+	} else if len((*z)) > 0 {
+		for key := range *z {
+			delete((*z), key)
+		}
+	}
+	for zb0003 > 0 {
+		zb0003--
+		var zb0001 string
+		var zb0002 BucketStats
+		zb0001, err = dc.ReadString()
+		if err != nil {
+			err = msgp.WrapError(err)
+			return
+		}
+		var field []byte
+		_ = field
+		var zb0004 uint32
+		zb0004, err = dc.ReadMapHeader()
+		if err != nil {
+			err = msgp.WrapError(err, zb0001)
+			return
+		}
+		for zb0004 > 0 {
+			zb0004--
+			field, err = dc.ReadMapKeyPtr()
+			if err != nil {
+				err = msgp.WrapError(err, zb0001)
+				return
+			}
+			switch msgp.UnsafeString(field) {
+			case "ReplicationStats":
+				err = zb0002.ReplicationStats.DecodeMsg(dc)
+				if err != nil {
+					err = msgp.WrapError(err, zb0001, "ReplicationStats")
+					return
+				}
+			default:
+				err = dc.Skip()
+				if err != nil {
+					err = msgp.WrapError(err, zb0001)
+					return
+				}
+			}
+		}
+		(*z)[zb0001] = zb0002
+	}
+	return
+}
+
+// EncodeMsg implements msgp.Encodable
+func (z BucketStatsMap) EncodeMsg(en *msgp.Writer) (err error) {
+	err = en.WriteMapHeader(uint32(len(z)))
+	if err != nil {
+		err = msgp.WrapError(err)
+		return
+	}
+	for zb0005, zb0006 := range z {
+		err = en.WriteString(zb0005)
+		if err != nil {
+			err = msgp.WrapError(err)
+			return
+		}
+		// map header, size 1
+		// write "ReplicationStats"
+		err = en.Append(0x81, 0xb0, 0x52, 0x65, 0x70, 0x6c, 0x69, 0x63, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x53, 0x74, 0x61, 0x74, 0x73)
+		if err != nil {
+			return
+		}
+		err = zb0006.ReplicationStats.EncodeMsg(en)
+		if err != nil {
+			err = msgp.WrapError(err, zb0005, "ReplicationStats")
+			return
+		}
+	}
+	return
+}
+
+// MarshalMsg implements msgp.Marshaler
+func (z BucketStatsMap) MarshalMsg(b []byte) (o []byte, err error) {
+	o = msgp.Require(b, z.Msgsize())
+	o = msgp.AppendMapHeader(o, uint32(len(z)))
+	for zb0005, zb0006 := range z {
+		o = msgp.AppendString(o, zb0005)
+		// map header, size 1
+		// string "ReplicationStats"
+		o = append(o, 0x81, 0xb0, 0x52, 0x65, 0x70, 0x6c, 0x69, 0x63, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x53, 0x74, 0x61, 0x74, 0x73)
+		o, err = zb0006.ReplicationStats.MarshalMsg(o)
+		if err != nil {
+			err = msgp.WrapError(err, zb0005, "ReplicationStats")
+			return
+		}
+	}
+	return
+}
+
+// UnmarshalMsg implements msgp.Unmarshaler
+func (z *BucketStatsMap) UnmarshalMsg(bts []byte) (o []byte, err error) {
+	var zb0003 uint32
+	zb0003, bts, err = msgp.ReadMapHeaderBytes(bts)
+	if err != nil {
+		err = msgp.WrapError(err)
+		return
+	}
+	if (*z) == nil {
+		(*z) = make(BucketStatsMap, zb0003)
+	} else if len((*z)) > 0 {
+		for key := range *z {
+			delete((*z), key)
+		}
+	}
+	for zb0003 > 0 {
+		var zb0001 string
+		var zb0002 BucketStats
+		zb0003--
+		zb0001, bts, err = msgp.ReadStringBytes(bts)
+		if err != nil {
+			err = msgp.WrapError(err)
+			return
+		}
+		var field []byte
+		_ = field
+		var zb0004 uint32
+		zb0004, bts, err = msgp.ReadMapHeaderBytes(bts)
+		if err != nil {
+			err = msgp.WrapError(err, zb0001)
+			return
+		}
+		for zb0004 > 0 {
+			zb0004--
+			field, bts, err = msgp.ReadMapKeyZC(bts)
+			if err != nil {
+				err = msgp.WrapError(err, zb0001)
+				return
+			}
+			switch msgp.UnsafeString(field) {
+			case "ReplicationStats":
+				bts, err = zb0002.ReplicationStats.UnmarshalMsg(bts)
+				if err != nil {
+					err = msgp.WrapError(err, zb0001, "ReplicationStats")
+					return
+				}
+			default:
+				bts, err = msgp.Skip(bts)
+				if err != nil {
+					err = msgp.WrapError(err, zb0001)
+					return
+				}
+			}
+		}
+		(*z)[zb0001] = zb0002
+	}
+	o = bts
+	return
+}
+
+// Msgsize returns an upper bound estimate of the number of bytes occupied by the serialized message
+func (z BucketStatsMap) Msgsize() (s int) {
+	s = msgp.MapHeaderSize
+	if z != nil {
+		for zb0005, zb0006 := range z {
+			_ = zb0006
+			s += msgp.StringPrefixSize + len(zb0005) + 1 + 17 + zb0006.ReplicationStats.Msgsize()
+		}
+	}
+	return
+}
+
 // DecodeMsg implements msgp.Decodable
 func (z *ReplicationLatency) DecodeMsg(dc *msgp.Reader) (err error) {
 	var field []byte

cmd/bucket-stats_gen_test.go

@@ -348,6 +348,119 @@ func BenchmarkDecodeBucketStats(b *testing.B) {
 	}
 }
 
+func TestMarshalUnmarshalBucketStatsMap(t *testing.T) {
+	v := BucketStatsMap{}
+	bts, err := v.MarshalMsg(nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	left, err := v.UnmarshalMsg(bts)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(left) > 0 {
+		t.Errorf("%d bytes left over after UnmarshalMsg(): %q", len(left), left)
+	}
+
+	left, err = msgp.Skip(bts)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(left) > 0 {
+		t.Errorf("%d bytes left over after Skip(): %q", len(left), left)
+	}
+}
+
+func BenchmarkMarshalMsgBucketStatsMap(b *testing.B) {
+	v := BucketStatsMap{}
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.MarshalMsg(nil)
+	}
+}
+
+func BenchmarkAppendMsgBucketStatsMap(b *testing.B) {
+	v := BucketStatsMap{}
+	bts := make([]byte, 0, v.Msgsize())
+	bts, _ = v.MarshalMsg(bts[0:0])
+	b.SetBytes(int64(len(bts)))
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		bts, _ = v.MarshalMsg(bts[0:0])
+	}
+}
+
+func BenchmarkUnmarshalBucketStatsMap(b *testing.B) {
+	v := BucketStatsMap{}
+	bts, _ := v.MarshalMsg(nil)
+	b.ReportAllocs()
+	b.SetBytes(int64(len(bts)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, err := v.UnmarshalMsg(bts)
+		if err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+func TestEncodeDecodeBucketStatsMap(t *testing.T) {
+	v := BucketStatsMap{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+
+	m := v.Msgsize()
+	if buf.Len() > m {
+		t.Log("WARNING: TestEncodeDecodeBucketStatsMap Msgsize() is inaccurate")
+	}
+
+	vn := BucketStatsMap{}
+	err := msgp.Decode(&buf, &vn)
+	if err != nil {
+		t.Error(err)
+	}
+
+	buf.Reset()
+	msgp.Encode(&buf, &v)
+	err = msgp.NewReader(&buf).Skip()
+	if err != nil {
+		t.Error(err)
+	}
+}
+
+func BenchmarkEncodeBucketStatsMap(b *testing.B) {
+	v := BucketStatsMap{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+	b.SetBytes(int64(buf.Len()))
+	en := msgp.NewWriter(msgp.Nowhere)
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.EncodeMsg(en)
+	}
+	en.Flush()
+}
+
+func BenchmarkDecodeBucketStatsMap(b *testing.B) {
+	v := BucketStatsMap{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+	b.SetBytes(int64(buf.Len()))
+	rd := msgp.NewEndlessReader(buf.Bytes(), b)
+	dc := msgp.NewReader(rd)
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		err := v.DecodeMsg(dc)
+		if err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
 func TestMarshalUnmarshalReplicationLatency(t *testing.T) {
 	v := ReplicationLatency{}
 	bts, err := v.MarshalMsg(nil)

cmd/bucket-targets.go

@@ -120,9 +120,6 @@ func (sys *BucketTargetSys) SetTarget(ctx context.Context, bucket string, tgt *m
 		return BucketRemoteConnectionErr{Bucket: tgt.TargetBucket, Err: err}
 	}
 	if tgt.Type == madmin.ReplicationService {
-		if !globalIsErasure {
-			return NotImplemented{Message: "Replication is not implemented in " + getMinioMode()}
-		}
 		if !globalBucketVersioningSys.Enabled(bucket) {
 			return BucketReplicationSourceNotVersioned{Bucket: bucket}
 		}
@@ -184,9 +181,6 @@ func (sys *BucketTargetSys) RemoveTarget(ctx context.Context, bucket, arnStr str
 	if globalIsGateway {
 		return nil
 	}
-	if !globalIsErasure {
-		return NotImplemented{Message: "Replication is not implemented in " + getMinioMode()}
-	}
 
 	if arnStr == "" {
 		return BucketRemoteArnInvalid{Bucket: bucket}
@@ -201,7 +195,7 @@ func (sys *BucketTargetSys) RemoveTarget(ctx context.Context, bucket, arnStr str
 		// reject removal of remote target if replication configuration is present
 		rcfg, err := getReplicationConfig(ctx, bucket)
 		if err == nil {
-			for _, tgtArn := range rcfg.FilterTargetArns(replication.ObjectOpts{}) {
+			for _, tgtArn := range rcfg.FilterTargetArns(replication.ObjectOpts{OpType: replication.AllReplicationType}) {
 				if err == nil && (tgtArn == arnStr || rcfg.RoleArn == arnStr) {
 					sys.RLock()
 					_, ok := sys.arnRemotesMap[arnStr]

cmd/bucket-versioning-handler.go

@@ -18,12 +18,14 @@
 package cmd
 
 import (
+	"encoding/base64"
 	"encoding/xml"
 	"io"
 	"net/http"
 
 	humanize "github.com/dustin/go-humanize"
 	"github.com/gorilla/mux"
+	"github.com/minio/madmin-go"
 	"github.com/minio/minio/internal/bucket/versioning"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/bucket/policy"
@@ -63,28 +65,28 @@ func (api objectAPIHandlers) PutBucketVersioningHandler(w http.ResponseWriter, r
 		return
 	}
 
-	if globalSiteReplicationSys.isEnabled() {
+	if globalSiteReplicationSys.isEnabled() && !v.Enabled() {
 		writeErrorResponse(ctx, w, APIError{
 			Code:           "InvalidBucketState",
-			Description:    "Cluster replication is enabled for this site, so the versioning state cannot be changed.",
-			HTTPStatusCode: http.StatusConflict,
+			Description:    "Cluster replication is enabled on this site, versioning cannot be suspended on bucket.",
+			HTTPStatusCode: http.StatusBadRequest,
 		}, r.URL)
 		return
 	}
 
-	if rcfg, _ := globalBucketObjectLockSys.Get(bucket); rcfg.LockEnabled && v.Suspended() {
+	if rcfg, _ := globalBucketObjectLockSys.Get(bucket); rcfg.LockEnabled && (v.Suspended() || v.PrefixesExcluded()) {
 		writeErrorResponse(ctx, w, APIError{
 			Code:           "InvalidBucketState",
-			Description:    "An Object Lock configuration is present on this bucket, so the versioning state cannot be changed.",
-			HTTPStatusCode: http.StatusConflict,
+			Description:    "An Object Lock configuration is present on this bucket, versioning cannot be suspended.",
+			HTTPStatusCode: http.StatusBadRequest,
 		}, r.URL)
 		return
 	}
 	if _, err := getReplicationConfig(ctx, bucket); err == nil && v.Suspended() {
 		writeErrorResponse(ctx, w, APIError{
 			Code:           "InvalidBucketState",
-			Description:    "A replication configuration is present on this bucket, so the versioning state cannot be changed.",
-			HTTPStatusCode: http.StatusConflict,
+			Description:    "A replication configuration is present on this bucket, bucket wide versioning cannot be suspended.",
+			HTTPStatusCode: http.StatusBadRequest,
 		}, r.URL)
 		return
 	}
@@ -100,6 +102,20 @@ func (api objectAPIHandlers) PutBucketVersioningHandler(w http.ResponseWriter, r
 		return
 	}
 
+	// Call site replication hook.
+	//
+	// We encode the xml bytes as base64 to ensure there are no encoding
+	// errors.
+	cfgStr := base64.StdEncoding.EncodeToString(configData)
+	if err = globalSiteReplicationSys.BucketMetaHook(ctx, madmin.SRBucketMeta{
+		Type:       madmin.SRBucketMetaTypeVersionConfig,
+		Bucket:     bucket,
+		Versioning: &cfgStr,
+	}); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
 	writeSuccessResponseHeadersOnly(w)
 }
 

cmd/bucket-versioning.go

@@ -17,44 +17,69 @@
 
 package cmd
 
-import "github.com/minio/minio/internal/bucket/versioning"
+import (
+	"strings"
+
+	"github.com/minio/minio/internal/bucket/versioning"
+	"github.com/minio/minio/internal/logger"
+)
 
 // BucketVersioningSys - policy subsystem.
 type BucketVersioningSys struct{}
 
 // Enabled enabled versioning?
 func (sys *BucketVersioningSys) Enabled(bucket string) bool {
-	vc, err := globalBucketMetadataSys.GetVersioningConfig(bucket)
+	vc, err := sys.Get(bucket)
 	if err != nil {
-		return false
+		logger.CriticalIf(GlobalContext, err)
 	}
 	return vc.Enabled()
 }
 
+// PrefixEnabled returns true is versioning is enabled at bucket level and if
+// the given prefix doesn't match any excluded prefixes pattern. This is
+// part of a MinIO versioning configuration extension.
+func (sys *BucketVersioningSys) PrefixEnabled(bucket, prefix string) bool {
+	vc, err := sys.Get(bucket)
+	if err != nil {
+		logger.CriticalIf(GlobalContext, err)
+	}
+	return vc.PrefixEnabled(prefix)
+}
+
 // Suspended suspended versioning?
 func (sys *BucketVersioningSys) Suspended(bucket string) bool {
-	vc, err := globalBucketMetadataSys.GetVersioningConfig(bucket)
+	vc, err := sys.Get(bucket)
 	if err != nil {
-		return false
+		logger.CriticalIf(GlobalContext, err)
 	}
 	return vc.Suspended()
 }
 
+// PrefixSuspended returns true if the given prefix matches an excluded prefix
+// pattern. This is part of a MinIO versioning configuration extension.
+func (sys *BucketVersioningSys) PrefixSuspended(bucket, prefix string) bool {
+	vc, err := sys.Get(bucket)
+	if err != nil {
+		logger.CriticalIf(GlobalContext, err)
+	}
+
+	return vc.PrefixSuspended(prefix)
+}
+
 // Get returns stored bucket policy
 func (sys *BucketVersioningSys) Get(bucket string) (*versioning.Versioning, error) {
 	if globalIsGateway {
-		objAPI := newObjectLayerFn()
-		if objAPI == nil {
-			return nil, errServerNotInitialized
-		}
-		return nil, NotImplemented{}
+		// Gateway does not implement versioning.
+		return &versioning.Versioning{XMLNS: "http://s3.amazonaws.com/doc/2006-03-01/"}, nil
 	}
-	return globalBucketMetadataSys.GetVersioningConfig(bucket)
-}
 
-// Reset BucketVersioningSys to initial state.
-func (sys *BucketVersioningSys) Reset() {
-	// There is currently no internal state.
+	if bucket == minioMetaBucket || strings.HasPrefix(bucket, minioMetaBucket) {
+		return &versioning.Versioning{XMLNS: "http://s3.amazonaws.com/doc/2006-03-01/"}, nil
+	}
+
+	vcfg, _, err := globalBucketMetadataSys.GetVersioningConfig(bucket)
+	return vcfg, err
 }
 
 // NewBucketVersioningSys - creates new versioning system.

cmd/callhome.go

@@ -0,0 +1,136 @@
+// Copyright (c) 2015-2022 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"
+	"fmt"
+	"math/rand"
+	"time"
+
+	"github.com/minio/madmin-go"
+	"github.com/minio/minio/internal/logger"
+	uatomic "go.uber.org/atomic"
+)
+
+const (
+	// callhomeSchemaVersion1 is callhome schema version 1
+	callhomeSchemaVersion1 = "1"
+
+	// callhomeSchemaVersion is current callhome schema version.
+	callhomeSchemaVersion = callhomeSchemaVersion1
+
+	// callhomeCycleDefault is the default interval between two callhome cycles (24hrs)
+	callhomeCycleDefault = 24 * time.Hour
+)
+
+// CallhomeInfo - Contains callhome information
+type CallhomeInfo struct {
+	SchemaVersion string             `json:"schema_version"`
+	AdminInfo     madmin.InfoMessage `json:"admin_info"`
+}
+
+var (
+	enableCallhome            = uatomic.NewBool(false)
+	callhomeLeaderLockTimeout = newDynamicTimeout(30*time.Second, 10*time.Second)
+	callhomeFreq              = uatomic.NewDuration(callhomeCycleDefault)
+)
+
+func updateCallhomeParams(ctx context.Context, objAPI ObjectLayer) {
+	alreadyEnabled := enableCallhome.Load()
+	enableCallhome.Store(globalCallhomeConfig.Enable)
+	callhomeFreq.Store(globalCallhomeConfig.Frequency)
+
+	// If callhome was disabled earlier and has now been enabled,
+	// initialize the callhome process again.
+	if !alreadyEnabled && enableCallhome.Load() {
+		initCallhome(ctx, objAPI)
+	}
+}
+
+// initCallhome will start the callhome task in the background.
+func initCallhome(ctx context.Context, objAPI ObjectLayer) {
+	go func() {
+		r := rand.New(rand.NewSource(time.Now().UnixNano()))
+		// Leader node (that successfully acquires the lock inside runCallhome)
+		// will keep performing the callhome. If the leader goes down for some reason,
+		// the lock will be released and another node will acquire it and take over
+		// because of this loop.
+		for {
+			runCallhome(ctx, objAPI)
+			if !enableCallhome.Load() {
+				return
+			}
+
+			// callhome running on a different node.
+			// sleep for some time and try again.
+			duration := time.Duration(r.Float64() * float64(callhomeFreq.Load()))
+			if duration < time.Second {
+				// Make sure to sleep atleast a second to avoid high CPU ticks.
+				duration = time.Second
+			}
+			time.Sleep(duration)
+
+			if !enableCallhome.Load() {
+				return
+			}
+		}
+	}()
+}
+
+func runCallhome(ctx context.Context, objAPI ObjectLayer) {
+	// Make sure only 1 callhome is running on the cluster.
+	locker := objAPI.NewNSLock(minioMetaBucket, "callhome/runCallhome.lock")
+	lkctx, err := locker.GetLock(ctx, callhomeLeaderLockTimeout)
+	if err != nil {
+		return
+	}
+
+	ctx = lkctx.Context()
+	defer locker.Unlock(lkctx.Cancel)
+
+	callhomeTimer := time.NewTimer(callhomeFreq.Load())
+	defer callhomeTimer.Stop()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-callhomeTimer.C:
+			if !enableCallhome.Load() {
+				// Stop the processing as callhome got disabled
+				return
+			}
+			performCallhome(ctx)
+
+			// Reset the timer for next cycle.
+			callhomeTimer.Reset(callhomeFreq.Load())
+		}
+	}
+}
+
+func performCallhome(ctx context.Context) {
+	err := sendCallhomeInfo(
+		CallhomeInfo{
+			SchemaVersion: callhomeSchemaVersion,
+			AdminInfo:     getServerInfo(ctx, nil),
+		})
+	if err != nil {
+		logger.LogIf(ctx, fmt.Errorf("Unable to perform callhome: %w", err))
+	}
+}

cmd/common-main.go

@@ -190,6 +190,10 @@ func minioConfigToConsoleFeatures() {
 		os.Setenv("CONSOLE_PROMETHEUS_URL", value)
 		if value := env.Get("MINIO_PROMETHEUS_JOB_ID", "minio-job"); value != "" {
 			os.Setenv("CONSOLE_PROMETHEUS_JOB_ID", value)
+			// Support additional labels for more granular filtering.
+			if value := env.Get("MINIO_PROMETHEUS_EXTRA_LABELS", ""); value != "" {
+				os.Setenv("CONSOLE_PROMETHEUS_EXTRA_LABELS", value)
+			}
 		}
 	}
 	// Enable if LDAP is enabled.
@@ -226,8 +230,8 @@ func minioConfigToConsoleFeatures() {
 	if globalSubnetConfig.APIKey != "" {
 		os.Setenv("CONSOLE_SUBNET_API_KEY", globalSubnetConfig.APIKey)
 	}
-	if globalSubnetConfig.Proxy != "" {
-		os.Setenv("CONSOLE_SUBNET_PROXY", globalSubnetConfig.Proxy)
+	if globalSubnetConfig.ProxyURL != nil {
+		os.Setenv("CONSOLE_SUBNET_PROXY", globalSubnetConfig.ProxyURL.String())
 	}
 }
 

cmd/config-current.go

@@ -28,15 +28,18 @@ import (
 	"github.com/minio/minio/internal/config"
 	"github.com/minio/minio/internal/config/api"
 	"github.com/minio/minio/internal/config/cache"
+	"github.com/minio/minio/internal/config/callhome"
 	"github.com/minio/minio/internal/config/compress"
 	"github.com/minio/minio/internal/config/dns"
 	"github.com/minio/minio/internal/config/etcd"
 	"github.com/minio/minio/internal/config/heal"
 	xldap "github.com/minio/minio/internal/config/identity/ldap"
 	"github.com/minio/minio/internal/config/identity/openid"
+	idplugin "github.com/minio/minio/internal/config/identity/plugin"
 	xtls "github.com/minio/minio/internal/config/identity/tls"
 	"github.com/minio/minio/internal/config/notify"
 	"github.com/minio/minio/internal/config/policy/opa"
+	polplugin "github.com/minio/minio/internal/config/policy/plugin"
 	"github.com/minio/minio/internal/config/scanner"
 	"github.com/minio/minio/internal/config/storageclass"
 	"github.com/minio/minio/internal/config/subnet"
@@ -55,7 +58,9 @@ func initHelp() {
 		config.IdentityLDAPSubSys:   xldap.DefaultKVS,
 		config.IdentityOpenIDSubSys: openid.DefaultKVS,
 		config.IdentityTLSSubSys:    xtls.DefaultKVS,
+		config.IdentityPluginSubSys: idplugin.DefaultKVS,
 		config.PolicyOPASubSys:      opa.DefaultKVS,
+		config.PolicyPluginSubSys:   polplugin.DefaultKVS,
 		config.SiteSubSys:           config.DefaultSiteKVS,
 		config.RegionSubSys:         config.DefaultRegionKVS,
 		config.APISubSys:            api.DefaultKVS,
@@ -66,6 +71,7 @@ func initHelp() {
 		config.HealSubSys:           heal.DefaultKVS,
 		config.ScannerSubSys:        scanner.DefaultKVS,
 		config.SubnetSubSys:         subnet.DefaultKVS,
+		config.CallhomeSubSys:       callhome.DefaultKVS,
 	}
 	for k, v := range notify.DefaultNotificationKVS {
 		kvs[k] = v
@@ -107,8 +113,12 @@ func initHelp() {
 			Description: "enable X.509 TLS certificate SSO support",
 		},
 		config.HelpKV{
-			Key:         config.PolicyOPASubSys,
-			Description: "[DEPRECATED] enable external OPA for policy enforcement",
+			Key:         config.IdentityPluginSubSys,
+			Description: "enable Identity Plugin via external hook",
+		},
+		config.HelpKV{
+			Key:         config.PolicyPluginSubSys,
+			Description: "enable Access Management Plugin for policy enforcement",
 		},
 		config.HelpKV{
 			Key:         config.APISubSys,
@@ -193,6 +203,12 @@ func initHelp() {
 			Description: "set subnet config for the cluster e.g. api key",
 			Optional:    true,
 		},
+		config.HelpKV{
+			Key:         config.CallhomeSubSys,
+			Type:        "string",
+			Description: "enable callhome for the cluster",
+			Optional:    true,
+		},
 	}
 
 	if globalIsErasure {
@@ -218,7 +234,9 @@ func initHelp() {
 		config.IdentityOpenIDSubSys: openid.Help,
 		config.IdentityLDAPSubSys:   xldap.Help,
 		config.IdentityTLSSubSys:    xtls.Help,
+		config.IdentityPluginSubSys: idplugin.Help,
 		config.PolicyOPASubSys:      opa.Help,
+		config.PolicyPluginSubSys:   polplugin.Help,
 		config.LoggerWebhookSubSys:  logger.Help,
 		config.AuditWebhookSubSys:   logger.HelpWebhook,
 		config.AuditKafkaSubSys:     logger.HelpKafka,
@@ -233,6 +251,7 @@ func initHelp() {
 		config.NotifyWebhookSubSys:  notify.HelpWebhook,
 		config.NotifyESSubSys:       notify.HelpES,
 		config.SubnetSubSys:         subnet.HelpSubnet,
+		config.CallhomeSubSys:       callhome.HelpCallhome,
 	}
 
 	config.RegisterHelpSubSys(helpMap)
@@ -243,6 +262,10 @@ func initHelp() {
 			Key:         config.RegionSubSys,
 			Description: "[DEPRECATED - use `site` instead] label the location of the server",
 		},
+		config.PolicyOPASubSys: {
+			Key:         config.PolicyOPASubSys,
+			Description: "[DEPRECATED - use `policy_plugin` instead] enable external OPA for policy enforcement",
+		},
 	}
 
 	config.RegisterHelpDeprecatedSubSys(deprecatedHelpKVMap)
@@ -335,14 +358,34 @@ func validateSubSysConfig(s config.Config, subSys string, objAPI ObjectLayer) er
 		if _, err := xtls.Lookup(s[config.IdentityTLSSubSys][config.Default]); err != nil {
 			return err
 		}
+	case config.IdentityPluginSubSys:
+		if _, err := idplugin.LookupConfig(s[config.IdentityPluginSubSys][config.Default],
+			NewGatewayHTTPTransport(), xhttp.DrainBody, globalSite.Region); err != nil {
+			return err
+		}
 	case config.SubnetSubSys:
 		if _, err := subnet.LookupConfig(s[config.SubnetSubSys][config.Default]); err != nil {
 			return err
 		}
+	case config.CallhomeSubSys:
+		if _, err := callhome.LookupConfig(s[config.CallhomeSubSys][config.Default]); err != nil {
+			return err
+		}
 	case config.PolicyOPASubSys:
-		if _, err := opa.LookupConfig(s[config.PolicyOPASubSys][config.Default],
+		// In case legacy OPA config is being set, we treat it as if the
+		// AuthZPlugin is being set.
+		subSys = config.PolicyPluginSubSys
+		fallthrough
+	case config.PolicyPluginSubSys:
+		if ppargs, err := polplugin.LookupConfig(s[config.PolicyPluginSubSys][config.Default],
 			NewGatewayHTTPTransport(), xhttp.DrainBody); err != nil {
 			return err
+		} else if ppargs.URL == nil {
+			// Check if legacy opa is configured.
+			if _, err := opa.LookupConfig(s[config.PolicyOPASubSys][config.Default],
+				NewGatewayHTTPTransport(), xhttp.DrainBody); err != nil {
+				return err
+			}
 		}
 	default:
 		if config.LoggerSubSystems.Contains(subSys) {
@@ -523,19 +566,39 @@ func lookupConfigs(s config.Config, objAPI ObjectLayer) {
 		logger.LogIf(ctx, fmt.Errorf("Unable to initialize OpenID: %w", err))
 	}
 
-	opaCfg, err := opa.LookupConfig(s[config.PolicyOPASubSys][config.Default],
-		NewGatewayHTTPTransport(), xhttp.DrainBody)
-	if err != nil {
-		logger.LogIf(ctx, fmt.Errorf("Unable to initialize OPA: %w", err))
-	}
-	globalPolicyOPA = opa.New(opaCfg)
-
 	globalLDAPConfig, err = xldap.Lookup(s[config.IdentityLDAPSubSys][config.Default],
 		globalRootCAs)
 	if err != nil {
 		logger.LogIf(ctx, fmt.Errorf("Unable to parse LDAP configuration: %w", err))
 	}
 
+	authNPluginCfg, err := idplugin.LookupConfig(s[config.IdentityPluginSubSys][config.Default],
+		NewGatewayHTTPTransport(), xhttp.DrainBody, globalSite.Region)
+	if err != nil {
+		logger.LogIf(ctx, fmt.Errorf("Unable to initialize AuthNPlugin: %w", err))
+	}
+	globalAuthNPlugin = idplugin.New(authNPluginCfg)
+
+	authZPluginCfg, err := polplugin.LookupConfig(s[config.PolicyPluginSubSys][config.Default],
+		NewGatewayHTTPTransport(), xhttp.DrainBody)
+	if err != nil {
+		logger.LogIf(ctx, fmt.Errorf("Unable to initialize AuthZPlugin: %w", err))
+	}
+	if authZPluginCfg.URL == nil {
+		opaCfg, err := opa.LookupConfig(s[config.PolicyOPASubSys][config.Default],
+			NewGatewayHTTPTransport(), xhttp.DrainBody)
+		if err != nil {
+			logger.LogIf(ctx, fmt.Errorf("Unable to initialize AuthZPlugin from legacy OPA config: %w", err))
+		} else {
+			authZPluginCfg.URL = opaCfg.URL
+			authZPluginCfg.AuthToken = opaCfg.AuthToken
+			authZPluginCfg.Transport = opaCfg.Transport
+			authZPluginCfg.CloseRespFn = opaCfg.CloseRespFn
+		}
+	}
+
+	setGlobalAuthZPlugin(polplugin.New(authZPluginCfg))
+
 	globalSubnetConfig, err = subnet.LookupConfig(s[config.SubnetSubSys][config.Default])
 	if err != nil {
 		logger.LogIf(ctx, fmt.Errorf("Unable to parse subnet configuration: %w", err))
@@ -599,7 +662,7 @@ func applyDynamicConfigForSubSys(ctx context.Context, objAPI ObjectLayer, s conf
 			return fmt.Errorf("Unable to apply scanner config: %w", err)
 		}
 		// update dynamic scanner values.
-		scannerCycle.Update(scannerCfg.Cycle)
+		scannerCycle.Store(scannerCfg.Cycle)
 		logger.LogIf(ctx, scannerSleeper.Update(scannerCfg.Delay, scannerCfg.MaxWait))
 	case config.LoggerWebhookSubSys:
 		loggerCfg, err := logger.LookupConfigForSubSys(s, config.LoggerWebhookSubSys)
@@ -609,7 +672,7 @@ func applyDynamicConfigForSubSys(ctx context.Context, objAPI ObjectLayer, s conf
 		userAgent := getUserAgent(getMinioMode())
 		for n, l := range loggerCfg.HTTP {
 			if l.Enabled {
-				l.LogOnce = logger.LogOnceIf
+				l.LogOnce = logger.LogOnceConsoleIf
 				l.UserAgent = userAgent
 				l.Transport = NewGatewayHTTPTransportWithClientCerts(l.ClientCert, l.ClientKey)
 				loggerCfg.HTTP[n] = l
@@ -627,7 +690,7 @@ func applyDynamicConfigForSubSys(ctx context.Context, objAPI ObjectLayer, s conf
 		userAgent := getUserAgent(getMinioMode())
 		for n, l := range loggerCfg.AuditWebhook {
 			if l.Enabled {
-				l.LogOnce = logger.LogOnceIf
+				l.LogOnce = logger.LogOnceConsoleIf
 				l.UserAgent = userAgent
 				l.Transport = NewGatewayHTTPTransportWithClientCerts(l.ClientCert, l.ClientKey)
 				loggerCfg.AuditWebhook[n] = l
@@ -669,6 +732,14 @@ func applyDynamicConfigForSubSys(ctx context.Context, objAPI ObjectLayer, s conf
 				}
 			}
 		}
+	case config.CallhomeSubSys:
+		callhomeCfg, err := callhome.LookupConfig(s[config.CallhomeSubSys][config.Default])
+		if err != nil {
+			logger.LogIf(ctx, fmt.Errorf("Unable to load callhome config: %w", err))
+		} else {
+			globalCallhomeConfig = callhomeCfg
+			updateCallhomeParams(ctx, objAPI)
+		}
 	}
 	globalServerConfigMu.Lock()
 	defer globalServerConfigMu.Unlock()

cmd/consolelogger.go

@@ -45,7 +45,7 @@ type HTTPConsoleLoggerSys struct {
 // NewConsoleLogger - creates new HTTPConsoleLoggerSys with all nodes subscribed to
 // the console logging pub sub system
 func NewConsoleLogger(ctx context.Context) *HTTPConsoleLoggerSys {
-	ps := pubsub.New()
+	ps := pubsub.New(8)
 	return &HTTPConsoleLoggerSys{
 		pubsub:  ps,
 		console: console.New(),
@@ -75,7 +75,7 @@ func (sys *HTTPConsoleLoggerSys) HasLogListeners() bool {
 }
 
 // Subscribe starts console logging for this node.
-func (sys *HTTPConsoleLoggerSys) Subscribe(subCh chan interface{}, doneCh <-chan struct{}, node string, last int, logKind string, filter func(entry interface{}) bool) {
+func (sys *HTTPConsoleLoggerSys) Subscribe(subCh chan interface{}, doneCh <-chan struct{}, node string, last int, logKind string, filter func(entry interface{}) bool) error {
 	// Enable console logging for remote client.
 	if !sys.HasLogListeners() {
 		logger.AddSystemTarget(sys)
@@ -111,11 +111,12 @@ func (sys *HTTPConsoleLoggerSys) Subscribe(subCh chan interface{}, doneCh <-chan
 			select {
 			case subCh <- entry:
 			case <-doneCh:
-				return
+				return nil
 			}
 		}
 	}
-	sys.pubsub.Subscribe(subCh, doneCh, filter)
+
+	return sys.pubsub.Subscribe(subCh, doneCh, filter)
 }
 
 // Init if HTTPConsoleLoggerSys is valid, always returns nil right now

cmd/data-scanner.go

@@ -45,6 +45,7 @@ import (
 	"github.com/minio/minio/internal/event"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/console"
+	uatomic "go.uber.org/atomic"
 )
 
 const (
@@ -66,9 +67,7 @@ var (
 	dataScannerLeaderLockTimeout = newDynamicTimeout(30*time.Second, 10*time.Second)
 	// Sleeper values are updated when config is loaded.
 	scannerSleeper = newDynamicSleeper(10, 10*time.Second)
-	scannerCycle   = &safeDuration{
-		t: dataScannerStartDelay,
-	}
+	scannerCycle   = uatomic.NewDuration(dataScannerStartDelay)
 )
 
 // initDataScanner will start the scanner in the background.
@@ -78,7 +77,7 @@ func initDataScanner(ctx context.Context, objAPI ObjectLayer) {
 		// Run the data scanner in a loop
 		for {
 			runDataScanner(ctx, objAPI)
-			duration := time.Duration(r.Float64() * float64(scannerCycle.Get()))
+			duration := time.Duration(r.Float64() * float64(scannerCycle.Load()))
 			if duration < time.Second {
 				// Make sure to sleep atleast a second to avoid high CPU ticks.
 				duration = time.Second
@@ -88,23 +87,6 @@ func initDataScanner(ctx context.Context, objAPI ObjectLayer) {
 	}()
 }
 
-type safeDuration struct {
-	sync.Mutex
-	t time.Duration
-}
-
-func (s *safeDuration) Update(t time.Duration) {
-	s.Lock()
-	defer s.Unlock()
-	s.t = t
-}
-
-func (s *safeDuration) Get() time.Duration {
-	s.Lock()
-	defer s.Unlock()
-	return s.t
-}
-
 func getCycleScanMode(currentCycle, bitrotStartCycle uint64, bitrotStartTime time.Time) madmin.HealScanMode {
 	bitrotCycle := globalHealConfig.BitrotScanCycle()
 	switch bitrotCycle {
@@ -189,7 +171,7 @@ func runDataScanner(pctx context.Context, objAPI ObjectLayer) {
 		}
 	}
 
-	scannerTimer := time.NewTimer(scannerCycle.Get())
+	scannerTimer := time.NewTimer(scannerCycle.Load())
 	defer scannerTimer.Stop()
 
 	for {
@@ -197,9 +179,6 @@ func runDataScanner(pctx context.Context, objAPI ObjectLayer) {
 		case <-ctx.Done():
 			return
 		case <-scannerTimer.C:
-			// Reset the timer for next cycle.
-			scannerTimer.Reset(scannerCycle.Get())
-
 			if intDataUpdateTracker.debug {
 				console.Debugln("starting scanner cycle")
 			}
@@ -232,6 +211,9 @@ func runDataScanner(pctx context.Context, objAPI ObjectLayer) {
 					logger.LogIf(ctx, err)
 				}
 			}
+
+			// Reset the timer for next cycle.
+			scannerTimer.Reset(scannerCycle.Load())
 		}
 	}
 }
@@ -1057,9 +1039,13 @@ func (i *scannerItem) applyNewerNoncurrentVersionLimit(ctx context.Context, _ Ob
 	fivs = fivs[:lim+1]
 
 	rcfg, _ := globalBucketObjectLockSys.Get(i.bucket)
+	vcfg, _ := globalBucketVersioningSys.Get(i.bucket)
+
+	versioned := vcfg != nil && vcfg.Versioned(i.objectPath())
+
 	toDel := make([]ObjectToDelete, 0, len(overflowVersions))
 	for _, fi := range overflowVersions {
-		obj := fi.ToObjectInfo(i.bucket, i.objectPath())
+		obj := fi.ToObjectInfo(i.bucket, i.objectPath(), versioned)
 		// skip versions with object locking enabled
 		if rcfg.LockEnabled && enforceRetentionForDeletion(ctx, obj) {
 			if i.debug {
@@ -1184,7 +1170,8 @@ func applyExpiryOnNonTransitionedObjects(ctx context.Context, objLayer ObjectLay
 		opts.VersionID = obj.VersionID
 	}
 	if opts.VersionID == "" {
-		opts.Versioned = globalBucketVersioningSys.Enabled(obj.Bucket)
+		opts.Versioned = globalBucketVersioningSys.PrefixEnabled(obj.Bucket, obj.Name)
+		opts.VersionSuspended = globalBucketVersioningSys.PrefixSuspended(obj.Bucket, obj.Name)
 	}
 
 	obj, err := objLayer.DeleteObject(ctx, obj.Bucket, obj.Name, opts)
@@ -1243,6 +1230,9 @@ func (i *scannerItem) objectPath() string {
 // healReplication will heal a scanned item that has failed replication.
 func (i *scannerItem) healReplication(ctx context.Context, o ObjectLayer, oi ObjectInfo, sizeS *sizeSummary) {
 	roi := getHealReplicateObjectInfo(oi, i.replication)
+	if !roi.Dsc.ReplicateAny() {
+		return
+	}
 
 	if oi.DeleteMarker || !oi.VersionPurgeStatus.Empty() {
 		// heal delete marker replication failure or versioned delete replication failure

cmd/disk-cache-backend.go

@@ -884,6 +884,7 @@ func (c *diskCache) put(ctx context.Context, bucket, object string, data io.Read
 	}
 
 	if err := os.MkdirAll(cachePath, 0o777); err != nil {
+		removeAll(cachePath)
 		return oi, err
 	}
 	metadata := cloneMSS(opts.UserDefined)
@@ -892,6 +893,7 @@ func (c *diskCache) put(ctx context.Context, bucket, object string, data io.Read
 	if globalCacheKMS != nil {
 		reader, err = newCacheEncryptReader(data, bucket, object, metadata)
 		if err != nil {
+			removeAll(cachePath)
 			return oi, err
 		}
 		actualSize, _ = sio.EncryptedSize(uint64(size))
@@ -1628,8 +1630,6 @@ func (c *diskCache) cleanupStaleUploads(ctx context.Context) {
 		case <-ctx.Done():
 			return
 		case <-timer.C:
-			// Reset for the next interval
-			timer.Reset(cacheStaleUploadCleanupInterval)
 			now := time.Now()
 			readDirFn(pathJoin(c.dir, minioMetaBucket, cacheMultipartDir), func(shaDir string, typ os.FileMode) error {
 				return readDirFn(pathJoin(c.dir, minioMetaBucket, cacheMultipartDir, shaDir), func(uploadIDDir string, typ os.FileMode) error {
@@ -1660,6 +1660,9 @@ func (c *diskCache) cleanupStaleUploads(ctx context.Context) {
 				}
 				return nil
 			})
+
+			// Reset for the next interval
+			timer.Reset(cacheStaleUploadCleanupInterval)
 		}
 	}
 }

cmd/disk-cache.go

@@ -33,7 +33,6 @@ import (
 	"github.com/minio/minio/internal/config/cache"
 	"github.com/minio/minio/internal/disk"
 	"github.com/minio/minio/internal/hash"
-	xhttp "github.com/minio/minio/internal/http"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/minio/internal/sync/errgroup"
 	"github.com/minio/pkg/wildcard"
@@ -739,35 +738,30 @@ func (c *cacheObjects) PutObject(ctx context.Context, bucket, object string, r *
 	// Initialize pipe to stream data to cache
 	rPipe, wPipe := io.Pipe()
 	infoCh := make(chan ObjectInfo)
-	errorCh := make(chan error)
 	go func() {
+		defer close(infoCh)
 		info, err := putObjectFn(ctx, bucket, object, NewPutObjReader(hashReader), opts)
-		if err != nil {
-			close(infoCh)
-			pipeReader.CloseWithError(err)
-			rPipe.CloseWithError(err)
-			errorCh <- err
-			return
+		pipeReader.CloseWithError(err)
+		rPipe.CloseWithError(err)
+		if err == nil {
+			infoCh <- info
 		}
-		close(errorCh)
-		infoCh <- info
 	}()
 
 	go func() {
 		_, err := dcache.put(lkctx.Context(), bucket, object, rPipe, r.Size(), nil, opts, false, false)
 		if err != nil {
-			rPipe.CloseWithError(err)
-			return
+			logger.LogIf(lkctx.Context(), err)
 		}
+		// We do not care about errors to cached backend.
+		rPipe.Close()
 	}()
 
 	mwriter := cacheMultiWriter(pipeWriter, wPipe)
 	_, err = io.Copy(mwriter, r)
 	pipeWriter.Close()
 	wPipe.Close()
-
 	if err != nil {
-		err = <-errorCh
 		return ObjectInfo{}, err
 	}
 	info := <-infoCh
@@ -803,8 +797,7 @@ func (c *cacheObjects) uploadObject(ctx context.Context, oi ObjectInfo) {
 		return
 	}
 	var opts ObjectOptions
-	opts.UserDefined = make(map[string]string)
-	opts.UserDefined[xhttp.ContentMD5] = oi.UserDefined["content-md5"]
+	opts.UserDefined = cloneMSS(oi.UserDefined)
 	objInfo, err := c.InnerPutObjectFn(ctx, oi.Bucket, oi.Name, NewPutObjReader(hashReader), opts)
 	wbCommitStatus := CommitComplete
 	size := objInfo.Size

cmd/endpoint.go

@@ -582,7 +582,7 @@ func CreateEndpoints(serverAddr string, foundLocal bool, args ...[]string) (Endp
 			return endpoints, setupType, config.ErrInvalidFSEndpoint(nil).Msg("use path style endpoint for FS setup")
 		}
 		endpoints = append(endpoints, endpoint)
-		setupType = FSSetupType
+		setupType = ErasureSDSetupType
 
 		// Check for cross device mounts if any.
 		if err = checkCrossDeviceMounts(endpoints); err != nil {

cmd/endpoint_test.go

@@ -231,10 +231,10 @@ func TestCreateEndpoints(t *testing.T) {
 	}{
 		{"localhost", [][]string{}, "", Endpoints{}, -1, fmt.Errorf("address localhost: missing port in address")},
 
-		// FS Setup
+		// Erasure Single Drive
 		{"localhost:9000", [][]string{{"http://localhost/d1"}}, "", Endpoints{}, -1, fmt.Errorf("use path style endpoint for FS setup")},
-		{":443", [][]string{{"/d1"}}, ":443", Endpoints{Endpoint{URL: &url.URL{Path: mustAbs("/d1")}, IsLocal: true}}, FSSetupType, nil},
-		{"localhost:10000", [][]string{{"/d1"}}, "localhost:10000", Endpoints{Endpoint{URL: &url.URL{Path: mustAbs("/d1")}, IsLocal: true}}, FSSetupType, nil},
+		{":443", [][]string{{"/d1"}}, ":443", Endpoints{Endpoint{URL: &url.URL{Path: mustAbs("/d1")}, IsLocal: true}}, ErasureSDSetupType, nil},
+		{"localhost:10000", [][]string{{"/d1"}}, "localhost:10000", Endpoints{Endpoint{URL: &url.URL{Path: mustAbs("/d1")}, IsLocal: true}}, ErasureSDSetupType, nil},
 		{"localhost:9000", [][]string{{"https://127.0.0.1:9000/d1", "https://localhost:9001/d1", "https://example.com/d1", "https://example.com/d2"}}, "", Endpoints{}, -1, fmt.Errorf("path '/d1' can not be served by different port on same address")},
 
 		// Erasure Setup with PathEndpointType

cmd/erasure-coding.go

@@ -41,7 +41,7 @@ type Erasure struct {
 // NewErasure creates a new ErasureStorage.
 func NewErasure(ctx context.Context, dataBlocks, parityBlocks int, blockSize int64) (e Erasure, err error) {
 	// Check the parameters for sanity now.
-	if dataBlocks <= 0 || parityBlocks <= 0 {
+	if dataBlocks <= 0 || parityBlocks < 0 {
 		return e, reedsolomon.ErrInvShardNum
 	}
 

cmd/erasure-healing.go

@@ -418,10 +418,19 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 		return result, nil
 	}
 
+	if !latestMeta.XLV1 && !latestMeta.Deleted && disksToHealCount > latestMeta.Erasure.ParityBlocks {
+		// When disk to heal count is greater than parity blocks we should simply error out.
+		err := fmt.Errorf("more disks are expected to heal than parity, returned errors: %v -> %s/%s(%s)", errs, bucket, object, versionID)
+		logger.LogIf(ctx, err)
+		return er.defaultHealResult(latestMeta, storageDisks, storageEndpoints, errs,
+			bucket, object, versionID), err
+	}
+
 	cleanFileInfo := func(fi FileInfo) FileInfo {
-		// Returns a copy of the 'fi' with checksums and parts nil'ed.
+		// Returns a copy of the 'fi' with erasure index, checksums and inline data niled.
 		nfi := fi
-		if !fi.IsRemote() {
+		if !nfi.IsRemote() {
+			nfi.Data = nil
 			nfi.Erasure.Index = 0
 			nfi.Erasure.Checksums = nil
 		}
@@ -432,12 +441,25 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 	tmpID := mustGetUUID()
 	migrateDataDir := mustGetUUID()
 
+	// Reorder so that we have data disks first and parity disks next.
+	latestDisks := shuffleDisks(availableDisks, latestMeta.Erasure.Distribution)
+	outDatedDisks = shuffleDisks(outDatedDisks, latestMeta.Erasure.Distribution)
+	partsMetadata = shufflePartsMetadata(partsMetadata, latestMeta.Erasure.Distribution)
+
 	copyPartsMetadata := make([]FileInfo, len(partsMetadata))
+	for i := range latestDisks {
+		if latestDisks[i] == nil {
+			continue
+		}
+		copyPartsMetadata[i] = partsMetadata[i]
+	}
+
 	for i := range outDatedDisks {
 		if outDatedDisks[i] == nil {
 			continue
 		}
-		copyPartsMetadata[i] = partsMetadata[i]
+		// Make sure to write the FileInfo information
+		// that is expected to be in quorum.
 		partsMetadata[i] = cleanFileInfo(latestMeta)
 	}
 
@@ -456,12 +478,6 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 		inlineBuffers = make([]*bytes.Buffer, len(outDatedDisks))
 	}
 
-	// Reorder so that we have data disks first and parity disks next.
-	latestDisks := shuffleDisks(availableDisks, latestMeta.Erasure.Distribution)
-	outDatedDisks = shuffleDisks(outDatedDisks, latestMeta.Erasure.Distribution)
-	partsMetadata = shufflePartsMetadata(partsMetadata, latestMeta.Erasure.Distribution)
-	copyPartsMetadata = shufflePartsMetadata(copyPartsMetadata, latestMeta.Erasure.Distribution)
-
 	if !latestMeta.Deleted && !latestMeta.IsRemote() {
 		// Heal each part. erasureHealFile() will write the healed
 		// part to .minio/tmp/uuid/ which needs to be renamed later to
@@ -486,7 +502,7 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 				}
 				checksumInfo := copyPartsMetadata[i].Erasure.GetChecksumInfo(partNumber)
 				partPath := pathJoin(object, srcDataDir, fmt.Sprintf("part.%d", partNumber))
-				readers[i] = newBitrotReader(disk, partsMetadata[i].Data, bucket, partPath, tillOffset, checksumAlgo,
+				readers[i] = newBitrotReader(disk, copyPartsMetadata[i].Data, bucket, partPath, tillOffset, checksumAlgo,
 					checksumInfo.Hash, erasure.ShardSize())
 			}
 			writers := make([]io.Writer, len(outDatedDisks))

cmd/erasure-metadata-utils.go

@@ -137,24 +137,26 @@ func readAllFileInfo(ctx context.Context, disks []StorageAPI, bucket, object, ve
 				return errDiskNotFound
 			}
 			metadataArray[index], err = disks[index].ReadVersion(ctx, bucket, object, versionID, readData)
-			if err != nil {
-				if !IsErr(err, []error{
-					errFileNotFound,
-					errVolumeNotFound,
-					errFileVersionNotFound,
-					errDiskNotFound,
-				}...) {
-					logger.LogOnceIf(ctx, fmt.Errorf("Drive %s, path (%s/%s) returned an error (%w)",
-						disks[index], bucket, object, err),
-						disks[index].String())
-				}
-			}
 			return err
 		}, index)
 	}
 
+	errs := g.Wait()
+	for index, err := range errs {
+		if !IsErr(err, []error{
+			errFileNotFound,
+			errVolumeNotFound,
+			errFileVersionNotFound,
+			errDiskNotFound,
+		}...) {
+			logger.LogOnceIf(ctx, fmt.Errorf("Drive %s, path (%s/%s) returned an error (%w)",
+				disks[index], bucket, object, err),
+				disks[index].String())
+		}
+	}
+
 	// Return all the metadata.
-	return metadataArray, g.Wait()
+	return metadataArray, errs
 }
 
 // shuffleDisksAndPartsMetadataByIndex this function should be always used by GetObjectNInfo()
@@ -178,10 +180,6 @@ func shuffleDisksAndPartsMetadataByIndex(disks []StorageAPI, metaArr []FileInfo,
 			inconsistent++
 			continue
 		}
-		if len(fi.Data) != len(meta.Data) {
-			inconsistent++
-			continue
-		}
 		if meta.XLV1 != fi.XLV1 {
 			inconsistent++
 			continue
@@ -229,12 +227,6 @@ func shuffleDisksAndPartsMetadata(disks []StorageAPI, partsMetadata []FileInfo,
 			// if object was ever written previously.
 			continue
 		}
-		if !init && len(fi.Data) != len(partsMetadata[index].Data) {
-			// Check for length of data parts only when
-			// fi.ModTime is not empty - ModTime is always set,
-			// if object was ever written previously.
-			continue
-		}
 		if !init && fi.XLV1 != partsMetadata[index].XLV1 {
 			continue
 		}

cmd/erasure-metadata.go

@@ -19,7 +19,6 @@ package cmd
 
 import (
 	"context"
-	"crypto/sha256"
 	"encoding/hex"
 	"fmt"
 	"net/http"
@@ -28,6 +27,7 @@ import (
 	"time"
 
 	"github.com/minio/minio/internal/bucket/replication"
+	"github.com/minio/minio/internal/hash/sha256"
 	xhttp "github.com/minio/minio/internal/http"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/minio/internal/sync/errgroup"
@@ -99,15 +99,15 @@ func (fi FileInfo) IsValid() bool {
 		fi.Erasure.Index <= dataBlocks+parityBlocks &&
 		len(fi.Erasure.Distribution) == (dataBlocks+parityBlocks))
 	return ((dataBlocks >= parityBlocks) &&
-		(dataBlocks != 0) && (parityBlocks != 0) &&
+		(dataBlocks > 0) && (parityBlocks >= 0) &&
 		correctIndexes)
 }
 
 // ToObjectInfo - Converts metadata to object info.
-func (fi FileInfo) ToObjectInfo(bucket, object string) ObjectInfo {
+func (fi FileInfo) ToObjectInfo(bucket, object string, versioned bool) ObjectInfo {
 	object = decodeDirObject(object)
 	versionID := fi.VersionID
-	if (globalBucketVersioningSys.Enabled(bucket) || globalBucketVersioningSys.Suspended(bucket)) && versionID == "" {
+	if versioned && versionID == "" {
 		versionID = nullVersionID
 	}
 
@@ -284,7 +284,7 @@ func (fi FileInfo) ObjectToPartOffset(ctx context.Context, offset int64) (partIn
 
 func findFileInfoInQuorum(ctx context.Context, metaArr []FileInfo, modTime time.Time, quorum int) (FileInfo, error) {
 	// with less quorum return error.
-	if quorum < 2 {
+	if quorum < 1 {
 		return FileInfo{}, errErasureReadQuorum
 	}
 	metaHashes := make([]string, len(metaArr))
@@ -297,8 +297,6 @@ func findFileInfoInQuorum(ctx context.Context, metaArr []FileInfo, modTime time.
 				fmt.Fprintf(h, "part.%d", part.Number)
 			}
 			fmt.Fprintf(h, "%v", meta.Erasure.Distribution)
-			// make sure that length of Data is same
-			fmt.Fprintf(h, "%v", len(meta.Data))
 
 			// ILM transition fields
 			fmt.Fprint(h, meta.TransitionStatus)
@@ -400,6 +398,10 @@ func writeUniqueFileInfo(ctx context.Context, disks []StorageAPI, bucket, prefix
 // readQuorum is the min required disks to read data.
 // writeQuorum is the min required disks to write data.
 func objectQuorumFromMeta(ctx context.Context, partsMetaData []FileInfo, errs []error, defaultParityCount int) (objectReadQuorum, objectWriteQuorum int, err error) {
+	if defaultParityCount == 0 {
+		return 1, 1, nil
+	}
+
 	// get the latest updated Metadata and a count of all the latest updated FileInfo(s)
 	latestFileInfo, err := getLatestFileInfo(ctx, partsMetaData, errs)
 	if err != nil {

cmd/erasure-multipart.go

@@ -190,6 +190,8 @@ func (er erasureObjects) cleanupStaleUploadsOnDisk(ctx context.Context, disk Sto
 // towards simplification of multipart APIs.
 // The resulting ListMultipartsInfo structure is unmarshalled directly as XML.
 func (er erasureObjects) ListMultipartUploads(ctx context.Context, bucket, object, keyMarker, uploadIDMarker, delimiter string, maxUploads int) (result ListMultipartsInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	result.MaxUploads = maxUploads
 	result.KeyMarker = keyMarker
 	result.Prefix = object
@@ -284,8 +286,10 @@ func (er erasureObjects) ListMultipartUploads(ctx context.Context, bucket, objec
 // disks. `uploads.json` carries metadata regarding on-going multipart
 // operation(s) on the object.
 func (er erasureObjects) newMultipartUpload(ctx context.Context, bucket string, object string, opts ObjectOptions) (string, error) {
+	userDefined := cloneMSS(opts.UserDefined)
+
 	onlineDisks := er.getDisks()
-	parityDrives := globalStorageClass.GetParityForSC(opts.UserDefined[xhttp.AmzStorageClass])
+	parityDrives := globalStorageClass.GetParityForSC(userDefined[xhttp.AmzStorageClass])
 	if parityDrives <= 0 {
 		parityDrives = er.defaultParityCount
 	}
@@ -306,7 +310,7 @@ func (er erasureObjects) newMultipartUpload(ctx context.Context, bucket string,
 		}
 	}
 	if parityOrig != parityDrives {
-		opts.UserDefined[minIOErasureUpgraded] = strconv.Itoa(parityOrig) + "->" + strconv.Itoa(parityDrives)
+		userDefined[minIOErasureUpgraded] = strconv.Itoa(parityOrig) + "->" + strconv.Itoa(parityDrives)
 	}
 
 	dataDrives := len(onlineDisks) - parityDrives
@@ -334,8 +338,8 @@ func (er erasureObjects) newMultipartUpload(ctx context.Context, bucket string,
 	}
 
 	// Guess content-type from the extension if possible.
-	if opts.UserDefined["content-type"] == "" {
-		opts.UserDefined["content-type"] = mimedb.TypeByExtension(path.Ext(object))
+	if userDefined["content-type"] == "" {
+		userDefined["content-type"] = mimedb.TypeByExtension(path.Ext(object))
 	}
 
 	modTime := opts.MTime
@@ -350,7 +354,7 @@ func (er erasureObjects) newMultipartUpload(ctx context.Context, bucket string,
 	for index := range partsMetadata {
 		partsMetadata[index].Fresh = true
 		partsMetadata[index].ModTime = modTime
-		partsMetadata[index].Metadata = opts.UserDefined
+		partsMetadata[index].Metadata = userDefined
 	}
 
 	uploadID := mustGetUUID()
@@ -371,10 +375,8 @@ func (er erasureObjects) newMultipartUpload(ctx context.Context, bucket string,
 //
 // Implements S3 compatible initiate multipart API.
 func (er erasureObjects) NewMultipartUpload(ctx context.Context, bucket, object string, opts ObjectOptions) (string, error) {
-	// No metadata is set, allocate a new one.
-	if opts.UserDefined == nil {
-		opts.UserDefined = make(map[string]string)
-	}
+	auditObjectErasureSet(ctx, object, &er)
+
 	return er.newMultipartUpload(ctx, bucket, object, opts)
 }
 
@@ -446,6 +448,8 @@ func renamePart(ctx context.Context, disks []StorageAPI, srcBucket, srcEntry, ds
 //
 // Implements S3 compatible Upload Part API.
 func (er erasureObjects) PutObjectPart(ctx context.Context, bucket, object, uploadID string, partID int, r *PutObjReader, opts ObjectOptions) (pi PartInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	// Write lock for this part ID.
 	// Held throughout the operation.
 	partIDLock := er.NewNSLock(bucket, pathJoin(object, uploadID, strconv.Itoa(partID)))
@@ -683,6 +687,8 @@ func (er erasureObjects) PutObjectPart(ctx context.Context, bucket, object, uplo
 // - encrypted
 // - compressed
 func (er erasureObjects) GetMultipartInfo(ctx context.Context, bucket, object, uploadID string, opts ObjectOptions) (MultipartInfo, error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	result := MultipartInfo{
 		Bucket:   bucket,
 		Object:   object,
@@ -739,6 +745,8 @@ func (er erasureObjects) GetMultipartInfo(ctx context.Context, bucket, object, u
 // ListPartsInfo structure is marshaled directly into XML and
 // replied back to the client.
 func (er erasureObjects) ListObjectParts(ctx context.Context, bucket, object, uploadID string, partNumberMarker, maxParts int, opts ObjectOptions) (result ListPartsInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	uploadIDLock := er.NewNSLock(bucket, pathJoin(object, uploadID))
 	lkctx, err := uploadIDLock.GetRLock(ctx, globalOperationTimeout)
 	if err != nil {
@@ -832,6 +840,8 @@ func (er erasureObjects) ListObjectParts(ctx context.Context, bucket, object, up
 //
 // Implements S3 compatible Complete multipart API.
 func (er erasureObjects) CompleteMultipartUpload(ctx context.Context, bucket string, object string, uploadID string, parts []CompletePart, opts ObjectOptions) (oi ObjectInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	// Hold read-locks to verify uploaded parts, also disallows
 	// parallel part uploads as well.
 	uploadIDLock := er.NewNSLock(bucket, pathJoin(object, uploadID))
@@ -1017,7 +1027,7 @@ func (er erasureObjects) CompleteMultipartUpload(ctx context.Context, bucket str
 	fi.IsLatest = true
 
 	// Success, return object info.
-	return fi.ToObjectInfo(bucket, object), nil
+	return fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended), nil
 }
 
 // AbortMultipartUpload - aborts an ongoing multipart operation
@@ -1028,6 +1038,8 @@ func (er erasureObjects) CompleteMultipartUpload(ctx context.Context, bucket str
 // would be removed from the system, rollback is not possible on this
 // operation.
 func (er erasureObjects) AbortMultipartUpload(ctx context.Context, bucket, object, uploadID string, opts ObjectOptions) (err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	lk := er.NewNSLock(bucket, pathJoin(object, uploadID))
 	lkctx, err := lk.GetLock(ctx, globalOperationTimeout)
 	if err != nil {

cmd/erasure-object.go

@@ -63,6 +63,8 @@ func countOnlineDisks(onlineDisks []StorageAPI) (online int) {
 // if source object and destination object are same we only
 // update metadata.
 func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, dstBucket, dstObject string, srcInfo ObjectInfo, srcOpts, dstOpts ObjectOptions) (oi ObjectInfo, err error) {
+	auditObjectErasureSet(ctx, dstObject, &er)
+
 	// This call shouldn't be used for anything other than metadata updates or adding self referential versions.
 	if !srcInfo.metadataOnly {
 		return oi, NotImplemented{}
@@ -115,7 +117,7 @@ func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, d
 		if srcOpts.VersionID == "" {
 			return oi, toObjectErr(errFileNotFound, srcBucket, srcObject)
 		}
-		return fi.ToObjectInfo(srcBucket, srcObject), toObjectErr(errMethodNotAllowed, srcBucket, srcObject)
+		return fi.ToObjectInfo(srcBucket, srcObject, srcOpts.Versioned || srcOpts.VersionSuspended), toObjectErr(errMethodNotAllowed, srcBucket, srcObject)
 	}
 
 	filterOnlineDisksInplace(fi, metaArr, onlineDisks)
@@ -131,15 +133,6 @@ func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, d
 		modTime = UTCNow()
 	}
 
-	// If the data is not inlined, we may end up incorrectly
-	// inlining the data here, that leads to an inconsistent
-	// situation where some objects are were not inlined
-	// were now inlined, make sure to `nil` the Data such
-	// that xl.meta is written as expected.
-	if !fi.InlineData() {
-		fi.Data = nil
-	}
-
 	fi.VersionID = versionID // set any new versionID we might have created
 	fi.ModTime = modTime     // set modTime for the new versionID
 	if !dstOpts.MTime.IsZero() {
@@ -149,12 +142,15 @@ func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, d
 	fi.Metadata = srcInfo.UserDefined
 	srcInfo.UserDefined["etag"] = srcInfo.ETag
 
+	inlineData := fi.InlineData()
+	freeVersionID := fi.TierFreeVersionID()
+	freeVersionMarker := fi.TierFreeVersion()
+
 	// Update `xl.meta` content on each disks.
 	for index := range metaArr {
 		if metaArr[index].IsValid() {
 			metaArr[index].ModTime = modTime
 			metaArr[index].VersionID = versionID
-			metaArr[index].Metadata = srcInfo.UserDefined
 			if !metaArr[index].InlineData() {
 				// If the data is not inlined, we may end up incorrectly
 				// inlining the data here, that leads to an inconsistent
@@ -163,6 +159,17 @@ func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, d
 				// that xl.meta is written as expected.
 				metaArr[index].Data = nil
 			}
+			metaArr[index].Metadata = srcInfo.UserDefined
+			// Preserve existing values
+			if inlineData {
+				metaArr[index].SetInlineData()
+			}
+			if freeVersionID != "" {
+				metaArr[index].SetTierFreeVersionID(freeVersionID)
+			}
+			if freeVersionMarker {
+				metaArr[index].SetTierFreeVersion()
+			}
 		}
 	}
 
@@ -171,12 +178,14 @@ func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, d
 		return oi, toObjectErr(err, srcBucket, srcObject)
 	}
 
-	return fi.ToObjectInfo(srcBucket, srcObject), nil
+	return fi.ToObjectInfo(srcBucket, srcObject, srcOpts.Versioned || srcOpts.VersionSuspended), nil
 }
 
 // GetObjectNInfo - returns object info and an object
 // Read(Closer). When err != nil, the returned reader is always nil.
 func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, lockType LockType, opts ObjectOptions) (gr *GetObjectReader, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	var unlockOnDefer bool
 	nsUnlocker := func() {}
 	defer func() {
@@ -236,7 +245,7 @@ func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object stri
 		}
 	}
 
-	objInfo := fi.ToObjectInfo(bucket, object)
+	objInfo := fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
 	if objInfo.DeleteMarker {
 		if opts.VersionID == "" {
 			return &GetObjectReader{
@@ -413,6 +422,8 @@ func (er erasureObjects) getObjectWithFileInfo(ctx context.Context, bucket, obje
 
 // GetObjectInfo - reads object metadata and replies back ObjectInfo.
 func (er erasureObjects) GetObjectInfo(ctx context.Context, bucket, object string, opts ObjectOptions) (info ObjectInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	if !opts.NoLock {
 		// Lock the object before reading.
 		lk := er.NewNSLock(bucket, object)
@@ -465,16 +476,6 @@ func readAllXL(ctx context.Context, disks []StorageAPI, bucket, object string, r
 			}
 			rf, err := disks[index].ReadXL(ctx, bucket, object, readData)
 			if err != nil {
-				if !IsErr(err, []error{
-					errFileNotFound,
-					errVolumeNotFound,
-					errFileVersionNotFound,
-					errDiskNotFound,
-				}...) {
-					logger.LogOnceIf(ctx, fmt.Errorf("Drive %s, path (%s/%s) returned an error (%w)",
-						disks[index], bucket, object, err),
-						disks[index].String())
-				}
 				return err
 			}
 
@@ -491,6 +492,19 @@ func readAllXL(ctx context.Context, disks []StorageAPI, bucket, object string, r
 	}
 
 	errs := g.Wait()
+	for index, err := range errs {
+		if !IsErr(err, []error{
+			errFileNotFound,
+			errVolumeNotFound,
+			errFileVersionNotFound,
+			errDiskNotFound,
+		}...) {
+			logger.LogOnceIf(ctx, fmt.Errorf("Drive %s, path (%s/%s) returned an error (%w)",
+				disks[index], bucket, object, err),
+				disks[index].String())
+		}
+	}
+
 	for index := range metadataArray {
 		if metadataArray[index] != nil {
 			metadataShallowVersions[index] = metadataArray[index].versions
@@ -498,23 +512,45 @@ func readAllXL(ctx context.Context, disks []StorageAPI, bucket, object string, r
 	}
 
 	readQuorum := (len(disks) + 1) / 2
-	merged := mergeXLV2Versions(readQuorum, false, 1, metadataShallowVersions...)
+	meta := &xlMetaV2{versions: mergeXLV2Versions(readQuorum, false, 1, metadataShallowVersions...)}
+	lfi, err := meta.ToFileInfo(bucket, object, "")
+	if err != nil {
+		for i := range errs {
+			if errs[i] == nil {
+				errs[i] = err
+			}
+		}
+		return metaFileInfos, errs
+	}
+	if !lfi.IsValid() {
+		for i := range errs {
+			if errs[i] == nil {
+				errs[i] = errCorruptedFormat
+			}
+		}
+		return metaFileInfos, errs
+	}
+
 	for index := range metadataArray {
 		if metadataArray[index] == nil {
 			continue
 		}
 
-		metadataArray[index].versions = merged
-
 		// make sure to preserve this for diskmtime based healing bugfix.
 		diskMTime := metaFileInfos[index].DiskMTime
 		metaFileInfos[index], errs[index] = metadataArray[index].ToFileInfo(bucket, object, "")
-		if errs[index] == nil {
+		if errs[index] != nil {
+			continue
+		}
+
+		if metaFileInfos[index].IsValid() && metaFileInfos[index].ModTime.Equal(lfi.ModTime) {
 			versionID := metaFileInfos[index].VersionID
 			if versionID == "" {
 				versionID = nullVersionID
 			}
-			metaFileInfos[index].Data = metadataArray[index].data.find(versionID)
+			if readData {
+				metaFileInfos[index].Data = metadataArray[index].data.find(versionID)
+			}
 			metaFileInfos[index].DiskMTime = diskMTime
 		}
 	}
@@ -582,6 +618,8 @@ func (er erasureObjects) getObjectFileInfo(ctx context.Context, bucket, object s
 		if metaArr[i].IsValid() && metaArr[i].ModTime.Equal(fi.ModTime) {
 			continue
 		}
+		metaArr[i] = FileInfo{}
+		onlineDisks[i] = nil
 		missingBlocks++
 	}
 
@@ -603,7 +641,7 @@ func (er erasureObjects) getObjectInfo(ctx context.Context, bucket, object strin
 	if err != nil {
 		return objInfo, toObjectErr(err, bucket, object)
 	}
-	objInfo = fi.ToObjectInfo(bucket, object)
+	objInfo = fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
 	if fi.Deleted {
 		if opts.VersionID == "" || opts.DeleteMarker {
 			return objInfo, toObjectErr(errFileNotFound, bucket, object)
@@ -627,7 +665,7 @@ func (er erasureObjects) getObjectInfoAndQuorum(ctx context.Context, bucket, obj
 		wquorum++
 	}
 
-	objInfo = fi.ToObjectInfo(bucket, object)
+	objInfo = fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
 	if !fi.VersionPurgeStatus().Empty() && opts.VersionID != "" {
 		// Make sure to return object info to provide extra information.
 		return objInfo, wquorum, toObjectErr(errMethodNotAllowed, bucket, object)
@@ -819,7 +857,7 @@ func (er erasureObjects) putMetacacheObject(ctx context.Context, key string, r *
 		return ObjectInfo{}, toObjectErr(err, minioMetaBucket, key)
 	}
 
-	return fi.ToObjectInfo(minioMetaBucket, key), nil
+	return fi.ToObjectInfo(minioMetaBucket, key, opts.Versioned || opts.VersionSuspended), nil
 }
 
 // PutObject - creates an object upon reading from the input stream
@@ -832,19 +870,18 @@ func (er erasureObjects) PutObject(ctx context.Context, bucket string, object st
 
 // putObject wrapper for erasureObjects PutObject
 func (er erasureObjects) putObject(ctx context.Context, bucket string, object string, r *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	data := r.Reader
 
-	// No metadata is set, allocate a new one.
-	if opts.UserDefined == nil {
-		opts.UserDefined = make(map[string]string)
-	}
+	userDefined := cloneMSS(opts.UserDefined)
 
 	storageDisks := er.getDisks()
 
 	parityDrives := len(storageDisks) / 2
 	if !opts.MaxParity {
 		// Get parity and data drive count based on storage class metadata
-		parityDrives = globalStorageClass.GetParityForSC(opts.UserDefined[xhttp.AmzStorageClass])
+		parityDrives = globalStorageClass.GetParityForSC(userDefined[xhttp.AmzStorageClass])
 		if parityDrives <= 0 {
 			parityDrives = er.defaultParityCount
 		}
@@ -882,7 +919,7 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 			parityDrives = len(storageDisks) / 2
 		}
 		if parityOrig != parityDrives {
-			opts.UserDefined[minIOErasureUpgraded] = strconv.Itoa(parityOrig) + "->" + strconv.Itoa(parityDrives)
+			userDefined[minIOErasureUpgraded] = strconv.Itoa(parityOrig) + "->" + strconv.Itoa(parityDrives)
 		}
 	}
 	dataDrives := len(storageDisks) - parityDrives
@@ -1058,13 +1095,13 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 			Hash:       bitrotWriterSum(w),
 		})
 	}
-	if opts.UserDefined["etag"] == "" {
-		opts.UserDefined["etag"] = r.MD5CurrentHexString()
+	if userDefined["etag"] == "" {
+		userDefined["etag"] = r.MD5CurrentHexString()
 	}
 
 	// Guess content-type from the extension if possible.
-	if opts.UserDefined["content-type"] == "" {
-		opts.UserDefined["content-type"] = mimedb.TypeByExtension(path.Ext(object))
+	if userDefined["content-type"] == "" {
+		userDefined["content-type"] = mimedb.TypeByExtension(path.Ext(object))
 	}
 
 	modTime := opts.MTime
@@ -1075,7 +1112,7 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 	// Fill all the necessary metadata.
 	// Update `xl.meta` content on each disks.
 	for index := range partsMetadata {
-		partsMetadata[index].Metadata = opts.UserDefined
+		partsMetadata[index].Metadata = userDefined
 		partsMetadata[index].Size = n
 		partsMetadata[index].ModTime = modTime
 	}
@@ -1125,7 +1162,7 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 	// we are adding a new version to this object under the namespace lock, so this is the latest version.
 	fi.IsLatest = true
 
-	return fi.ToObjectInfo(bucket, object), nil
+	return fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended), nil
 }
 
 func (er erasureObjects) deleteObjectVersion(ctx context.Context, bucket, object string, writeQuorum int, fi FileInfo, forceDelMarker bool) error {
@@ -1148,6 +1185,10 @@ func (er erasureObjects) deleteObjectVersion(ctx context.Context, bucket, object
 // into smaller bulks if some object names are found to be duplicated in the delete list, splitting
 // into smaller bulks will avoid holding twice the write lock of the duplicated object names.
 func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objects []ObjectToDelete, opts ObjectOptions) ([]DeletedObject, []error) {
+	for _, obj := range objects {
+		auditObjectErasureSet(ctx, obj.ObjectV.ObjectName, &er)
+	}
+
 	errs := make([]error, len(objects))
 	dobjects := make([]DeletedObject, len(objects))
 	writeQuorums := make([]int, len(objects))
@@ -1178,7 +1219,13 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 		// VersionID is not set means delete is not specific about
 		// any version, look for if the bucket is versioned or not.
 		if objects[i].VersionID == "" {
-			if opts.Versioned || opts.VersionSuspended {
+			// MinIO extension to bucket version configuration
+			suspended := opts.VersionSuspended
+			versioned := opts.Versioned
+			if opts.PrefixEnabledFn != nil {
+				versioned = opts.PrefixEnabledFn(objects[i].ObjectName)
+			}
+			if versioned || suspended {
 				// Bucket is versioned and no version was explicitly
 				// mentioned for deletes, create a delete marker instead.
 				vr.ModTime = UTCNow()
@@ -1186,7 +1233,7 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 				// Versioning suspended means that we add a `null` version
 				// delete marker, if not add a new version for this delete
 				// marker.
-				if opts.Versioned {
+				if versioned {
 					vr.VersionID = mustGetUUID()
 				}
 			}
@@ -1307,12 +1354,17 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 func (er erasureObjects) deletePrefix(ctx context.Context, bucket, prefix string) error {
 	disks := er.getDisks()
 	g := errgroup.WithNErrs(len(disks))
+	dirPrefix := encodeDirObject(prefix)
 	for index := range disks {
 		index := index
 		g.Go(func() error {
 			if disks[index] == nil {
 				return nil
 			}
+			// Deletes
+			// - The prefix and its children
+			// - The prefix__XLDIR__
+			defer disks[index].Delete(ctx, bucket, dirPrefix, true)
 			return disks[index].Delete(ctx, bucket, prefix, true)
 		}, index)
 	}
@@ -1328,6 +1380,8 @@ func (er erasureObjects) deletePrefix(ctx context.Context, bucket, prefix string
 // any error as it is not necessary for the handler to reply back a
 // response to the client request.
 func (er erasureObjects) DeleteObject(ctx context.Context, bucket, object string, opts ObjectOptions) (objInfo ObjectInfo, err error) {
+	auditObjectErasureSet(ctx, object, &er)
+
 	if opts.DeletePrefix {
 		return ObjectInfo{}, toObjectErr(er.deletePrefix(ctx, bucket, object), bucket, object)
 	}
@@ -1482,7 +1536,7 @@ func (er erasureObjects) DeleteObject(ctx context.Context, bucket, object string
 			if err = er.deleteObjectVersion(ctx, bucket, object, writeQuorum, fi, opts.DeleteMarker); err != nil {
 				return objInfo, toObjectErr(err, bucket, object)
 			}
-			return fi.ToObjectInfo(bucket, object), nil
+			return fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended), nil
 		}
 	}
 
@@ -1587,7 +1641,7 @@ func (er erasureObjects) PutObjectMetadata(ctx context.Context, bucket, object s
 		opts.VersionID = fi.VersionID
 	}
 
-	objInfo := fi.ToObjectInfo(bucket, object)
+	objInfo := fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
 	if opts.EvalMetadataFn != nil {
 		if err := opts.EvalMetadataFn(objInfo); err != nil {
 			return ObjectInfo{}, err
@@ -1603,7 +1657,7 @@ func (er erasureObjects) PutObjectMetadata(ctx context.Context, bucket, object s
 		return ObjectInfo{}, toObjectErr(err, bucket, object)
 	}
 
-	return fi.ToObjectInfo(bucket, object), nil
+	return fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended), nil
 }
 
 // PutObjectTags - replace or add tags to an existing object
@@ -1667,7 +1721,7 @@ func (er erasureObjects) PutObjectTags(ctx context.Context, bucket, object strin
 		return ObjectInfo{}, toObjectErr(err, bucket, object)
 	}
 
-	return fi.ToObjectInfo(bucket, object), nil
+	return fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended), nil
 }
 
 // updateObjectMeta will update the metadata of a file.
@@ -1805,7 +1859,7 @@ func (er erasureObjects) TransitionObject(ctx context.Context, bucket, object st
 		break
 	}
 
-	objInfo := fi.ToObjectInfo(bucket, object)
+	objInfo := fi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
 	sendEvent(eventArgs{
 		EventName:  eventName,
 		BucketName: bucket,
@@ -1860,7 +1914,7 @@ func (er erasureObjects) restoreTransitionedObject(ctx context.Context, bucket s
 		return setRestoreHeaderFn(oi, toObjectErr(err, bucket, object))
 	}
 
-	oi = actualfi.ToObjectInfo(bucket, object)
+	oi = actualfi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
 	ropts := putRestoreOpts(bucket, object, opts.Transition.RestoreRequest, oi)
 	if len(oi.Parts) == 1 {
 		var rs *HTTPRangeSpec

cmd/erasure-object_test.go

@@ -20,11 +20,14 @@ package cmd
 import (
 	"bytes"
 	"context"
+	"crypto/md5"
 	crand "crypto/rand"
 	"errors"
+	"fmt"
 	"io"
 	"io/ioutil"
 	"os"
+	"path/filepath"
 	"strconv"
 	"testing"
 
@@ -1048,3 +1051,157 @@ func testObjectQuorumFromMeta(obj ObjectLayer, instanceType string, dirs []strin
 		})
 	}
 }
+
+// In some deployments, one object has data inlined in one disk and not inlined in other disks.
+func TestGetObjectInlineNotInline(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// Create a backend with 4 disks named disk{1...4}, this name convention
+	// because we will unzip some object data from a sample archive.
+	const numDisks = 4
+	path, err := ioutil.TempDir(globalTestTmpDir, "minio-")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var fsDirs []string
+	for i := 1; i <= numDisks; i++ {
+		fsDirs = append(fsDirs, filepath.Join(path, fmt.Sprintf("disk%d", i)))
+	}
+
+	objLayer, _, err := initObjectLayer(ctx, mustGetPoolEndpoints(fsDirs...))
+	if err != nil {
+		removeRoots(fsDirs)
+		t.Fatal(err)
+	}
+
+	// cleaning up of temporary test directories
+	defer objLayer.Shutdown(context.Background())
+	defer removeRoots(fsDirs)
+
+	// Create a testbucket
+	err = objLayer.MakeBucketWithLocation(ctx, "testbucket", BucketOptions{})
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Unzip sample object data to the existing disks
+	err = unzipArchive("testdata/xl-meta-inline-notinline.zip", path)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Try to read the object and check its md5sum
+	gr, err := objLayer.GetObjectNInfo(ctx, "testbucket", "file", nil, nil, readLock, ObjectOptions{})
+	if err != nil {
+		t.Fatalf("Expected GetObject to succeed, but failed with %v", err)
+	}
+
+	h := md5.New()
+	_, err = io.Copy(h, gr)
+	if err != nil {
+		t.Fatalf("Expected GetObject reading data to succeed, but failed with %v", err)
+	}
+	gr.Close()
+
+	const expectedHash = "fffb6377948ebea75ad2b8058e849ef5"
+	foundHash := fmt.Sprintf("%x", h.Sum(nil))
+	if foundHash != expectedHash {
+		t.Fatalf("Expected data to have md5sum = `%s`, found `%s`", expectedHash, foundHash)
+	}
+}
+
+// Test reading an object with some outdated data in some disks
+func TestGetObjectWithOutdatedDisks(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// Create an instance of xl backend.
+	obj, fsDirs, err := prepareErasure(ctx, 6)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Cleanup backend directories.
+	defer obj.Shutdown(context.Background())
+	defer removeRoots(fsDirs)
+
+	z := obj.(*erasureServerPools)
+	sets := z.serverPools[0]
+	xl := sets.sets[0]
+
+	origErasureDisks := xl.getDisks()
+
+	testCases := []struct {
+		bucket    string
+		versioned bool
+		object    string
+		content   []byte
+	}{
+		{"bucket1", false, "object1", []byte("aaaaaaaaaaaaaaaa")},
+		{"bucket2", false, "object2", bytes.Repeat([]byte{'a'}, smallFileThreshold*2)},
+		{"bucket3", true, "version1", []byte("aaaaaaaaaaaaaaaa")},
+		{"bucket4", true, "version2", bytes.Repeat([]byte{'a'}, smallFileThreshold*2)},
+	}
+
+	for i, testCase := range testCases {
+		// Step 1: create a bucket
+		err = z.MakeBucketWithLocation(ctx, testCase.bucket, BucketOptions{VersioningEnabled: testCase.versioned})
+		if err != nil {
+			t.Fatalf("Test %d: Failed to create a bucket: %v", i+1, err)
+		}
+
+		// Step 2: Upload an object with a random content
+		initialData := bytes.Repeat([]byte{'b'}, len(testCase.content))
+		sets.erasureDisksMu.Lock()
+		xl.getDisks = func() []StorageAPI { return origErasureDisks }
+		sets.erasureDisksMu.Unlock()
+		_, err = z.PutObject(ctx, testCase.bucket, testCase.object, mustGetPutObjReader(t, bytes.NewReader(initialData), int64(len(initialData)), "", ""),
+			ObjectOptions{Versioned: testCase.versioned})
+		if err != nil {
+			t.Fatalf("Test %d: Failed to upload a random object: %v", i+1, err)
+		}
+
+		// Step 3: Upload the object with some disks offline
+		sets.erasureDisksMu.Lock()
+		xl.getDisks = func() []StorageAPI {
+			disks := make([]StorageAPI, len(origErasureDisks))
+			copy(disks, origErasureDisks)
+			disks[0] = nil
+			disks[1] = nil
+			return disks
+		}
+		sets.erasureDisksMu.Unlock()
+		_, err = z.PutObject(ctx, testCase.bucket, testCase.object, mustGetPutObjReader(t, bytes.NewReader(testCase.content), int64(len(testCase.content)), "", ""),
+			ObjectOptions{Versioned: testCase.versioned})
+		if err != nil {
+			t.Fatalf("Test %d: Failed to upload the final object: %v", i+1, err)
+		}
+
+		// Step 4: Try to read the object back and check its md5sum
+		sets.erasureDisksMu.Lock()
+		xl.getDisks = func() []StorageAPI { return origErasureDisks }
+		sets.erasureDisksMu.Unlock()
+		gr, err := z.GetObjectNInfo(ctx, testCase.bucket, testCase.object, nil, nil, readLock, ObjectOptions{})
+		if err != nil {
+			t.Fatalf("Expected GetObject to succeed, but failed with %v", err)
+		}
+
+		h := md5.New()
+		h.Write(testCase.content)
+		expectedHash := h.Sum(nil)
+
+		h.Reset()
+		_, err = io.Copy(h, gr)
+		if err != nil {
+			t.Fatalf("Test %d: Failed to calculate md5sum of the object: %v", i+1, err)
+		}
+		gr.Close()
+		foundHash := h.Sum(nil)
+
+		if !bytes.Equal(foundHash, expectedHash) {
+			t.Fatalf("Test %d: Expected data to have md5sum = `%x`, found `%x`", i+1, expectedHash, foundHash)
+		}
+	}
+}

cmd/erasure-server-pool-decom.go

@@ -242,39 +242,24 @@ func (p *poolMeta) Decommission(idx int, pi poolSpaceInfo) error {
 		}
 	}
 
+	// Return an error when there is decommission on going - the user needs
+	// to explicitly cancel it first in order to restart decommissioning again.
+	if p.Pools[idx].Decommission != nil &&
+		!p.Pools[idx].Decommission.Complete &&
+		!p.Pools[idx].Decommission.Failed &&
+		!p.Pools[idx].Decommission.Canceled {
+		return errDecommissionAlreadyRunning
+	}
+
 	now := UTCNow()
-	if p.Pools[idx].Decommission == nil {
-		p.Pools[idx].LastUpdate = now
-		p.Pools[idx].Decommission = &PoolDecommissionInfo{
-			StartTime:   now,
-			StartSize:   pi.Free,
-			CurrentSize: pi.Free,
-			TotalSize:   pi.Total,
-		}
-		return nil
+	p.Pools[idx].LastUpdate = now
+	p.Pools[idx].Decommission = &PoolDecommissionInfo{
+		StartTime:   now,
+		StartSize:   pi.Free,
+		CurrentSize: pi.Free,
+		TotalSize:   pi.Total,
 	}
-
-	// Completed pool doesn't need to be decommissioned again.
-	if p.Pools[idx].Decommission.Complete {
-		return errDecommissionComplete
-	}
-
-	// Canceled or Failed decommission can be triggered again.
-	if p.Pools[idx].Decommission.StartTime.IsZero() {
-		if p.Pools[idx].Decommission.Canceled || p.Pools[idx].Decommission.Failed {
-			p.Pools[idx].LastUpdate = now
-			p.Pools[idx].Decommission = &PoolDecommissionInfo{
-				StartTime:   now,
-				StartSize:   pi.Free,
-				CurrentSize: pi.Free,
-				TotalSize:   pi.Total,
-			}
-			return nil
-		}
-	} // In-progress pool doesn't need to be decommissioned again.
-
-	// In all other scenarios an active decommissioning is in progress.
-	return errDecommissionAlreadyRunning
+	return nil
 }
 
 func (p poolMeta) IsSuspended(idx int) bool {
@@ -556,8 +541,13 @@ func (z *erasureServerPools) Init(ctx context.Context) error {
 }
 
 func (z *erasureServerPools) decommissionObject(ctx context.Context, bucket string, gr *GetObjectReader) (err error) {
-	defer gr.Close()
 	objInfo := gr.ObjInfo
+
+	defer func() {
+		gr.Close()
+		auditLogDecom(ctx, "DecomCopyData", objInfo.Bucket, objInfo.Name, objInfo.VersionID, err)
+	}()
+
 	if objInfo.isMultipart() {
 		uploadID, err := z.NewMultipartUpload(ctx, bucket, objInfo.Name, ObjectOptions{
 			VersionID:   objInfo.VersionID,
@@ -618,6 +608,8 @@ func (v versionsSorter) reverse() {
 }
 
 func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool *erasureSets, bName string) error {
+	ctx = logger.SetReqInfo(ctx, &logger.ReqInfo{})
+
 	var wg sync.WaitGroup
 	wStr := env.Get("_MINIO_DECOMMISSION_WORKERS", strconv.Itoa(len(pool.sets)))
 	workerSize, err := strconv.Atoi(wStr)
@@ -627,7 +619,6 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 
 	parallelWorkers := make(chan struct{}, workerSize)
 
-	versioned := globalBucketVersioningSys.Enabled(bName)
 	for _, set := range pool.sets {
 		set := set
 		disks := set.getOnlineDisks()
@@ -637,6 +628,7 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 			continue
 		}
 
+		vc, _ := globalBucketVersioningSys.Get(bName)
 		decommissionEntry := func(entry metaCacheEntry) {
 			defer func() {
 				<-parallelWorkers
@@ -675,7 +667,7 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 						bName,
 						version.Name,
 						ObjectOptions{
-							Versioned:         versioned,
+							Versioned:         vc.PrefixEnabled(version.Name),
 							VersionID:         version.VersionID,
 							MTime:             version.ModTime,
 							DeleteReplication: version.ReplicationState,
@@ -695,27 +687,34 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 					continue
 				}
 
-				gr, err := set.GetObjectNInfo(ctx,
-					bName,
-					encodeDirObject(version.Name),
-					nil,
-					http.Header{},
-					noLock, // all mutations are blocked reads are safe without locks.
-					ObjectOptions{
-						VersionID: version.VersionID,
-					})
-				if err != nil {
-					logger.LogIf(ctx, err)
-					z.poolMetaMutex.Lock()
-					z.poolMeta.CountItem(idx, version.Size, true)
-					z.poolMetaMutex.Unlock()
-					break // break out on first error
-				}
 				var failure bool
 				// gr.Close() is ensured by decommissionObject().
-				if err = z.decommissionObject(ctx, bName, gr); err != nil {
-					logger.LogIf(ctx, err)
-					failure = true
+				for try := 0; try < 3; try++ {
+					gr, err := set.GetObjectNInfo(ctx,
+						bName,
+						encodeDirObject(version.Name),
+						nil,
+						http.Header{},
+						noLock, // all mutations are blocked reads are safe without locks.
+						ObjectOptions{
+							VersionID: version.VersionID,
+						})
+					if isErrObjectNotFound(err) {
+						// object deleted by the application, nothing to do here we move on.
+						return
+					}
+					if err != nil {
+						failure = true
+						logger.LogIf(ctx, err)
+						continue
+					}
+					if err = z.decommissionObject(ctx, bName, gr); err != nil {
+						failure = true
+						logger.LogIf(ctx, err)
+						continue
+					}
+					failure = false
+					break
 				}
 				z.poolMetaMutex.Lock()
 				z.poolMeta.CountItem(idx, version.Size, failure)
@@ -728,13 +727,17 @@ func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool
 
 			// if all versions were decommissioned, then we can delete the object versions.
 			if decommissionedCount == len(fivs.Versions) {
-				set.DeleteObject(ctx,
+				_, err := set.DeleteObject(ctx,
 					bName,
-					entry.name,
+					encodeDirObject(entry.name),
 					ObjectOptions{
 						DeletePrefix: true, // use prefix delete to delete all versions at once.
 					},
 				)
+				auditLogDecom(ctx, "DecomDeleteObject", bName, entry.name, "", err)
+				if err != nil {
+					logger.LogIf(ctx, err)
+				}
 			}
 			z.poolMetaMutex.Lock()
 			z.poolMeta.TrackCurrentBucketObject(idx, bName, entry.name)
@@ -819,6 +822,9 @@ func (z *erasureServerPools) doDecommissionInRoutine(ctx context.Context, idx in
 	dctx, z.decommissionCancelers[idx] = context.WithCancel(GlobalContext)
 	z.poolMetaMutex.Unlock()
 
+	// Generate an empty request info so it can be directly modified later by audit
+	dctx = logger.SetReqInfo(dctx, &logger.ReqInfo{})
+
 	if err := z.decommissionInBackground(dctx, idx); err != nil {
 		logger.LogIf(GlobalContext, err)
 		logger.LogIf(GlobalContext, z.DecommissionFailed(dctx, idx))
@@ -1090,3 +1096,16 @@ func (z *erasureServerPools) StartDecommission(ctx context.Context, idx int) (er
 	globalNotificationSys.ReloadPoolMeta(ctx)
 	return nil
 }
+
+func auditLogDecom(ctx context.Context, apiName, bucket, object, versionID string, err error) {
+	errStr := ""
+	if err != nil {
+		errStr = err.Error()
+	}
+	auditLogInternal(ctx, bucket, object, AuditLogOptions{
+		Trigger:   "decommissioning",
+		APIName:   apiName,
+		VersionID: versionID,
+		Error:     errStr,
+	})
+}

cmd/erasure-server-pool.go

@@ -61,6 +61,22 @@ func (z *erasureServerPools) SinglePool() bool {
 
 // Initialize new pool of erasure sets.
 func newErasureServerPools(ctx context.Context, endpointServerPools EndpointServerPools) (ObjectLayer, error) {
+	if endpointServerPools.NEndpoints() == 1 {
+		ep := endpointServerPools[0]
+		storageDisks, format, err := waitForFormatErasure(true, ep.Endpoints, 1, ep.SetCount, ep.DrivesPerSet, "", "")
+		if err != nil {
+			return nil, err
+		}
+
+		objLayer, err := newErasureSingle(ctx, storageDisks[0], format)
+		if err != nil {
+			return nil, err
+		}
+
+		globalLocalDrives = storageDisks
+		return objLayer, nil
+	}
+
 	var (
 		deploymentID       string
 		distributionAlgo   string
@@ -233,8 +249,9 @@ func (z *erasureServerPools) SetDriveCounts() []int {
 type serverPoolsAvailableSpace []poolAvailableSpace
 
 type poolAvailableSpace struct {
-	Index     int
-	Available uint64
+	Index      int
+	Available  uint64
+	MaxUsedPct int // Used disk percentage of most filled disk, rounded down.
 }
 
 // TotalAvailable - total available space
@@ -246,10 +263,41 @@ func (p serverPoolsAvailableSpace) TotalAvailable() uint64 {
 	return total
 }
 
+// FilterMaxUsed will filter out any pools that has used percent bigger than max,
+// unless all have that, in which case all are preserved.
+func (p serverPoolsAvailableSpace) FilterMaxUsed(max int) {
+	// We aren't modifying p, only entries in it, so we don't need to receive a pointer.
+	if len(p) <= 1 {
+		// Nothing to do.
+		return
+	}
+	var ok bool
+	for _, z := range p {
+		if z.MaxUsedPct < max {
+			ok = true
+			break
+		}
+	}
+	if !ok {
+		// All above limit.
+		// Do not modify
+		return
+	}
+
+	// Remove entries that are above.
+	for i, z := range p {
+		if z.MaxUsedPct < max {
+			continue
+		}
+		p[i].Available = 0
+	}
+}
+
 // getAvailablePoolIdx will return an index that can hold size bytes.
 // -1 is returned if no serverPools have available space for the size given.
 func (z *erasureServerPools) getAvailablePoolIdx(ctx context.Context, bucket, object string, size int64) int {
 	serverPools := z.getServerPoolsAvailableSpace(ctx, bucket, object, size)
+	serverPools.FilterMaxUsed(100 - (100 * diskReserveFraction))
 	total := serverPools.TotalAvailable()
 	if total == 0 {
 		return -1
@@ -270,11 +318,13 @@ func (z *erasureServerPools) getAvailablePoolIdx(ctx context.Context, bucket, ob
 
 // getServerPoolsAvailableSpace will return the available space of each pool after storing the content.
 // If there is not enough space the pool will return 0 bytes available.
+// The size of each will be multiplied by the number of sets.
 // Negative sizes are seen as 0 bytes.
 func (z *erasureServerPools) getServerPoolsAvailableSpace(ctx context.Context, bucket, object string, size int64) serverPoolsAvailableSpace {
 	serverPools := make(serverPoolsAvailableSpace, len(z.serverPools))
 
 	storageInfos := make([][]*DiskInfo, len(z.serverPools))
+	nSets := make([]int, len(z.serverPools))
 	g := errgroup.WithNErrs(len(z.serverPools))
 	for index := range z.serverPools {
 		index := index
@@ -282,9 +332,11 @@ func (z *erasureServerPools) getServerPoolsAvailableSpace(ctx context.Context, b
 		if z.IsSuspended(index) {
 			continue
 		}
+		pool := z.serverPools[index]
+		nSets[index] = pool.setCount
 		g.Go(func() error {
 			// Get the set where it would be placed.
-			storageInfos[index] = getDiskInfos(ctx, z.serverPools[index].getHashedSet(object).getDisks())
+			storageInfos[index] = getDiskInfos(ctx, pool.getHashedSet(object).getDisks()...)
 			return nil
 		}, index)
 	}
@@ -298,15 +350,30 @@ func (z *erasureServerPools) getServerPoolsAvailableSpace(ctx context.Context, b
 			serverPools[i] = poolAvailableSpace{Index: i}
 			continue
 		}
+		var maxUsedPct int
 		for _, disk := range zinfo {
-			if disk == nil {
+			if disk == nil || disk.Total == 0 {
 				continue
 			}
 			available += disk.Total - disk.Used
+
+			// set maxUsedPct to the value from the disk with the least space percentage.
+			if pctUsed := int(disk.Used * 100 / disk.Total); pctUsed > maxUsedPct {
+				maxUsedPct = pctUsed
+			}
 		}
+
+		// Since we are comparing pools that may have a different number of sets
+		// we multiply by the number of sets in the pool.
+		// This will compensate for differences in set sizes
+		// when choosing destination pool.
+		// Different set sizes are already compensated by less disks.
+		available *= uint64(nSets[i])
+
 		serverPools[i] = poolAvailableSpace{
-			Index:     i,
-			Available: available,
+			Index:      i,
+			Available:  available,
+			MaxUsedPct: maxUsedPct,
 		}
 	}
 	return serverPools
@@ -774,7 +841,16 @@ func (z *erasureServerPools) GetObjectNInfo(ctx context.Context, bucket, object
 	}
 
 	lockType = noLock // do not take locks at lower levels for GetObjectNInfo()
-	return z.serverPools[zIdx].GetObjectNInfo(ctx, bucket, object, rs, h, lockType, opts)
+	gr, err = z.serverPools[zIdx].GetObjectNInfo(ctx, bucket, object, rs, h, lockType, opts)
+	if err != nil {
+		return nil, err
+	}
+
+	if unlockOnDefer {
+		unlockOnDefer = false
+		return gr.WithCleanupFuncs(nsUnlocker), nil
+	}
+	return gr, nil
 }
 
 // getLatestObjectInfoWithIdx returns the objectInfo of the latest object from multiple pools (this function
@@ -873,7 +949,7 @@ func (z *erasureServerPools) PutObject(ctx context.Context, bucket string, objec
 	object = encodeDirObject(object)
 
 	if z.SinglePool() {
-		if !isMinioMetaBucketName(bucket) && !hasSpaceFor(getDiskInfos(ctx, z.serverPools[0].getHashedSet(object).getDisks()), data.Size()) {
+		if !isMinioMetaBucketName(bucket) && !hasSpaceFor(getDiskInfos(ctx, z.serverPools[0].getHashedSet(object).getDisks()...), data.Size()) {
 			return ObjectInfo{}, toObjectErr(errDiskFull)
 		}
 		return z.serverPools[0].PutObject(ctx, bucket, object, data, opts)
@@ -1265,7 +1341,7 @@ func (z *erasureServerPools) NewMultipartUpload(ctx context.Context, bucket, obj
 	}
 
 	if z.SinglePool() {
-		if !isMinioMetaBucketName(bucket) && !hasSpaceFor(getDiskInfos(ctx, z.serverPools[0].getHashedSet(object).getDisks()), -1) {
+		if !isMinioMetaBucketName(bucket) && !hasSpaceFor(getDiskInfos(ctx, z.serverPools[0].getHashedSet(object).getDisks()...), -1) {
 			return "", toObjectErr(errDiskFull)
 		}
 		return z.serverPools[0].NewMultipartUpload(ctx, bucket, object, opts)
@@ -1675,6 +1751,8 @@ func (z *erasureServerPools) Walk(ctx context.Context, bucket, prefix string, re
 		return err
 	}
 
+	vcfg, _ := globalBucketVersioningSys.Get(bucket)
+
 	ctx, cancel := context.WithCancel(ctx)
 	go func() {
 		defer cancel()
@@ -1707,12 +1785,15 @@ func (z *erasureServerPools) Walk(ctx context.Context, bucket, prefix string, re
 						if opts.WalkAscending {
 							for i := len(fivs.Versions) - 1; i >= 0; i-- {
 								version := fivs.Versions[i]
-								results <- version.ToObjectInfo(bucket, version.Name)
+								versioned := vcfg != nil && vcfg.Versioned(version.Name)
+
+								results <- version.ToObjectInfo(bucket, version.Name, versioned)
 							}
 							return
 						}
 						for _, version := range fivs.Versions {
-							results <- version.ToObjectInfo(bucket, version.Name)
+							versioned := vcfg != nil && vcfg.Versioned(version.Name)
+							results <- version.ToObjectInfo(bucket, version.Name, versioned)
 						}
 					}
 

cmd/erasure-sets.go

@@ -301,14 +301,14 @@ func (s *erasureSets) monitorAndConnectEndpoints(ctx context.Context, monitorInt
 		case <-ctx.Done():
 			return
 		case <-monitor.C:
-			// Reset the timer once fired for required interval.
-			monitor.Reset(monitorInterval)
-
 			if serverDebugLog {
 				console.Debugln("running disk monitoring")
 			}
 
 			s.connectDisks()
+
+			// Reset the timer for next interval
+			monitor.Reset(monitorInterval)
 		}
 	}
 }
@@ -485,7 +485,9 @@ func newErasureSets(ctx context.Context, endpoints PoolEndpoints, storageDisks [
 	go s.cleanupDeletedObjects(ctx)
 
 	// Start the disk monitoring and connect routine.
-	go s.monitorAndConnectEndpoints(ctx, defaultMonitorConnectEndpointInterval)
+	if !globalIsTesting {
+		go s.monitorAndConnectEndpoints(ctx, defaultMonitorConnectEndpointInterval)
+	}
 
 	return s, nil
 }
@@ -502,9 +504,6 @@ func (s *erasureSets) cleanupDeletedObjects(ctx context.Context) {
 		case <-ctx.Done():
 			return
 		case <-timer.C:
-			// Reset for the next interval
-			timer.Reset(globalAPIConfig.getDeleteCleanupInterval())
-
 			var wg sync.WaitGroup
 			for _, set := range s.sets {
 				wg.Add(1)
@@ -517,6 +516,9 @@ func (s *erasureSets) cleanupDeletedObjects(ctx context.Context) {
 				}(set)
 			}
 			wg.Wait()
+
+			// Reset for the next interval
+			timer.Reset(globalAPIConfig.getDeleteCleanupInterval())
 		}
 	}
 }
@@ -530,9 +532,6 @@ func (s *erasureSets) cleanupStaleUploads(ctx context.Context) {
 		case <-ctx.Done():
 			return
 		case <-timer.C:
-			// Reset for the next interval
-			timer.Reset(globalAPIConfig.getStaleUploadsCleanupInterval())
-
 			var wg sync.WaitGroup
 			for _, set := range s.sets {
 				wg.Add(1)
@@ -545,6 +544,9 @@ func (s *erasureSets) cleanupStaleUploads(ctx context.Context) {
 				}(set)
 			}
 			wg.Wait()
+
+			// Reset for the next interval
+			timer.Reset(globalAPIConfig.getStaleUploadsCleanupInterval())
 		}
 	}
 }
@@ -572,6 +574,7 @@ func (a *auditObjectErasureMap) MarshalJSON() ([]byte, error) {
 	return json.Marshal(mapCopy)
 }
 
+// Add erasure set information to the current context
 func auditObjectErasureSet(ctx context.Context, object string, set *erasureObjects) {
 	if len(logger.AuditTargets()) == 0 {
 		return
@@ -933,21 +936,18 @@ func (s *erasureSets) ListBuckets(ctx context.Context) (buckets []BucketInfo, er
 // GetObjectNInfo - returns object info and locked object ReadCloser
 func (s *erasureSets) GetObjectNInfo(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, lockType LockType, opts ObjectOptions) (gr *GetObjectReader, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.GetObjectNInfo(ctx, bucket, object, rs, h, lockType, opts)
 }
 
 // PutObject - writes an object to hashedSet based on the object name.
 func (s *erasureSets) PutObject(ctx context.Context, bucket string, object string, data *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.PutObject(ctx, bucket, object, data, opts)
 }
 
 // GetObjectInfo - reads object metadata from the hashedSet based on the object name.
 func (s *erasureSets) GetObjectInfo(ctx context.Context, bucket, object string, opts ObjectOptions) (objInfo ObjectInfo, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.GetObjectInfo(ctx, bucket, object, opts)
 }
 
@@ -967,13 +967,11 @@ func (s *erasureSets) deletePrefix(ctx context.Context, bucket string, prefix st
 
 // DeleteObject - deletes an object from the hashedSet based on the object name.
 func (s *erasureSets) DeleteObject(ctx context.Context, bucket string, object string, opts ObjectOptions) (objInfo ObjectInfo, err error) {
-	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
-
 	if opts.DeletePrefix {
 		err := s.deletePrefix(ctx, bucket, object)
 		return ObjectInfo{}, err
 	}
+	set := s.getHashedSet(object)
 	return set.DeleteObject(ctx, bucket, object, opts)
 }
 
@@ -1030,9 +1028,6 @@ func (s *erasureSets) DeleteObjects(ctx context.Context, bucket string, objects
 			for i, obj := range group {
 				delErrs[obj.origIndex] = errs[i]
 				delObjects[obj.origIndex] = dobjects[i]
-				if errs[i] == nil {
-					auditObjectErasureSet(ctx, obj.object.ObjectName, set)
-				}
 			}
 		}(s.sets[setIdx], objsGroup)
 	}
@@ -1046,8 +1041,6 @@ func (s *erasureSets) CopyObject(ctx context.Context, srcBucket, srcObject, dstB
 	srcSet := s.getHashedSet(srcObject)
 	dstSet := s.getHashedSet(dstObject)
 
-	auditObjectErasureSet(ctx, dstObject, dstSet)
-
 	cpSrcDstSame := srcSet == dstSet
 	// Check if this request is only metadata update.
 	if cpSrcDstSame && srcInfo.metadataOnly {
@@ -1086,14 +1079,12 @@ func (s *erasureSets) ListMultipartUploads(ctx context.Context, bucket, prefix,
 	// In list multipart uploads we are going to treat input prefix as the object,
 	// this means that we are not supporting directory navigation.
 	set := s.getHashedSet(prefix)
-	auditObjectErasureSet(ctx, prefix, set)
 	return set.ListMultipartUploads(ctx, bucket, prefix, keyMarker, uploadIDMarker, delimiter, maxUploads)
 }
 
 // Initiate a new multipart upload on a hashedSet based on object name.
 func (s *erasureSets) NewMultipartUpload(ctx context.Context, bucket, object string, opts ObjectOptions) (uploadID string, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.NewMultipartUpload(ctx, bucket, object, opts)
 }
 
@@ -1102,42 +1093,36 @@ func (s *erasureSets) CopyObjectPart(ctx context.Context, srcBucket, srcObject,
 	startOffset int64, length int64, srcInfo ObjectInfo, srcOpts, dstOpts ObjectOptions,
 ) (partInfo PartInfo, err error) {
 	destSet := s.getHashedSet(destObject)
-	auditObjectErasureSet(ctx, destObject, destSet)
 	return destSet.PutObjectPart(ctx, destBucket, destObject, uploadID, partID, NewPutObjReader(srcInfo.Reader), dstOpts)
 }
 
 // PutObjectPart - writes part of an object to hashedSet based on the object name.
 func (s *erasureSets) PutObjectPart(ctx context.Context, bucket, object, uploadID string, partID int, data *PutObjReader, opts ObjectOptions) (info PartInfo, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.PutObjectPart(ctx, bucket, object, uploadID, partID, data, opts)
 }
 
 // GetMultipartInfo - return multipart metadata info uploaded at hashedSet.
 func (s *erasureSets) GetMultipartInfo(ctx context.Context, bucket, object, uploadID string, opts ObjectOptions) (result MultipartInfo, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.GetMultipartInfo(ctx, bucket, object, uploadID, opts)
 }
 
 // ListObjectParts - lists all uploaded parts to an object in hashedSet.
 func (s *erasureSets) ListObjectParts(ctx context.Context, bucket, object, uploadID string, partNumberMarker int, maxParts int, opts ObjectOptions) (result ListPartsInfo, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.ListObjectParts(ctx, bucket, object, uploadID, partNumberMarker, maxParts, opts)
 }
 
 // Aborts an in-progress multipart operation on hashedSet based on the object name.
 func (s *erasureSets) AbortMultipartUpload(ctx context.Context, bucket, object, uploadID string, opts ObjectOptions) error {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.AbortMultipartUpload(ctx, bucket, object, uploadID, opts)
 }
 
 // CompleteMultipartUpload - completes a pending multipart transaction, on hashedSet based on object name.
 func (s *erasureSets) CompleteMultipartUpload(ctx context.Context, bucket, object, uploadID string, uploadedParts []CompletePart, opts ObjectOptions) (objInfo ObjectInfo, err error) {
 	set := s.getHashedSet(object)
-	auditObjectErasureSet(ctx, object, set)
 	return set.CompleteMultipartUpload(ctx, bucket, object, uploadID, uploadedParts, opts)
 }
 
@@ -1267,7 +1252,7 @@ func (s *erasureSets) HealFormat(ctx context.Context, dryRun bool) (res madmin.H
 
 	defer func(storageDisks []StorageAPI) {
 		if err != nil {
-			closeStorageDisks(storageDisks)
+			closeStorageDisks(storageDisks...)
 		}
 	}(storageDisks)
 

cmd/erasure.go

@@ -79,7 +79,7 @@ func (er erasureObjects) NewNSLock(bucket string, objects ...string) RWLocker {
 // Shutdown function for object storage interface.
 func (er erasureObjects) Shutdown(ctx context.Context) error {
 	// Add any object layer shutdown activities here.
-	closeStorageDisks(er.getDisks())
+	closeStorageDisks(er.getDisks()...)
 	return nil
 }
 

cmd/format-erasure.go

@@ -40,6 +40,9 @@ const (
 	// Represents Erasure backend.
 	formatBackendErasure = "xl"
 
+	// Represents Erasure backend - single drive
+	formatBackendErasureSingle = "xl-single"
+
 	// formatErasureV1.Erasure.Version - version '1'.
 	formatErasureVersionV1 = "1"
 
@@ -146,6 +149,9 @@ func newFormatErasureV3(numSets int, setLen int) *formatErasureV3 {
 	format := &formatErasureV3{}
 	format.Version = formatMetaVersionV1
 	format.Format = formatBackendErasure
+	if setLen == 1 {
+		format.Format = formatBackendErasureSingle
+	}
 	format.ID = mustGetUUID()
 	format.Erasure.Version = formatErasureVersionV3
 	format.Erasure.DistributionAlgo = formatErasureVersionV3DistributionAlgoV3
@@ -170,8 +176,8 @@ func formatGetBackendErasureVersion(b []byte) (string, error) {
 	if meta.Version != formatMetaVersionV1 {
 		return "", fmt.Errorf(`format.Version expected: %s, got: %s`, formatMetaVersionV1, meta.Version)
 	}
-	if meta.Format != formatBackendErasure {
-		return "", fmt.Errorf(`found backend type %s, expected %s`, meta.Format, formatBackendErasure)
+	if meta.Format != formatBackendErasure && meta.Format != formatBackendErasureSingle {
+		return "", fmt.Errorf(`found backend type %s, expected %s or %s`, meta.Format, formatBackendErasure, formatBackendErasureSingle)
 	}
 	// Erasure backend found, proceed to detect version.
 	format := &formatErasureVersionDetect{}
@@ -291,7 +297,7 @@ func formatErasureMigrateV2ToV3(data []byte, export, version string) ([]byte, er
 func countErrs(errs []error, err error) int {
 	i := 0
 	for _, err1 := range errs {
-		if err1 == err {
+		if err1 == err || errors.Is(err1, err) {
 			i++
 		}
 	}
@@ -410,7 +416,7 @@ func checkFormatErasureValue(formatErasure *formatErasureV3, disk StorageAPI) er
 	if formatErasure.Version != formatMetaVersionV1 {
 		return fmt.Errorf("Unsupported version of backend format [%s] found on %s", formatErasure.Version, disk)
 	}
-	if formatErasure.Format != formatBackendErasure {
+	if formatErasure.Format != formatBackendErasure && formatErasure.Format != formatBackendErasureSingle {
 		return fmt.Errorf("Unsupported backend format [%s] found on %s", formatErasure.Format, disk)
 	}
 	if formatErasure.Erasure.Version != formatErasureVersionV3 {
@@ -643,7 +649,7 @@ func saveFormatErasureAll(ctx context.Context, storageDisks []StorageAPI, format
 }
 
 // relinquishes the underlying connection for all storage disks.
-func closeStorageDisks(storageDisks []StorageAPI) {
+func closeStorageDisks(storageDisks ...StorageAPI) {
 	var wg sync.WaitGroup
 	for _, disk := range storageDisks {
 		if disk == nil {

cmd/format-fs.go

@@ -240,6 +240,9 @@ func initFormatFS(ctx context.Context, fsPath string) (rlk *lock.RLockedFile, er
 		if err != nil {
 			return nil, err
 		}
+		if formatBackend == formatBackendErasureSingle {
+			return nil, errFreshDisk
+		}
 		if formatBackend != formatBackendFS {
 			return nil, fmt.Errorf(`%s file: expected format-type: %s, found: %s`, formatConfigFile, formatBackendFS, formatBackend)
 		}
@@ -319,6 +322,10 @@ func formatFSFixDeploymentID(ctx context.Context, fsFormatPath string) error {
 		rlk.Close()
 		return err
 	}
+	if formatBackend == formatBackendErasureSingle {
+		rlk.Close()
+		return errFreshDisk
+	}
 	if formatBackend != formatBackendFS {
 		rlk.Close()
 		return fmt.Errorf(`%s file: expected format-type: %s, found: %s`, formatConfigFile, formatBackendFS, formatBackend)

cmd/fs-v1-metadata_test.go

@@ -45,6 +45,8 @@ func TestFSV1MetadataObjInfo(t *testing.T) {
 
 // TestReadFSMetadata - readFSMetadata testing with a healthy and faulty disk
 func TestReadFSMetadata(t *testing.T) {
+	t.Skip()
+
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)
 
@@ -80,6 +82,7 @@ func TestReadFSMetadata(t *testing.T) {
 
 // TestWriteFSMetadata - tests of writeFSMetadata with healthy disk.
 func TestWriteFSMetadata(t *testing.T) {
+	t.Skip()
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)
 

cmd/fs-v1-multipart.go

@@ -887,8 +887,6 @@ func (fs *FSObjects) cleanupStaleUploads(ctx context.Context) {
 		case <-ctx.Done():
 			return
 		case <-bgAppendTmpCleaner.C:
-			bgAppendTmpCleaner.Reset(bgAppendsCleanupInterval)
-
 			foundUploadIDs := fs.getAllUploadIDs(ctx)
 
 			// Remove background append map from the memory
@@ -915,10 +913,8 @@ func (fs *FSObjects) cleanupStaleUploads(ctx context.Context) {
 				}
 			}
 
+			bgAppendTmpCleaner.Reset(bgAppendsCleanupInterval)
 		case <-expiryUploadsTimer.C:
-			// Reset for the next interval
-			expiryUploadsTimer.Reset(globalAPIConfig.getStaleUploadsCleanupInterval())
-
 			expiry := globalAPIConfig.getStaleUploadsExpiry()
 			now := time.Now()
 
@@ -944,6 +940,9 @@ func (fs *FSObjects) cleanupStaleUploads(ctx context.Context) {
 					fs.appendFileMapMu.Unlock()
 				}
 			}
+
+			// Reset for the next interval
+			expiryUploadsTimer.Reset(globalAPIConfig.getStaleUploadsCleanupInterval())
 		}
 	}
 }

cmd/fs-v1-multipart_test.go

@@ -32,6 +32,7 @@ import (
 
 // Tests cleanup multipart uploads for filesystem backend.
 func TestFSCleanupMultipartUploadsInRoutine(t *testing.T) {
+	t.Skip()
 	// Prepare for tests
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)
@@ -88,6 +89,7 @@ func TestFSCleanupMultipartUploadsInRoutine(t *testing.T) {
 
 // TestNewMultipartUploadFaultyDisk - test NewMultipartUpload with faulty disks
 func TestNewMultipartUploadFaultyDisk(t *testing.T) {
+	t.Skip()
 	// Prepare for tests
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)

cmd/fs-v1.go

@@ -142,6 +142,11 @@ func NewFSObjectLayer(fsPath string) (ObjectLayer, error) {
 		return nil, config.ErrUnableToWriteInBackend(err).Hint(hint)
 	}
 
+	fsFormatPath := pathJoin(fsPath, minioMetaBucket, formatConfigFile)
+	if _, err = fsStat(ctx, fsFormatPath); err != nil && os.IsNotExist(err) {
+		return nil, errFreshDisk
+	}
+
 	// Assign a new UUID for FS minio mode. Each server instance
 	// gets its own UUID for temporary file transaction.
 	fsUUID := mustGetUUID()

cmd/fs-v1_test.go

@@ -51,6 +51,8 @@ func TestNewFS(t *testing.T) {
 // TestFSShutdown - initialize a new FS object layer then calls
 // Shutdown to check returned results
 func TestFSShutdown(t *testing.T) {
+	t.Skip()
+
 	bucketName := "testbucket"
 	objectName := "object"
 	// Create and return an fsObject with its path in the disk
@@ -83,6 +85,8 @@ func TestFSShutdown(t *testing.T) {
 
 // TestFSGetBucketInfo - test GetBucketInfo with healty and faulty disks
 func TestFSGetBucketInfo(t *testing.T) {
+	t.Skip()
+
 	// Prepare for testing
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)
@@ -165,6 +169,7 @@ func TestFSPutObject(t *testing.T) {
 
 // TestFSDeleteObject - test fs.DeleteObject() with healthy and corrupted disks
 func TestFSDeleteObject(t *testing.T) {
+	t.Skip()
 	// Prepare for tests
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)
@@ -209,6 +214,7 @@ func TestFSDeleteObject(t *testing.T) {
 
 // TestFSDeleteBucket - tests for fs DeleteBucket
 func TestFSDeleteBucket(t *testing.T) {
+	t.Skip()
 	// Prepare for testing
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)
@@ -249,6 +255,7 @@ func TestFSDeleteBucket(t *testing.T) {
 
 // TestFSListBuckets - tests for fs ListBuckets
 func TestFSListBuckets(t *testing.T) {
+	t.Skip()
 	// Prepare for tests
 	disk := filepath.Join(globalTestTmpDir, "minio-"+nextSuffix())
 	defer os.RemoveAll(disk)

cmd/gateway-main.go

@@ -19,6 +19,7 @@ package cmd
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"io/ioutil"
 	"log"
@@ -32,6 +33,7 @@ import (
 	"github.com/minio/cli"
 	"github.com/minio/madmin-go"
 	"github.com/minio/minio/internal/color"
+	"github.com/minio/minio/internal/config"
 	xhttp "github.com/minio/minio/internal/http"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/certs"
@@ -292,6 +294,9 @@ func StartGateway(ctx *cli.Context, gw Gateway) {
 		SecretKey: globalActiveCred.SecretKey,
 	})
 	if err != nil {
+		if errors.Is(err, errFreshDisk) {
+			err = config.ErrInvalidFSValue(err)
+		}
 		logger.FatalIf(err, "Unable to initialize gateway backend")
 	}
 	newObject = NewGatewayLayerWithLocker(newObject)

cmd/gateway/s3/gateway-s3-sse.go

@@ -364,6 +364,7 @@ func (l *s3EncObjects) GetObjectInfo(ctx context.Context, bucket string, object
 // CopyObject copies an object from source bucket to a destination bucket.
 func (l *s3EncObjects) CopyObject(ctx context.Context, srcBucket string, srcObject string, dstBucket string, dstObject string, srcInfo minio.ObjectInfo, s, d minio.ObjectOptions) (objInfo minio.ObjectInfo, err error) {
 	cpSrcDstSame := path.Join(srcBucket, srcObject) == path.Join(dstBucket, dstObject)
+	userDefined := minio.CloneMSS(srcInfo.UserDefined)
 	if cpSrcDstSame {
 		var gwMeta gwMetaV1
 		if s.ServerSideEncryption != nil && d.ServerSideEncryption != nil &&
@@ -378,16 +379,16 @@ func (l *s3EncObjects) CopyObject(ctx context.Context, srcBucket string, srcObje
 				d.ServerSideEncryption.Marshal(header)
 			}
 			for k, v := range header {
-				srcInfo.UserDefined[k] = v[0]
+				userDefined[k] = v[0]
 			}
-			gwMeta.Meta = srcInfo.UserDefined
+			gwMeta.Meta = userDefined
 			if err = l.writeGWMetadata(ctx, dstBucket, getDareMetaPath(dstObject), gwMeta, minio.ObjectOptions{}); err != nil {
 				return objInfo, minio.ErrorRespToObjectError(err)
 			}
 			return gwMeta.ToObjectInfo(dstBucket, dstObject), nil
 		}
 	}
-	dstOpts := minio.ObjectOptions{ServerSideEncryption: d.ServerSideEncryption, UserDefined: srcInfo.UserDefined}
+	dstOpts := minio.ObjectOptions{ServerSideEncryption: d.ServerSideEncryption, UserDefined: userDefined}
 	return l.PutObject(ctx, dstBucket, dstObject, srcInfo.PutObjReader, dstOpts)
 }
 

cmd/gateway/s3/gateway-s3.go

@@ -502,17 +502,20 @@ func (l *s3Objects) GetObjectInfo(ctx context.Context, bucket string, object str
 // PutObject creates a new object with the incoming data,
 func (l *s3Objects) PutObject(ctx context.Context, bucket string, object string, r *minio.PutObjReader, opts minio.ObjectOptions) (objInfo minio.ObjectInfo, err error) {
 	data := r.Reader
+
+	userDefined := minio.CloneMSS(opts.UserDefined)
+
 	var tagMap map[string]string
-	if tagstr, ok := opts.UserDefined[xhttp.AmzObjectTagging]; ok && tagstr != "" {
+	if tagstr, ok := userDefined[xhttp.AmzObjectTagging]; ok && tagstr != "" {
 		tagObj, err := tags.ParseObjectTags(tagstr)
 		if err != nil {
 			return objInfo, minio.ErrorRespToObjectError(err, bucket, object)
 		}
 		tagMap = tagObj.ToMap()
-		delete(opts.UserDefined, xhttp.AmzObjectTagging)
+		delete(userDefined, xhttp.AmzObjectTagging)
 	}
 	putOpts := miniogo.PutObjectOptions{
-		UserMetadata:         opts.UserDefined,
+		UserMetadata:         userDefined,
 		ServerSideEncryption: opts.ServerSideEncryption,
 		UserTags:             tagMap,
 		// Content-Md5 is needed for buckets with object locking,
@@ -529,7 +532,7 @@ func (l *s3Objects) PutObject(ctx context.Context, bucket string, object string,
 		ETag:     ui.ETag,
 		Size:     ui.Size,
 		Key:      object,
-		Metadata: minio.ToMinioClientObjectInfoMetadata(opts.UserDefined),
+		Metadata: minio.ToMinioClientObjectInfoMetadata(userDefined),
 	}
 
 	return minio.FromMinioClientObjectInfo(bucket, oi), nil
@@ -544,8 +547,9 @@ func (l *s3Objects) CopyObject(ctx context.Context, srcBucket string, srcObject
 	// metadata input is already a trickled down value from interpreting x-amz-metadata-directive at
 	// handler layer. So what we have right now is supposed to be applied on the destination object anyways.
 	// So preserve it by adding "REPLACE" directive to save all the metadata set by CopyObject API.
-	srcInfo.UserDefined["x-amz-metadata-directive"] = "REPLACE"
-	srcInfo.UserDefined["x-amz-copy-source-if-match"] = srcInfo.ETag
+	userDefined := minio.CloneMSS(srcInfo.UserDefined)
+	userDefined["x-amz-metadata-directive"] = "REPLACE"
+	userDefined["x-amz-copy-source-if-match"] = srcInfo.ETag
 	header := make(http.Header)
 	if srcOpts.ServerSideEncryption != nil {
 		encrypt.SSECopy(srcOpts.ServerSideEncryption).Marshal(header)
@@ -556,10 +560,10 @@ func (l *s3Objects) CopyObject(ctx context.Context, srcBucket string, srcObject
 	}
 
 	for k, v := range header {
-		srcInfo.UserDefined[k] = v[0]
+		userDefined[k] = v[0]
 	}
 
-	if _, err = l.Client.CopyObject(ctx, srcBucket, srcObject, dstBucket, dstObject, srcInfo.UserDefined, miniogo.CopySrcOptions{}, miniogo.PutObjectOptions{}); err != nil {
+	if _, err = l.Client.CopyObject(ctx, srcBucket, srcObject, dstBucket, dstObject, userDefined, miniogo.CopySrcOptions{}, miniogo.PutObjectOptions{}); err != nil {
 		return objInfo, minio.ErrorRespToObjectError(err, srcBucket, srcObject)
 	}
 	return l.GetObjectInfo(ctx, dstBucket, dstObject, dstOpts)
@@ -605,17 +609,18 @@ func (l *s3Objects) ListMultipartUploads(ctx context.Context, bucket string, pre
 // NewMultipartUpload upload object in multiple parts
 func (l *s3Objects) NewMultipartUpload(ctx context.Context, bucket string, object string, o minio.ObjectOptions) (uploadID string, err error) {
 	var tagMap map[string]string
-	if tagStr, ok := o.UserDefined[xhttp.AmzObjectTagging]; ok {
+	userDefined := minio.CloneMSS(o.UserDefined)
+	if tagStr, ok := userDefined[xhttp.AmzObjectTagging]; ok {
 		tagObj, err := tags.Parse(tagStr, true)
 		if err != nil {
 			return uploadID, minio.ErrorRespToObjectError(err, bucket, object)
 		}
 		tagMap = tagObj.ToMap()
-		delete(o.UserDefined, xhttp.AmzObjectTagging)
+		delete(userDefined, xhttp.AmzObjectTagging)
 	}
 	// Create PutObject options
 	opts := miniogo.PutObjectOptions{
-		UserMetadata:         o.UserDefined,
+		UserMetadata:         userDefined,
 		ServerSideEncryption: o.ServerSideEncryption,
 		UserTags:             tagMap,
 	}
@@ -645,9 +650,8 @@ func (l *s3Objects) CopyObjectPart(ctx context.Context, srcBucket, srcObject, de
 	if srcOpts.CheckPrecondFn != nil && srcOpts.CheckPrecondFn(srcInfo) {
 		return minio.PartInfo{}, minio.PreConditionFailed{}
 	}
-	srcInfo.UserDefined = map[string]string{
-		"x-amz-copy-source-if-match": srcInfo.ETag,
-	}
+	userDefined := minio.CloneMSS(srcInfo.UserDefined)
+	userDefined["x-amz-copy-source-if-match"] = srcInfo.ETag
 	header := make(http.Header)
 	if srcOpts.ServerSideEncryption != nil {
 		encrypt.SSECopy(srcOpts.ServerSideEncryption).Marshal(header)
@@ -657,11 +661,11 @@ func (l *s3Objects) CopyObjectPart(ctx context.Context, srcBucket, srcObject, de
 		dstOpts.ServerSideEncryption.Marshal(header)
 	}
 	for k, v := range header {
-		srcInfo.UserDefined[k] = v[0]
+		userDefined[k] = v[0]
 	}
 
 	completePart, err := l.Client.CopyObjectPart(ctx, srcBucket, srcObject, destBucket, destObject,
-		uploadID, partID, startOffset, length, srcInfo.UserDefined)
+		uploadID, partID, startOffset, length, userDefined)
 	if err != nil {
 		return p, minio.ErrorRespToObjectError(err, srcBucket, srcObject)
 	}

cmd/generic-handlers.go

@@ -18,6 +18,7 @@
 package cmd
 
 import (
+	"fmt"
 	"net"
 	"net/http"
 	"path"
@@ -141,6 +142,14 @@ func setBrowserRedirectHandler(h http.Handler) http.Handler {
 	})
 }
 
+var redirectPrefixes = map[string]struct{}{
+	"favicon-16x16.png": {},
+	"favicon-32x32.png": {},
+	"favicon-96x96.png": {},
+	"index.html":        {},
+	minioReservedBucket: {},
+}
+
 // Fetch redirect location if urlPath satisfies certain
 // criteria. Some special names are considered to be
 // redirectable, this is purely internal function and
@@ -151,32 +160,25 @@ func getRedirectLocation(r *http.Request) *xnet.URL {
 	if err != nil {
 		return nil
 	}
-	for _, prefix := range []string{
-		"favicon-16x16.png",
-		"favicon-32x32.png",
-		"favicon-96x96.png",
-		"index.html",
-		minioReservedBucket,
-	} {
-		bucket, _ := path2BucketObject(resource)
-		if path.Clean(bucket) == prefix || resource == slashSeparator {
-			if globalBrowserRedirectURL != nil {
-				return globalBrowserRedirectURL
-			}
-			hostname, _, _ := net.SplitHostPort(r.Host)
-			if hostname == "" {
-				hostname = r.Host
-			}
-			return &xnet.URL{
-				Host: net.JoinHostPort(hostname, globalMinioConsolePort),
-				Scheme: func() string {
-					scheme := "http"
-					if r.TLS != nil {
-						scheme = "https"
-					}
-					return scheme
-				}(),
-			}
+	bucket, _ := path2BucketObject(resource)
+	_, redirect := redirectPrefixes[path.Clean(bucket)]
+	if redirect || resource == slashSeparator {
+		if globalBrowserRedirectURL != nil {
+			return globalBrowserRedirectURL
+		}
+		xhost, err := xnet.ParseHost(r.Host)
+		if err != nil {
+			return nil
+		}
+		return &xnet.URL{
+			Host: net.JoinHostPort(xhost.Name, globalMinioConsolePort),
+			Scheme: func() string {
+				scheme := "http"
+				if r.TLS != nil {
+					scheme = "https"
+				}
+				return scheme
+			}(),
 		}
 	}
 	return nil
@@ -294,6 +296,15 @@ const (
 	dotComponent    = "."
 )
 
+func hasBadHost(host string) error {
+	if globalIsCICD && strings.TrimSpace(host) == "" {
+		// under CI/CD test setups ignore empty hosts as invalid hosts
+		return nil
+	}
+	_, err := xnet.ParseHost(host)
+	return err
+}
+
 // Check if the incoming path has bad path components,
 // such as ".." and "."
 func hasBadPathComponent(path string) bool {
@@ -312,7 +323,11 @@ func hasBadPathComponent(path string) bool {
 // Check if client is sending a malicious request.
 func hasMultipleAuth(r *http.Request) bool {
 	authTypeCount := 0
-	for _, hasValidAuth := range []func(*http.Request) bool{isRequestSignatureV2, isRequestPresignedSignatureV2, isRequestSignatureV4, isRequestPresignedSignatureV4, isRequestJWT, isRequestPostPolicySignatureV4} {
+	for _, hasValidAuth := range []func(*http.Request) bool{
+		isRequestSignatureV2, isRequestPresignedSignatureV2,
+		isRequestSignatureV4, isRequestPresignedSignatureV4,
+		isRequestJWT, isRequestPostPolicySignatureV4,
+	} {
 		if hasValidAuth(r) {
 			authTypeCount++
 		}
@@ -324,6 +339,14 @@ func hasMultipleAuth(r *http.Request) bool {
 // any malicious requests.
 func setRequestValidityHandler(h http.Handler) http.Handler {
 	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if err := hasBadHost(r.Host); err != nil {
+			invalidReq := errorCodes.ToAPIErr(ErrInvalidRequest)
+			invalidReq.Description = fmt.Sprintf("%s (%s)", invalidReq.Description, err)
+			writeErrorResponse(r.Context(), w, invalidReq, r.URL)
+			atomic.AddUint64(&globalHTTPStats.rejectedRequestsInvalid, 1)
+			return
+		}
+
 		// Check for bad components in URL path.
 		if hasBadPathComponent(r.URL.Path) {
 			writeErrorResponse(r.Context(), w, errorCodes.ToAPIErr(ErrInvalidResourceName), r.URL)
@@ -341,7 +364,9 @@ func setRequestValidityHandler(h http.Handler) http.Handler {
 			}
 		}
 		if hasMultipleAuth(r) {
-			writeErrorResponse(r.Context(), w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
+			invalidReq := errorCodes.ToAPIErr(ErrInvalidRequest)
+			invalidReq.Description = fmt.Sprintf("%s (request has multiple authentication types, please use one)", invalidReq.Description)
+			writeErrorResponse(r.Context(), w, invalidReq, r.URL)
 			atomic.AddUint64(&globalHTTPStats.rejectedRequestsInvalid, 1)
 			return
 		}

cmd/global-heal.go

@@ -340,7 +340,9 @@ func (er *erasureObjects) healErasureSet(ctx context.Context, buckets []string,
 // healObject heals given object path in deep to fix bitrot.
 func healObject(bucket, object, versionID string, scan madmin.HealScanMode) {
 	// Get background heal sequence to send elements to heal
+	globalHealStateLK.Lock()
 	bgSeq, ok := globalBackgroundHealState.getHealSequenceByToken(bgHealingUUID)
+	globalHealStateLK.Unlock()
 	if ok {
 		bgSeq.queueHealTask(healSource{
 			bucket:    bucket,

cmd/globals.go

@@ -37,12 +37,14 @@ import (
 	"github.com/dustin/go-humanize"
 	"github.com/minio/minio/internal/auth"
 	"github.com/minio/minio/internal/config/cache"
+	"github.com/minio/minio/internal/config/callhome"
 	"github.com/minio/minio/internal/config/compress"
 	"github.com/minio/minio/internal/config/dns"
 	xldap "github.com/minio/minio/internal/config/identity/ldap"
 	"github.com/minio/minio/internal/config/identity/openid"
+	idplugin "github.com/minio/minio/internal/config/identity/plugin"
 	xtls "github.com/minio/minio/internal/config/identity/tls"
-	"github.com/minio/minio/internal/config/policy/opa"
+	polplugin "github.com/minio/minio/internal/config/policy/plugin"
 	"github.com/minio/minio/internal/config/storageclass"
 	"github.com/minio/minio/internal/config/subnet"
 	xhttp "github.com/minio/minio/internal/http"
@@ -73,6 +75,7 @@ const (
 	globalWindowsOSName            = "windows"
 	globalMacOSName                = "darwin"
 	globalMinioModeFS              = "mode-server-fs"
+	globalMinioModeErasureSD       = "mode-server-xl-single"
 	globalMinioModeErasure         = "mode-server-xl"
 	globalMinioModeDistErasure     = "mode-server-distributed-xl"
 	globalMinioModeGatewayPrefix   = "mode-gateway-"
@@ -113,6 +116,10 @@ const (
 	// diskFillFraction is the fraction of a disk we allow to be filled.
 	diskFillFraction = 0.99
 
+	// diskReserveFraction is the fraction of a disk where we will fill other server pools first.
+	// If all pools reach this, we will use all pools with regular placement.
+	diskReserveFraction = 0.15
+
 	// diskAssumeUnknownSize is the size to assume when an unknown size upload is requested.
 	diskAssumeUnknownSize = 1 << 30
 
@@ -136,9 +143,15 @@ var (
 	// Indicates if the running minio server is an erasure-code backend.
 	globalIsErasure = false
 
+	// Indicates if the running minio server is in single drive XL mode.
+	globalIsErasureSD = false
+
 	// Indicates if the running minio is in gateway mode.
 	globalIsGateway = false
 
+	// Indicates if server code should go through testing path.
+	globalIsTesting = false
+
 	// Name of gateway server, e.g S3, NAS etc
 	globalGatewayName = ""
 
@@ -193,10 +206,13 @@ var (
 	globalAPIConfig = apiConfig{listQuorum: "strict"}
 
 	globalStorageClass storageclass.Config
+
 	globalLDAPConfig   xldap.Config
 	globalOpenIDConfig openid.Config
 	globalSTSTLSConfig xtls.Config
 
+	globalAuthNPlugin *idplugin.AuthNPlugin
+
 	// CA root certificates, a nil value means system certs pool will be used
 	globalRootCAs *x509.CertPool
 
@@ -211,10 +227,10 @@ var (
 
 	// global Trace system to send HTTP request/response
 	// and Storage/OS calls info to registered listeners.
-	globalTrace = pubsub.New()
+	globalTrace = pubsub.New(8)
 
 	// global Listen system to send S3 API events to registered listeners
-	globalHTTPListen = pubsub.New()
+	globalHTTPListen = pubsub.New(0)
 
 	// global console system to send console logs to
 	// registered listeners
@@ -228,6 +244,9 @@ var (
 	// The global subnet config
 	globalSubnetConfig subnet.Config
 
+	// The global callhome config
+	globalCallhomeConfig callhome.Config
+
 	globalRemoteEndpoints map[string]Endpoint
 
 	// Global server's network statistics
@@ -290,8 +309,8 @@ var (
 	// Some standard content-types which we strictly dis-allow for compression.
 	standardExcludeCompressContentTypes = []string{"video/*", "audio/*", "application/zip", "application/x-gzip", "application/x-zip-compressed", " application/x-compress", "application/x-spoon"}
 
-	// OPA policy system.
-	globalPolicyOPA *opa.Opa
+	// AuthZ Plugin system.
+	globalAuthZPlugin *polplugin.AuthZPlugin
 
 	// Deployment ID - unique per deployment
 	globalDeploymentID string
@@ -352,4 +371,18 @@ var (
 	// Add new variable global values here.
 )
 
+var globalAuthZPluginMutex sync.Mutex
+
+func newGlobalAuthZPluginFn() *polplugin.AuthZPlugin {
+	globalAuthZPluginMutex.Lock()
+	defer globalAuthZPluginMutex.Unlock()
+	return globalAuthZPlugin
+}
+
+func setGlobalAuthZPlugin(authz *polplugin.AuthZPlugin) {
+	globalAuthZPluginMutex.Lock()
+	globalAuthZPlugin = authz
+	globalAuthZPluginMutex.Unlock()
+}
+
 var errSelfTestFailure = errors.New("self test failed. unsafe to start server")

cmd/handler-api.go

@@ -116,10 +116,7 @@ func (t *apiConfig) init(cfg api.Config, setDriveCounts []int) {
 		//    + 2 * 10MiB (default erasure block size v1) + 2 * 1MiB (default erasure block size v2)
 		blockSize := xioutil.BlockSizeLarge + xioutil.BlockSizeSmall
 		apiRequestsMaxPerNode = int(maxMem / uint64(maxSetDrives*blockSize+int(blockSizeV1*2+blockSizeV2*2)))
-
-		if globalIsErasure {
-			logger.Info("Automatically configured API requests per node based on available memory on the system: %d", apiRequestsMaxPerNode)
-		}
+		logger.Info("Automatically configured API requests per node based on available memory on the system: %d", apiRequestsMaxPerNode)
 	} else {
 		apiRequestsMaxPerNode = cfg.RequestsMax
 		if len(globalEndpoints.Hostnames()) > 0 {

cmd/handler-utils.go

@@ -25,7 +25,6 @@ import (
 	"io"
 	"io/ioutil"
 	"mime/multipart"
-	"net"
 	"net/http"
 	"net/textproto"
 	"net/url"
@@ -37,6 +36,7 @@ import (
 	"github.com/minio/minio/internal/handlers"
 	xhttp "github.com/minio/minio/internal/http"
 	"github.com/minio/minio/internal/logger"
+	xnet "github.com/minio/pkg/net"
 )
 
 const (
@@ -237,6 +237,10 @@ func extractReqParams(r *http.Request) map[string]string {
 		"sourceIPAddress": handlers.GetSourceIP(r),
 		// Add more fields here.
 	}
+	if rangeField := r.Header.Get(xhttp.Range); rangeField != "" {
+		m["range"] = rangeField
+	}
+
 	if _, ok := r.Header[xhttp.MinIOSourceReplicationRequest]; ok {
 		m[xhttp.MinIOSourceReplicationRequest] = ""
 	}
@@ -395,26 +399,25 @@ func getResource(path string, host string, domains []string) (string, error) {
 	if len(domains) == 0 {
 		return path, nil
 	}
+
 	// If virtual-host-style is enabled construct the "resource" properly.
-	if strings.Contains(host, ":") {
-		// In bucket.mydomain.com:9000, strip out :9000
-		var err error
-		if host, _, err = net.SplitHostPort(host); err != nil {
-			reqInfo := (&logger.ReqInfo{}).AppendTags("host", host)
-			reqInfo.AppendTags("path", path)
-			ctx := logger.SetReqInfo(GlobalContext, reqInfo)
-			logger.LogIf(ctx, err)
-			return "", err
-		}
+	xhost, err := xnet.ParseHost(host)
+	if err != nil {
+		reqInfo := (&logger.ReqInfo{}).AppendTags("host", host)
+		reqInfo.AppendTags("path", path)
+		ctx := logger.SetReqInfo(context.Background(), reqInfo)
+		logger.LogIf(ctx, err)
+		return "", err
 	}
+
 	for _, domain := range domains {
-		if host == minioReservedBucket+"."+domain {
+		if xhost.Name == minioReservedBucket+"."+domain {
 			continue
 		}
-		if !strings.HasSuffix(host, "."+domain) {
+		if !strings.HasSuffix(xhost.Name, "."+domain) {
 			continue
 		}
-		bucket := strings.TrimSuffix(host, "."+domain)
+		bucket := strings.TrimSuffix(xhost.Name, "."+domain)
 		return SlashSeparator + pathJoin(bucket, path), nil
 	}
 	return path, nil

cmd/handler-utils_test.go

@@ -221,6 +221,9 @@ func TestGetResource(t *testing.T) {
 		expectedResource string
 	}{
 		{"/a/b/c", "test.mydomain.com", []string{"mydomain.com"}, "/test/a/b/c"},
+		{"/a/b/c", "[2001:0db8:85a3:0000:0000:8a2e:0370:7334]:17000", []string{"mydomain.com"}, "/a/b/c"},
+		{"/a/b/c", "[2001:0db8:85a3:0000:0000:8a2e:0370:7334]", []string{"mydomain.com"}, "/a/b/c"},
+		{"/a/b/c", "192.168.1.1:9000", []string{"mydomain.com"}, "/a/b/c"},
 		{"/a/b/c", "test.mydomain.com", []string{"notmydomain.com"}, "/a/b/c"},
 		{"/a/b/c", "test.mydomain.com", nil, "/a/b/c"},
 	}

cmd/hasher.go

@@ -19,8 +19,9 @@ package cmd
 
 import (
 	"crypto/md5"
-	"crypto/sha256"
 	"encoding/hex"
+
+	"github.com/minio/minio/internal/hash/sha256"
 )
 
 // getSHA256Hash returns SHA-256 hash in hex encoding of given data.

cmd/iam-store.go

@@ -158,8 +158,7 @@ type MappedPolicy struct {
 func (mp MappedPolicy) toSlice() []string {
 	var policies []string
 	for _, policy := range strings.Split(mp.Policies, ",") {
-		policy = strings.TrimSpace(policy)
-		if policy == "" {
+		if strings.TrimSpace(policy) == "" {
 			continue
 		}
 		policies = append(policies, policy)
@@ -245,6 +244,8 @@ type iamWatchEvent struct {
 
 // iamCache contains in-memory cache of IAM data.
 type iamCache struct {
+	updatedAt time.Time
+
 	// map of policy names to policy definitions
 	iamPolicyDocsMap map[string]PolicyDoc
 	// map of usernames to credentials
@@ -433,8 +434,7 @@ func setDefaultCannedPolicies(policies map[string]PolicyDoc) {
 func (store *IAMStoreSys) LoadIAMCache(ctx context.Context) error {
 	newCache := newIamCache()
 
-	cache := store.lock()
-	defer store.unlock()
+	loadedAt := time.Now()
 
 	if iamOS, ok := store.IAMStorageAPI.(*IAMObjectStore); ok {
 		err := iamOS.loadAllFromObjStore(ctx, newCache)
@@ -487,12 +487,28 @@ func (store *IAMStoreSys) LoadIAMCache(ctx context.Context) error {
 		newCache.buildUserGroupMemberships()
 	}
 
-	cache.iamGroupPolicyMap = newCache.iamGroupPolicyMap
-	cache.iamGroupsMap = newCache.iamGroupsMap
-	cache.iamPolicyDocsMap = newCache.iamPolicyDocsMap
-	cache.iamUserGroupMemberships = newCache.iamUserGroupMemberships
-	cache.iamUserPolicyMap = newCache.iamUserPolicyMap
-	cache.iamUsersMap = newCache.iamUsersMap
+	cache := store.lock()
+	defer store.unlock()
+
+	// We should only update the in-memory cache if there were no changes
+	// to the in-memory cache since the disk loading began. If there
+	// were changes to the in-memory cache we should wait for the next
+	// cycle until we can safely update the in-memory cache.
+	//
+	// An in-memory cache must be replaced only if we know for sure that
+	// the values loaded from disk are not stale. They might be stale
+	// if the cached.updatedAt is recent than the refresh cycle began.
+	if cache.updatedAt.Before(loadedAt) {
+		// No one has updated anything since the config was loaded,
+		// so we just replace whatever is on the disk into memory.
+		cache.iamGroupPolicyMap = newCache.iamGroupPolicyMap
+		cache.iamGroupsMap = newCache.iamGroupsMap
+		cache.iamPolicyDocsMap = newCache.iamPolicyDocsMap
+		cache.iamUserGroupMemberships = newCache.iamUserGroupMemberships
+		cache.iamUserPolicyMap = newCache.iamUserPolicyMap
+		cache.iamUsersMap = newCache.iamUsersMap
+		cache.updatedAt = time.Now()
+	}
 
 	return nil
 }
@@ -549,6 +565,8 @@ func (store *IAMStoreSys) GroupNotificationHandler(ctx context.Context, group st
 		cache.removeGroupFromMembershipsMap(group)
 		delete(cache.iamGroupsMap, group)
 		delete(cache.iamGroupPolicyMap, group)
+
+		cache.updatedAt = time.Now()
 		return nil
 	}
 
@@ -563,6 +581,7 @@ func (store *IAMStoreSys) GroupNotificationHandler(ctx context.Context, group st
 	// removed, the cache stays current.
 	cache.removeGroupFromMembershipsMap(group)
 	cache.updateGroupMembershipsMap(group, &gi)
+	cache.updatedAt = time.Now()
 	return nil
 }
 
@@ -640,6 +659,8 @@ func (store *IAMStoreSys) AddUsersToGroup(ctx context.Context, group string, mem
 		cache.iamUserGroupMemberships[member] = gset
 	}
 
+	cache.updatedAt = time.Now()
+
 	return nil
 }
 
@@ -673,6 +694,7 @@ func removeMembersFromGroup(ctx context.Context, store *IAMStoreSys, cache *iamC
 		cache.iamUserGroupMemberships[member] = gset
 	}
 
+	cache.updatedAt = time.Now()
 	return nil
 }
 
@@ -721,6 +743,7 @@ func (store *IAMStoreSys) RemoveUsersFromGroup(ctx context.Context, group string
 		// Delete from server memory
 		delete(cache.iamGroupsMap, group)
 		delete(cache.iamGroupPolicyMap, group)
+		cache.updatedAt = time.Now()
 		return nil
 	}
 
@@ -750,7 +773,10 @@ func (store *IAMStoreSys) SetGroupStatus(ctx context.Context, group string, enab
 	if err := store.saveGroupInfo(ctx, group, gi); err != nil {
 		return err
 	}
+
 	cache.iamGroupsMap[group] = gi
+	cache.updatedAt = time.Now()
+
 	return nil
 }
 
@@ -801,7 +827,7 @@ func (store *IAMStoreSys) ListGroups(ctx context.Context) (res []string, err err
 			return
 		}
 		cache.iamGroupsMap = m
-
+		cache.updatedAt = time.Now()
 		for k := range cache.iamGroupsMap {
 			res = append(res, k)
 		}
@@ -814,6 +840,7 @@ func (store *IAMStoreSys) ListGroups(ctx context.Context) (res []string, err err
 			return
 		}
 		cache.iamGroupPolicyMap = m
+		cache.updatedAt = time.Now()
 		for k := range cache.iamGroupPolicyMap {
 			res = append(res, k)
 		}
@@ -862,6 +889,8 @@ func (store *IAMStoreSys) PolicyDBSet(ctx context.Context, name, policy string,
 		} else {
 			delete(cache.iamGroupPolicyMap, name)
 		}
+		cache.updatedAt = time.Now()
+
 		return nil
 	}
 
@@ -882,6 +911,7 @@ func (store *IAMStoreSys) PolicyDBSet(ctx context.Context, name, policy string,
 	} else {
 		cache.iamGroupPolicyMap[name] = mp
 	}
+	cache.updatedAt = time.Now()
 	return nil
 }
 
@@ -931,6 +961,7 @@ func (store *IAMStoreSys) PolicyNotificationHandler(ctx context.Context, policy
 			cache.iamGroupPolicyMap[g] = newMappedPolicy(strings.Join(pset.ToSlice(), ","))
 		}
 
+		cache.updatedAt = time.Now()
 		return nil
 	}
 	return err
@@ -988,6 +1019,8 @@ func (store *IAMStoreSys) DeletePolicy(ctx context.Context, policy string) error
 	}
 
 	delete(cache.iamPolicyDocsMap, policy)
+	cache.updatedAt = time.Now()
+
 	return nil
 }
 
@@ -1058,6 +1091,8 @@ func (store *IAMStoreSys) SetPolicy(ctx context.Context, name string, policy iam
 	}
 
 	cache.iamPolicyDocsMap[name] = d
+	cache.updatedAt = time.Now()
+
 	return nil
 }
 
@@ -1077,6 +1112,7 @@ func (store *IAMStoreSys) ListPolicies(ctx context.Context, bucketName string) (
 	setDefaultCannedPolicies(m)
 
 	cache.iamPolicyDocsMap = m
+	cache.updatedAt = time.Now()
 
 	ret := map[string]iampolicy.Policy{}
 	for k, v := range m {
@@ -1104,6 +1140,7 @@ func (store *IAMStoreSys) ListPolicyDocs(ctx context.Context, bucketName string)
 	setDefaultCannedPolicies(m)
 
 	cache.iamPolicyDocsMap = m
+	cache.updatedAt = time.Now()
 
 	ret := map[string]PolicyDoc{}
 	for k, v := range m {
@@ -1300,6 +1337,8 @@ func (store *IAMStoreSys) PolicyMappingNotificationHandler(ctx context.Context,
 		// This means that the policy mapping was deleted, so we update
 		// the cache.
 		delete(m, userOrGroup)
+		cache.updatedAt = time.Now()
+
 		err = nil
 	}
 	return err
@@ -1348,11 +1387,15 @@ func (store *IAMStoreSys) UserNotificationHandler(ctx context.Context, accessKey
 
 		// 3. Delete any mapped policy
 		delete(cache.iamUserPolicyMap, accessKey)
+
+		cache.updatedAt = time.Now()
 		return nil
 	}
+
 	if err != nil {
 		return err
 	}
+
 	if userType != svcUser {
 		err = store.loadMappedPolicy(ctx, accessKey, userType, false, cache.iamUserPolicyMap)
 		// Ignore policy not mapped error
@@ -1374,6 +1417,7 @@ func (store *IAMStoreSys) UserNotificationHandler(ctx context.Context, accessKey
 		if cred.IsTemp() && cred.ParentUser != "" && cred.ParentUser != globalActiveCred.AccessKey {
 			if _, ok := cache.iamUserPolicyMap[cred.ParentUser]; !ok {
 				cache.iamUserPolicyMap[cred.ParentUser] = cache.iamUserPolicyMap[accessKey]
+				cache.updatedAt = time.Now()
 			}
 		}
 	}
@@ -1431,6 +1475,8 @@ func (store *IAMStoreSys) DeleteUser(ctx context.Context, accessKey string, user
 	}
 	delete(cache.iamUsersMap, accessKey)
 
+	cache.updatedAt = time.Now()
+
 	return err
 }
 
@@ -1470,6 +1516,9 @@ func (store *IAMStoreSys) SetTempUser(ctx context.Context, accessKey string, cre
 	}
 
 	cache.iamUsersMap[accessKey] = cred
+
+	cache.updatedAt = time.Now()
+
 	return nil
 }
 
@@ -1480,6 +1529,7 @@ func (store *IAMStoreSys) DeleteUsers(ctx context.Context, users []string) error
 	cache := store.lock()
 	defer store.unlock()
 
+	var deleted bool
 	usersToDelete := set.CreateStringSet(users...)
 	for user, cred := range cache.iamUsersMap {
 		userType := regUser
@@ -1498,9 +1548,15 @@ func (store *IAMStoreSys) DeleteUsers(ctx context.Context, users []string) error
 			err := store.deleteUserIdentity(ctx, user, userType)
 			logger.LogIf(GlobalContext, err)
 			delete(cache.iamUsersMap, user)
+
+			deleted = true
 		}
 	}
 
+	if deleted {
+		cache.updatedAt = time.Now()
+	}
+
 	return nil
 }
 
@@ -1611,6 +1667,8 @@ func (store *IAMStoreSys) SetUserStatus(ctx context.Context, accessKey string, s
 	}
 
 	cache.iamUsersMap[accessKey] = uinfo.Credentials
+	cache.updatedAt = time.Now()
+
 	return nil
 }
 
@@ -1644,6 +1702,7 @@ func (store *IAMStoreSys) AddServiceAccount(ctx context.Context, cred auth.Crede
 	}
 
 	cache.iamUsersMap[u.Credentials.AccessKey] = u.Credentials
+	cache.updatedAt = time.Now()
 
 	return nil
 }
@@ -1685,6 +1744,17 @@ func (store *IAMStoreSys) UpdateServiceAccount(ctx context.Context, accessKey st
 		return fmt.Errorf("unable to get svc acc claims: %v", err)
 	}
 
+	// Extracted session policy name string can be removed as its not useful
+	// at this point.
+	delete(m, sessionPolicyNameExtracted)
+
+	// sessionPolicy is nil and there is embedded policy attached we remove
+	// rembedded policy at that point.
+	if _, ok := m[iampolicy.SessionPolicyName]; ok && opts.sessionPolicy == nil {
+		delete(m, iampolicy.SessionPolicyName)
+		m[iamPolicyClaimNameSA()] = inheritedPolicyType
+	}
+
 	if opts.sessionPolicy != nil {
 		if err := opts.sessionPolicy.Validate(); err != nil {
 			return err
@@ -1701,7 +1771,7 @@ func (store *IAMStoreSys) UpdateServiceAccount(ctx context.Context, accessKey st
 
 		// Overwrite session policy claims.
 		m[iampolicy.SessionPolicyName] = base64.StdEncoding.EncodeToString(policyBuf)
-		m[iamPolicyClaimNameSA()] = "embedded-policy"
+		m[iamPolicyClaimNameSA()] = embeddedPolicyType
 	}
 
 	cr.SessionToken, err = auth.JWTSignWithAccessKey(accessKey, m, cr.SecretKey)
@@ -1715,6 +1785,7 @@ func (store *IAMStoreSys) UpdateServiceAccount(ctx context.Context, accessKey st
 	}
 
 	cache.iamUsersMap[u.Credentials.AccessKey] = u.Credentials
+	cache.updatedAt = time.Now()
 
 	return nil
 }
@@ -1724,14 +1795,29 @@ func (store *IAMStoreSys) ListTempAccounts(ctx context.Context, accessKey string
 	cache := store.rlock()
 	defer store.runlock()
 
+	userExists := false
 	var tempAccounts []auth.Credentials
 	for _, v := range cache.iamUsersMap {
-		if v.IsTemp() && v.ParentUser == accessKey {
-			// Hide secret key & session key here
-			v.SecretKey = ""
-			v.SessionToken = ""
-			tempAccounts = append(tempAccounts, v)
+		isDerived := false
+		if v.IsServiceAccount() || v.IsTemp() {
+			isDerived = true
 		}
+
+		if !isDerived && v.AccessKey == accessKey {
+			userExists = true
+		} else if isDerived && v.ParentUser == accessKey {
+			userExists = true
+			if v.IsTemp() {
+				// Hide secret key & session key here
+				v.SecretKey = ""
+				v.SessionToken = ""
+				tempAccounts = append(tempAccounts, v)
+			}
+		}
+	}
+
+	if !userExists {
+		return nil, errNoSuchUser
 	}
 
 	return tempAccounts, nil
@@ -1742,14 +1828,29 @@ func (store *IAMStoreSys) ListServiceAccounts(ctx context.Context, accessKey str
 	cache := store.rlock()
 	defer store.runlock()
 
+	userExists := false
 	var serviceAccounts []auth.Credentials
 	for _, v := range cache.iamUsersMap {
-		if v.IsServiceAccount() && v.ParentUser == accessKey {
-			// Hide secret key & session key here
-			v.SecretKey = ""
-			v.SessionToken = ""
-			serviceAccounts = append(serviceAccounts, v)
+		isDerived := false
+		if v.IsServiceAccount() || v.IsTemp() {
+			isDerived = true
 		}
+
+		if !isDerived && v.AccessKey == accessKey {
+			userExists = true
+		} else if isDerived && v.ParentUser == accessKey {
+			userExists = true
+			if v.IsServiceAccount() {
+				// Hide secret key & session key here
+				v.SecretKey = ""
+				v.SessionToken = ""
+				serviceAccounts = append(serviceAccounts, v)
+			}
+		}
+	}
+
+	if !userExists {
+		return nil, errNoSuchUser
 	}
 
 	return serviceAccounts, nil
@@ -1760,6 +1861,8 @@ func (store *IAMStoreSys) AddUser(ctx context.Context, accessKey string, ureq ma
 	cache := store.lock()
 	defer store.unlock()
 
+	cache.updatedAt = time.Now()
+
 	cr, ok := cache.iamUsersMap[accessKey]
 
 	// It is not possible to update an STS account.
@@ -1793,6 +1896,8 @@ func (store *IAMStoreSys) UpdateUserSecretKey(ctx context.Context, accessKey, se
 	cache := store.lock()
 	defer store.unlock()
 
+	cache.updatedAt = time.Now()
+
 	cred, ok := cache.iamUsersMap[accessKey]
 	if !ok {
 		return errNoSuchUser
@@ -1827,6 +1932,8 @@ func (store *IAMStoreSys) UpdateUserIdentity(ctx context.Context, cred auth.Cred
 	cache := store.lock()
 	defer store.unlock()
 
+	cache.updatedAt = time.Now()
+
 	userType := regUser
 	if cred.IsServiceAccount() {
 		userType = svcUser
@@ -1848,6 +1955,8 @@ func (store *IAMStoreSys) LoadUser(ctx context.Context, accessKey string) {
 	cache := store.lock()
 	defer store.unlock()
 
+	cache.updatedAt = time.Now()
+
 	_, found := cache.iamUsersMap[accessKey]
 	if !found {
 		store.loadUser(ctx, accessKey, regUser, cache.iamUsersMap)

cmd/iam.go

@@ -38,6 +38,8 @@ import (
 	"github.com/minio/minio/internal/arn"
 	"github.com/minio/minio/internal/auth"
 	"github.com/minio/minio/internal/color"
+	xldap "github.com/minio/minio/internal/config/identity/ldap"
+	"github.com/minio/minio/internal/config/identity/openid"
 	"github.com/minio/minio/internal/jwt"
 	"github.com/minio/minio/internal/logger"
 	iampolicy "github.com/minio/pkg/iam/policy"
@@ -63,6 +65,11 @@ const (
 	statusDisabled = "disabled"
 )
 
+const (
+	embeddedPolicyType  = "embedded-policy"
+	inheritedPolicyType = "inherited-policy"
+)
+
 // IAMSys - config system.
 type IAMSys struct {
 	// Need to keep them here to keep alignment - ref: https://golang.org/pkg/sync/atomic/#pkg-note-BUG
@@ -75,6 +82,8 @@ type IAMSys struct {
 	sync.Mutex
 
 	iamRefreshInterval time.Duration
+	ldapConfig         xldap.Config  // only valid if usersSysType is LDAPUsers
+	openIDConfig       openid.Config // only valid if OpenID is configured
 
 	usersSysType UsersSysType
 
@@ -159,7 +168,7 @@ func (sys *IAMSys) doIAMConfigMigration(ctx context.Context) error {
 
 // initStore initializes IAM stores
 func (sys *IAMSys) initStore(objAPI ObjectLayer, etcdClient *etcd.Client) {
-	if globalLDAPConfig.Enabled {
+	if sys.ldapConfig.Enabled {
 		sys.EnableLDAPSys()
 	}
 
@@ -219,6 +228,8 @@ func (sys *IAMSys) Init(ctx context.Context, objAPI ObjectLayer, etcdClient *etc
 	sys.Lock()
 	defer sys.Unlock()
 
+	sys.ldapConfig = globalLDAPConfig.Clone()
+	sys.openIDConfig = globalOpenIDConfig.Clone()
 	sys.iamRefreshInterval = iamRefreshInterval
 
 	// Initialize IAM store
@@ -303,30 +314,37 @@ func (sys *IAMSys) Init(ctx context.Context, objAPI ObjectLayer, etcdClient *etc
 		break
 	}
 
+	refreshInterval := sys.iamRefreshInterval
+
 	// Set up polling for expired accounts and credentials purging.
 	switch {
-	case globalOpenIDConfig.ProviderEnabled():
+	case sys.openIDConfig.ProviderEnabled():
 		go func() {
-			ticker := time.NewTicker(sys.iamRefreshInterval)
-			defer ticker.Stop()
+			timer := time.NewTimer(refreshInterval)
+			defer timer.Stop()
 			for {
 				select {
-				case <-ticker.C:
+				case <-timer.C:
 					sys.purgeExpiredCredentialsForExternalSSO(ctx)
+
+					timer.Reset(refreshInterval)
 				case <-ctx.Done():
 					return
 				}
 			}
 		}()
-	case globalLDAPConfig.Enabled:
+	case sys.ldapConfig.Enabled:
 		go func() {
-			ticker := time.NewTicker(sys.iamRefreshInterval)
-			defer ticker.Stop()
+			timer := time.NewTimer(refreshInterval)
+			defer timer.Stop()
+
 			for {
 				select {
-				case <-ticker.C:
+				case <-timer.C:
 					sys.purgeExpiredCredentialsForLDAP(ctx)
 					sys.updateGroupMembershipsForLDAP(ctx)
+
+					timer.Reset(refreshInterval)
 				case <-ctx.Done():
 					return
 				}
@@ -337,20 +355,18 @@ func (sys *IAMSys) Init(ctx context.Context, objAPI ObjectLayer, etcdClient *etc
 	// Start watching changes to storage.
 	go sys.watch(ctx)
 
-	// Load RoleARN
-	if rolePolicyMap := globalOpenIDConfig.GetRoleInfo(); rolePolicyMap != nil {
-		// Validate that policies associated with roles are defined.
-		for _, rolePolicies := range rolePolicyMap {
-			ps := newMappedPolicy(rolePolicies).toSlice()
-			numPolicies := len(ps)
-			validPolicies, _ := sys.store.FilterPolicies(rolePolicies, "")
-			numValidPolicies := len(strings.Split(validPolicies, ","))
-			if numPolicies != numValidPolicies {
-				logger.LogIf(ctx, fmt.Errorf("Some specified role policies (in %s) were not defined - role policies will not be enabled.", rolePolicies))
-				return
-			}
-		}
-		sys.rolesMap = rolePolicyMap
+	// Load RoleARNs
+	sys.rolesMap = make(map[arn.ARN]string)
+
+	// From OpenID
+	if riMap := globalOpenIDConfig.GetRoleInfo(); riMap != nil {
+		sys.validateAndAddRolePolicyMappings(ctx, riMap)
+	}
+
+	// From AuthN plugin if enabled.
+	if globalAuthNPlugin != nil {
+		riMap := globalAuthNPlugin.GetRoleInfo()
+		sys.validateAndAddRolePolicyMappings(ctx, riMap)
 	}
 
 	sys.printIAMRoles()
@@ -359,6 +375,32 @@ func (sys *IAMSys) Init(ctx context.Context, objAPI ObjectLayer, etcdClient *etc
 	logger.Info("Finished loading IAM sub-system (took %.1fs of %.1fs to load data).", now.Sub(iamLoadStart).Seconds(), now.Sub(iamInitStart).Seconds())
 }
 
+func (sys *IAMSys) validateAndAddRolePolicyMappings(ctx context.Context, m map[arn.ARN]string) {
+	// Validate that policies associated with roles are defined. If
+	// authZ plugin is set, role policies are just claims sent to
+	// the plugin and they need not exist.
+	//
+	// If some mapped policies do not exist, we print some error
+	// messages but continue any way - they can be fixed in the
+	// running server by creating the policies after start up.
+	for arn, rolePolicies := range m {
+		specifiedPoliciesSet := newMappedPolicy(rolePolicies).policySet()
+		validPolicies, _ := sys.store.FilterPolicies(rolePolicies, "")
+		knownPoliciesSet := newMappedPolicy(validPolicies).policySet()
+		unknownPoliciesSet := specifiedPoliciesSet.Difference(knownPoliciesSet)
+		if len(unknownPoliciesSet) > 0 {
+			if globalAuthZPlugin == nil {
+				// Print a warning that some policies mapped to a role are not defined.
+				errMsg := fmt.Errorf(
+					"The policies \"%s\" mapped to role ARN %s are not defined - this role may not work as expected.",
+					unknownPoliciesSet.ToSlice(), arn.String())
+				logger.LogIf(ctx, errMsg)
+			}
+		}
+		sys.rolesMap[arn] = rolePolicies
+	}
+}
+
 // Prints IAM role ARNs.
 func (sys *IAMSys) printIAMRoles() {
 	if len(sys.rolesMap) == 0 {
@@ -398,12 +440,12 @@ func (sys *IAMSys) watch(ctx context.Context) {
 
 	var maxRefreshDurationSecondsForLog float64 = 10
 
-	// Fall back to loading all items periodically
-	ticker := time.NewTicker(sys.iamRefreshInterval)
-	defer ticker.Stop()
+	// Load all items periodically
+	timer := time.NewTimer(sys.iamRefreshInterval)
+	defer timer.Stop()
 	for {
 		select {
-		case <-ticker.C:
+		case <-timer.C:
 			refreshStart := time.Now()
 			if err := sys.Load(ctx); err != nil {
 				logger.LogIf(ctx, fmt.Errorf("Failure in periodic refresh for IAM (took %.2fs): %v", time.Since(refreshStart).Seconds(), err))
@@ -415,6 +457,7 @@ func (sys *IAMSys) watch(ctx context.Context) {
 				}
 			}
 
+			timer.Reset(sys.iamRefreshInterval)
 		case <-ctx.Done():
 			return
 		}
@@ -676,7 +719,7 @@ func (sys *IAMSys) SetTempUser(ctx context.Context, accessKey string, cred auth.
 		return errServerNotInitialized
 	}
 
-	if globalPolicyOPA != nil {
+	if newGlobalAuthZPluginFn() != nil {
 		// If OPA is set, we do not need to set a policy mapping.
 		policyName = ""
 	}
@@ -866,9 +909,9 @@ func (sys *IAMSys) NewServiceAccount(ctx context.Context, parentUser string, gro
 
 	if len(policyBuf) > 0 {
 		m[iampolicy.SessionPolicyName] = base64.StdEncoding.EncodeToString(policyBuf)
-		m[iamPolicyClaimNameSA()] = "embedded-policy"
+		m[iamPolicyClaimNameSA()] = embeddedPolicyType
 	} else {
-		m[iamPolicyClaimNameSA()] = "inherited-policy"
+		m[iamPolicyClaimNameSA()] = inheritedPolicyType
 	}
 
 	// Add all the necessary claims for the service accounts.
@@ -989,7 +1032,7 @@ func (sys *IAMSys) getServiceAccount(ctx context.Context, accessKey string) (aut
 	}
 	pt, ptok := jwtClaims.Lookup(iamPolicyClaimNameSA())
 	sp, spok := jwtClaims.Lookup(iampolicy.SessionPolicyName)
-	if ptok && spok && pt == "embedded-policy" {
+	if ptok && spok && pt == embeddedPolicyType {
 		policyBytes, err := base64.StdEncoding.DecodeString(sp)
 		if err != nil {
 			return auth.Credentials{}, nil, err
@@ -1127,7 +1170,7 @@ func (sys *IAMSys) purgeExpiredCredentialsForExternalSSO(ctx context.Context) {
 			continue
 		}
 		roleArn = roleArns[0]
-		u, err := globalOpenIDConfig.LookupUser(roleArn, puInfo.subClaimValue)
+		u, err := sys.openIDConfig.LookupUser(roleArn, puInfo.subClaimValue)
 		if err != nil {
 			logger.LogIf(GlobalContext, err)
 			continue
@@ -1149,14 +1192,14 @@ func (sys *IAMSys) purgeExpiredCredentialsForLDAP(ctx context.Context) {
 	parentUsers := sys.store.GetAllParentUsers()
 	var allDistNames []string
 	for parentUser := range parentUsers {
-		if !globalLDAPConfig.IsLDAPUserDN(parentUser) {
+		if !sys.ldapConfig.IsLDAPUserDN(parentUser) {
 			continue
 		}
 
 		allDistNames = append(allDistNames, parentUser)
 	}
 
-	expiredUsers, err := globalLDAPConfig.GetNonEligibleUserDistNames(allDistNames)
+	expiredUsers, err := sys.ldapConfig.GetNonEligibleUserDistNames(allDistNames)
 	if err != nil {
 		// Log and return on error - perhaps it'll work the next time.
 		logger.LogIf(GlobalContext, err)
@@ -1178,7 +1221,7 @@ func (sys *IAMSys) updateGroupMembershipsForLDAP(ctx context.Context) {
 	// DN to ldap username mapping for each LDAP user
 	parentUserToLDAPUsernameMap := make(map[string]string)
 	for _, cred := range allCreds {
-		if !globalLDAPConfig.IsLDAPUserDN(cred.ParentUser) {
+		if !sys.ldapConfig.IsLDAPUserDN(cred.ParentUser) {
 			continue
 		}
 		// Check if this is the first time we are
@@ -1226,7 +1269,7 @@ func (sys *IAMSys) updateGroupMembershipsForLDAP(ctx context.Context) {
 	}
 
 	// 2. Query LDAP server for groups of the LDAP users collected.
-	updatedGroups, err := globalLDAPConfig.LookupGroupMemberships(parentUsers, parentUserToLDAPUsernameMap)
+	updatedGroups, err := sys.ldapConfig.LookupGroupMemberships(parentUsers, parentUserToLDAPUsernameMap)
 	if err != nil {
 		// Log and return on error - perhaps it'll work the next time.
 		logger.LogIf(GlobalContext, err)
@@ -1417,6 +1460,8 @@ func (sys *IAMSys) PolicyDBGet(name string, isGroup bool, groups ...string) ([]s
 	return sys.store.PolicyDBGet(name, isGroup, groups...)
 }
 
+const sessionPolicyNameExtracted = iampolicy.SessionPolicyName + "-extracted"
+
 // IsAllowedServiceAccount - checks if the given service account is allowed to perform
 // actions. The permission of the parent user is checked first
 func (sys *IAMSys) IsAllowedServiceAccount(args iampolicy.Args, parentUser string) bool {
@@ -1493,12 +1538,12 @@ func (sys *IAMSys) IsAllowedServiceAccount(args iampolicy.Args, parentUser strin
 		return false
 	}
 
-	if saPolicyClaimStr == "inherited-policy" {
+	if saPolicyClaimStr == inheritedPolicyType {
 		return combinedPolicy.IsAllowed(parentArgs)
 	}
 
 	// Now check if we have a sessionPolicy.
-	spolicy, ok := args.Claims[iampolicy.SessionPolicyName]
+	spolicy, ok := args.Claims[sessionPolicyNameExtracted]
 	if !ok {
 		return false
 	}
@@ -1530,54 +1575,10 @@ func (sys *IAMSys) IsAllowedServiceAccount(args iampolicy.Args, parentUser strin
 	return combinedPolicy.IsAllowed(parentArgs) && subPolicy.IsAllowed(parentArgs)
 }
 
-// IsAllowedLDAPSTS - checks for LDAP specific claims and values
-func (sys *IAMSys) IsAllowedLDAPSTS(args iampolicy.Args, parentUser string) bool {
-	// parentUser value must match the ldap user in the claim.
-	if parentInClaimIface, ok := args.Claims[ldapUser]; !ok {
-		// no ldapUser claim present reject it.
-		return false
-	} else if parentInClaim, ok := parentInClaimIface.(string); !ok {
-		// not the right type, reject it.
-		return false
-	} else if parentInClaim != parentUser {
-		// ldap claim has been modified maliciously reject it.
-		return false
-	}
-
-	// Check policy for this LDAP user.
-	ldapPolicies, err := sys.PolicyDBGet(parentUser, false, args.Groups...)
-	if err != nil {
-		return false
-	}
-
-	if len(ldapPolicies) == 0 {
-		return false
-	}
-
-	// Policies were found, evaluate all of them.
-	availablePoliciesStr, combinedPolicy := sys.store.FilterPolicies(strings.Join(ldapPolicies, ","), "")
-	if availablePoliciesStr == "" {
-		return false
-	}
-
-	hasSessionPolicy, isAllowedSP := isAllowedBySessionPolicy(args)
-	if hasSessionPolicy {
-		return isAllowedSP && combinedPolicy.IsAllowed(args)
-	}
-
-	return combinedPolicy.IsAllowed(args)
-}
-
 // IsAllowedSTS is meant for STS based temporary credentials,
 // which implements claims validation and verification other than
 // applying policies.
 func (sys *IAMSys) IsAllowedSTS(args iampolicy.Args, parentUser string) bool {
-	// If it is an LDAP request, check that user and group
-	// policies allow the request.
-	if sys.usersSysType == LDAPUsersSysType {
-		return sys.IsAllowedLDAPSTS(args, parentUser)
-	}
-
 	var policies []string
 	roleArn := args.GetRoleArn()
 	if roleArn != "" {
@@ -1612,6 +1613,11 @@ func (sys *IAMSys) IsAllowedSTS(args iampolicy.Args, parentUser string) bool {
 		}
 	}
 
+	// Defensive code: Do not allow any operation if no policy is found in the session token
+	if len(policies) == 0 {
+		return false
+	}
+
 	combinedPolicy, err := sys.store.GetPolicy(strings.Join(policies, ","))
 	if err == errNoSuchPolicy {
 		for _, pname := range policies {
@@ -1646,7 +1652,7 @@ func isAllowedBySessionPolicy(args iampolicy.Args) (hasSessionPolicy bool, isAll
 	isAllowed = false
 
 	// Now check if we have a sessionPolicy.
-	spolicy, ok := args.Claims[iampolicy.SessionPolicyName]
+	spolicy, ok := args.Claims[sessionPolicyNameExtracted]
 	if !ok {
 		return
 	}
@@ -1686,8 +1692,8 @@ func (sys *IAMSys) GetCombinedPolicy(policies ...string) iampolicy.Policy {
 // IsAllowed - checks given policy args is allowed to continue the Rest API.
 func (sys *IAMSys) IsAllowed(args iampolicy.Args) bool {
 	// If opa is configured, use OPA always.
-	if globalPolicyOPA != nil {
-		ok, err := globalPolicyOPA.IsAllowed(args)
+	if authz := newGlobalAuthZPluginFn(); authz != nil {
+		ok, err := authz.IsAllowed(args)
 		if err != nil {
 			logger.LogIf(GlobalContext, err)
 		}

cmd/listen-notification-handlers.go

@@ -127,7 +127,7 @@ func (api objectAPIHandlers) ListenNotificationHandler(w http.ResponseWriter, r
 
 	peers, _ := newPeerRestClients(globalEndpoints)
 
-	globalHTTPListen.Subscribe(listenCh, ctx.Done(), func(evI interface{}) bool {
+	listenFn := func(evI interface{}) bool {
 		ev, ok := evI.(event.Event)
 		if !ok {
 			return false
@@ -138,7 +138,13 @@ func (api objectAPIHandlers) ListenNotificationHandler(w http.ResponseWriter, r
 			}
 		}
 		return rulesMap.MatchSimple(ev.EventName, ev.S3.Object.Key)
-	})
+	}
+
+	err := globalHTTPListen.Subscribe(listenCh, ctx.Done(), listenFn)
+	if err != nil {
+		writeErrorResponse(ctx, w, errorCodes.ToAPIErr(ErrSlowDown), r.URL)
+		return
+	}
 
 	if bucketName != "" {
 		values.Set(peerRESTListenBucket, bucketName)

cmd/local-locker_test.go

@@ -344,3 +344,96 @@ func Test_localLocker_expireOldLocksExpire(t *testing.T) {
 		})
 	}
 }
+
+func Test_localLocker_RUnlock(t *testing.T) {
+	rng := rand.New(rand.NewSource(0))
+	// Numbers of unique locks
+	for _, locks := range []int{1, 100, 1000, 1e6} {
+		if testing.Short() && locks > 100 {
+			continue
+		}
+		t.Run(fmt.Sprintf("%d-locks", locks), func(t *testing.T) {
+			// Number of readers per lock...
+			for _, readers := range []int{1, 10, 100} {
+				if locks > 1000 && readers > 1 {
+					continue
+				}
+				if testing.Short() && readers > 10 {
+					continue
+				}
+				t.Run(fmt.Sprintf("%d-read", readers), func(t *testing.T) {
+					l := newLocker()
+					for i := 0; i < locks; i++ {
+						var tmp [16]byte
+						rng.Read(tmp[:])
+						res := []string{hex.EncodeToString(tmp[:])}
+
+						for i := 0; i < readers; i++ {
+							rng.Read(tmp[:])
+							ok, err := l.RLock(context.Background(), dsync.LockArgs{
+								UID:       uuid.NewString(),
+								Resources: res,
+								Source:    hex.EncodeToString(tmp[:8]),
+								Owner:     hex.EncodeToString(tmp[8:]),
+								Quorum:    0,
+							})
+							if !ok || err != nil {
+								t.Fatal("failed:", err, ok)
+							}
+						}
+					}
+
+					// Expire 50%
+					toUnLock := make([]dsync.LockArgs, 0, locks*readers)
+					for k, v := range l.lockMap {
+						for _, lock := range v {
+							if rng.Intn(2) == 0 {
+								toUnLock = append(toUnLock, dsync.LockArgs{Resources: []string{k}, UID: lock.UID})
+							}
+						}
+					}
+					start := time.Now()
+					for _, lock := range toUnLock {
+						ok, err := l.ForceUnlock(context.Background(), lock)
+						if err != nil || !ok {
+							t.Fatal(err)
+						}
+					}
+					t.Logf("Expire 50%% took: %v. Left: %d/%d", time.Since(start).Round(time.Millisecond), len(l.lockUID), len(l.lockMap))
+
+					if len(l.lockUID) == locks*readers {
+						t.Fatalf("objects uids all remain, unlikely")
+					}
+					if len(l.lockMap) == 0 && locks > 10 {
+						t.Fatalf("objects all deleted, 0 remains")
+					}
+					if len(l.lockUID) != locks*readers-len(toUnLock) {
+						t.Fatalf("want %d objects uids all deleted, %d remains", len(l.lockUID), locks*readers-len(toUnLock))
+					}
+
+					toUnLock = toUnLock[:0]
+					for k, v := range l.lockMap {
+						for _, lock := range v {
+							toUnLock = append(toUnLock, dsync.LockArgs{Resources: []string{k}, UID: lock.UID, Owner: lock.Owner})
+						}
+					}
+					start = time.Now()
+					for _, lock := range toUnLock {
+						ok, err := l.RUnlock(nil, lock)
+						if err != nil || !ok {
+							t.Fatal(err)
+						}
+					}
+					t.Logf("Expire rest took: %v. Left: %d/%d", time.Since(start).Round(time.Millisecond), len(l.lockUID), len(l.lockMap))
+
+					if len(l.lockMap) != 0 {
+						t.Fatalf("objects not deleted, want %d != got %d", 0, len(l.lockMap))
+					}
+					if len(l.lockUID) != 0 {
+						t.Fatalf("objects not deleted, want %d != got %d", 0, len(l.lockUID))
+					}
+				})
+			}
+		})
+	}
+}

cmd/lock-rest-server.go

@@ -209,22 +209,7 @@ func (l *lockRESTServer) ForceUnlockHandler(w http.ResponseWriter, r *http.Reque
 // lockMaintenance loops over all locks and discards locks
 // that have not been refreshed for some time.
 func lockMaintenance(ctx context.Context) {
-	// Wait until the object API is ready
-	// no need to start the lock maintenance
-	// if ObjectAPI is not initialized.
-
-	var objAPI ObjectLayer
-
-	for {
-		objAPI = newObjectLayerFn()
-		if objAPI == nil {
-			time.Sleep(time.Second)
-			continue
-		}
-		break
-	}
-
-	if _, ok := objAPI.(*erasureServerPools); !ok {
+	if !globalIsDistErasure {
 		return
 	}
 
@@ -239,10 +224,10 @@ func lockMaintenance(ctx context.Context) {
 		case <-ctx.Done():
 			return
 		case <-lkTimer.C:
+			globalLockServer.expireOldLocks(lockValidityDuration)
+
 			// Reset the timer for next cycle.
 			lkTimer.Reset(lockMaintenanceInterval)
-
-			globalLockServer.expireOldLocks(lockValidityDuration)
 		}
 	}
 }

cmd/main.go

@@ -20,6 +20,7 @@ package cmd
 import (
 	"os"
 	"path/filepath"
+	"runtime"
 	"sort"
 
 	"github.com/minio/cli"
@@ -68,6 +69,11 @@ var GlobalFlags = []cli.Flag{
 	},
 }
 
+var versionFlag = cli.BoolFlag{
+	Name:  "version, v",
+	Usage: "print version information",
+}
+
 // Help template for minio.
 var minioHelpTemplate = `NAME:
   {{.Name}} - {{.Usage}}
@@ -132,14 +138,19 @@ func newApp(name string) *cli.App {
 		Usage: "show help",
 	}
 
+	topLevelFlags := make([]cli.Flag, len(GlobalFlags)+1)
+	copy(topLevelFlags, GlobalFlags)
+	topLevelFlags[len(GlobalFlags)] = versionFlag
+
 	app := cli.NewApp()
 	app.Name = name
 	app.Author = "MinIO, Inc."
-	app.Version = ReleaseTag
+	app.Action = versionAndHelpAction
 	app.Usage = "High Performance Object Storage"
 	app.Description = `Build high performance data infrastructure for machine learning, analytics and application data workloads with MinIO`
-	app.Flags = GlobalFlags
+	app.Flags = topLevelFlags
 	app.HideHelpCommand = true // Hide `help, h` command, we already have `minio --help`.
+	app.HideVersion = true
 	app.Commands = commands
 	app.CustomAppHelpTemplate = minioHelpTemplate
 	app.CommandNotFound = func(ctx *cli.Context, command string) {
@@ -159,6 +170,18 @@ func newApp(name string) *cli.App {
 	return app
 }
 
+func versionAndHelpAction(ctx *cli.Context) {
+	if ctx.IsSet("version") {
+		console.Printf("%s version %s\n", ctx.App.Name, ReleaseTag)
+		console.Printf("commit: %s\n", CommitID)
+		console.Printf("go version: %s\n", runtime.Version())
+
+		return
+	}
+
+	cli.ShowAppHelpAndExit(ctx, 1)
+}
+
 // Main main for minio server.
 func Main(args []string) {
 	// Set the minio app name.

cmd/metacache-bucket.go

@@ -51,10 +51,12 @@ func newBucketMetacache(bucket string, cleanup bool) *bucketMetacache {
 	if cleanup {
 		// Recursively delete all caches.
 		objAPI := newObjectLayerFn()
-		ez, ok := objAPI.(*erasureServerPools)
-		if ok {
-			ctx := context.Background()
-			ez.renameAll(ctx, minioMetaBucket, metacachePrefixForID(bucket, slashSeparator))
+		if objAPI != nil {
+			ez, ok := objAPI.(renameAllStorager)
+			if ok {
+				ctx := context.Background()
+				ez.renameAll(ctx, minioMetaBucket, metacachePrefixForID(bucket, slashSeparator))
+			}
 		}
 	}
 	return &bucketMetacache{
@@ -207,9 +209,15 @@ func (b *bucketMetacache) cloneCaches() (map[string]metacache, map[string][]stri
 // Deletes are performed concurrently.
 func (b *bucketMetacache) deleteAll() {
 	ctx := context.Background()
-	ez, ok := newObjectLayerFn().(*erasureServerPools)
+
+	objAPI := newObjectLayerFn()
+	if objAPI == nil {
+		return
+	}
+
+	ez, ok := objAPI.(renameAllStorager)
 	if !ok {
-		logger.LogIf(ctx, errors.New("bucketMetacache: expected objAPI to be *erasurePools"))
+		logger.LogIf(ctx, errors.New("bucketMetacache: expected objAPI to be 'renameAllStorager'"))
 		return
 	}
 

cmd/metacache-entries.go

@@ -168,8 +168,10 @@ func (e *metaCacheEntry) isLatestDeletemarker() bool {
 	if !isXL2V1Format(e.metadata) {
 		return false
 	}
-	if meta, _ := isIndexedMetaV2(e.metadata); meta != nil {
+	if meta, _, err := isIndexedMetaV2(e.metadata); meta != nil {
 		return meta.IsLatestDeleteMarker()
+	} else if err != nil {
+		return true
 	}
 	// Fall back...
 	xlMeta, err := e.xlmeta()
@@ -438,6 +440,8 @@ func (m metaCacheEntriesSorted) shallowClone() metaCacheEntriesSorted {
 func (m *metaCacheEntriesSorted) fileInfoVersions(bucket, prefix, delimiter, afterV string) (versions []ObjectInfo) {
 	versions = make([]ObjectInfo, 0, m.len())
 	prevPrefix := ""
+	vcfg, _ := globalBucketVersioningSys.Get(bucket)
+
 	for _, entry := range m.o {
 		if entry.isObject() {
 			if delimiter != "" {
@@ -473,7 +477,8 @@ func (m *metaCacheEntriesSorted) fileInfoVersions(bucket, prefix, delimiter, aft
 			}
 
 			for _, version := range fiVersions {
-				versions = append(versions, version.ToObjectInfo(bucket, entry.name))
+				versioned := vcfg != nil && vcfg.Versioned(entry.name)
+				versions = append(versions, version.ToObjectInfo(bucket, entry.name, versioned))
 			}
 
 			continue
@@ -509,6 +514,9 @@ func (m *metaCacheEntriesSorted) fileInfoVersions(bucket, prefix, delimiter, aft
 func (m *metaCacheEntriesSorted) fileInfos(bucket, prefix, delimiter string) (objects []ObjectInfo) {
 	objects = make([]ObjectInfo, 0, m.len())
 	prevPrefix := ""
+
+	vcfg, _ := globalBucketVersioningSys.Get(bucket)
+
 	for _, entry := range m.o {
 		if entry.isObject() {
 			if delimiter != "" {
@@ -531,7 +539,8 @@ func (m *metaCacheEntriesSorted) fileInfos(bucket, prefix, delimiter string) (ob
 
 			fi, err := entry.fileInfo(bucket)
 			if err == nil {
-				objects = append(objects, fi.ToObjectInfo(bucket, entry.name))
+				versioned := vcfg != nil && vcfg.Versioned(entry.name)
+				objects = append(objects, fi.ToObjectInfo(bucket, entry.name, versioned))
 			}
 			continue
 		}
@@ -686,8 +695,12 @@ func mergeEntryChannels(ctx context.Context, in []chan metaCacheEntry, out chan<
 			}
 		}
 		if best.name > last {
-			out <- *best
-			last = best.name
+			select {
+			case <-ctxDone:
+				return ctx.Err()
+			case out <- *best:
+				last = best.name
+			}
 		} else if serverDebugLog {
 			console.Debugln("mergeEntryChannels: discarding duplicate", best.name, "<=", last)
 		}

cmd/metacache-manager.go

@@ -56,7 +56,7 @@ func (m *metacacheManager) initManager() {
 			objAPI = newObjectLayerFn()
 		}
 
-		if !globalIsErasure {
+		if globalIsGateway {
 			return
 		}
 

cmd/metacache-server-pool.go

@@ -259,6 +259,286 @@ func (z *erasureServerPools) listPath(ctx context.Context, o *listPathOptions) (
 	return entries, nil
 }
 
+// listPath will return the requested entries.
+// If no more entries are in the listing io.EOF is returned,
+// otherwise nil or an unexpected error is returned.
+// The listPathOptions given will be checked and modified internally.
+// Required important fields are Bucket, Prefix, Separator.
+// Other important fields are Limit, Marker.
+// List ID always derived from the Marker.
+func (es *erasureSingle) listPath(ctx context.Context, o *listPathOptions) (entries metaCacheEntriesSorted, err error) {
+	if err := checkListObjsArgs(ctx, o.Bucket, o.Prefix, o.Marker, es); err != nil {
+		return entries, err
+	}
+
+	// Marker is set validate pre-condition.
+	if o.Marker != "" && o.Prefix != "" {
+		// Marker not common with prefix is not implemented. Send an empty response
+		if !HasPrefix(o.Marker, o.Prefix) {
+			return entries, io.EOF
+		}
+	}
+
+	// With max keys of zero we have reached eof, return right here.
+	if o.Limit == 0 {
+		return entries, io.EOF
+	}
+
+	// For delimiter and prefix as '/' we do not list anything at all
+	// along // with the prefix. On a flat namespace with 'prefix'
+	// as '/' we don't have any entries, since all the keys are
+	// of form 'keyName/...'
+	if strings.HasPrefix(o.Prefix, SlashSeparator) {
+		return entries, io.EOF
+	}
+
+	// If delimiter is slashSeparator we must return directories of
+	// the non-recursive scan unless explicitly requested.
+	o.IncludeDirectories = o.Separator == slashSeparator
+	if (o.Separator == slashSeparator || o.Separator == "") && !o.Recursive {
+		o.Recursive = o.Separator != slashSeparator
+		o.Separator = slashSeparator
+	} else {
+		// Default is recursive, if delimiter is set then list non recursive.
+		o.Recursive = true
+	}
+
+	// Decode and get the optional list id from the marker.
+	o.parseMarker()
+	o.BaseDir = baseDirFromPrefix(o.Prefix)
+	o.Transient = o.Transient || isReservedOrInvalidBucket(o.Bucket, false)
+	o.SetFilter()
+	if o.Transient {
+		o.Create = false
+	}
+
+	// We have 2 cases:
+	// 1) Cold listing, just list.
+	// 2) Returning, but with no id. Start async listing.
+	// 3) Returning, with ID, stream from list.
+	//
+	// If we don't have a list id we must ask the server if it has a cache or create a new.
+	if o.ID != "" && !o.Transient {
+		// Create or ping with handout...
+		rpc := globalNotificationSys.restClientFromHash(pathJoin(o.Bucket, o.Prefix))
+		ctx, cancel := context.WithTimeout(ctx, 5*time.Second)
+		defer cancel()
+		var c *metacache
+		if rpc == nil {
+			resp := localMetacacheMgr.getBucket(ctx, o.Bucket).findCache(*o)
+			c = &resp
+		} else {
+			c, err = rpc.GetMetacacheListing(ctx, *o)
+		}
+		if err != nil {
+			if errors.Is(err, context.Canceled) {
+				// Context is canceled, return at once.
+				// request canceled, no entries to return
+				return entries, io.EOF
+			}
+			if !errors.Is(err, context.DeadlineExceeded) {
+				o.debugln("listPath: got error", err)
+			}
+			o.Transient = true
+			o.Create = false
+			o.ID = mustGetUUID()
+		} else {
+			if c.fileNotFound {
+				// No cache found, no entries found.
+				return entries, io.EOF
+			}
+			if c.status == scanStateError || c.status == scanStateNone {
+				o.ID = ""
+				o.Create = false
+				o.debugln("scan status", c.status, " - waiting a roundtrip to create")
+			} else {
+				// Continue listing
+				o.ID = c.id
+				go func(meta metacache) {
+					// Continuously update while we wait.
+					t := time.NewTicker(metacacheMaxClientWait / 10)
+					defer t.Stop()
+					select {
+					case <-ctx.Done():
+						// Request is done, stop updating.
+						return
+					case <-t.C:
+						meta.lastHandout = time.Now()
+						if rpc == nil {
+							meta, _ = localMetacacheMgr.updateCacheEntry(meta)
+						}
+						meta, _ = rpc.UpdateMetacacheListing(ctx, meta)
+					}
+				}(*c)
+			}
+		}
+
+		// We have an existing list ID, continue streaming.
+		if o.Create {
+			o.debugln("Creating", o)
+			entries, err = es.listAndSave(ctx, o)
+			if err == nil || err == io.EOF {
+				return entries, err
+			}
+			entries.truncate(0)
+		} else {
+			o.debugln("Resuming", o)
+			entries, err = es.streamMetadataParts(ctx, *o)
+			entries.reuse = true // We read from stream and are not sharing results.
+			if err == nil {
+				return entries, nil
+			}
+		}
+		if IsErr(err, []error{
+			nil,
+			context.Canceled,
+			context.DeadlineExceeded,
+			// io.EOF is expected and should be returned but no need to log it.
+			io.EOF,
+		}...) {
+			// Expected good errors we don't need to return error.
+			return entries, err
+		}
+		entries.truncate(0)
+		go func() {
+			rpc := globalNotificationSys.restClientFromHash(pathJoin(o.Bucket, o.Prefix))
+			if rpc != nil {
+				ctx, cancel := context.WithTimeout(GlobalContext, 5*time.Second)
+				defer cancel()
+				c, err := rpc.GetMetacacheListing(ctx, *o)
+				if err == nil {
+					c.error = "no longer used"
+					c.status = scanStateError
+					rpc.UpdateMetacacheListing(ctx, *c)
+				}
+			}
+		}()
+		o.ID = ""
+
+		if err != nil {
+			logger.LogIf(ctx, fmt.Errorf("Resuming listing from drives failed %w, proceeding to do raw listing", err))
+		}
+	}
+
+	// Do listing in-place.
+	// Create output for our results.
+	// Create filter for results.
+	o.debugln("Raw List", o)
+	filterCh := make(chan metaCacheEntry, o.Limit)
+	listCtx, cancelList := context.WithCancel(ctx)
+	filteredResults := o.gatherResults(listCtx, filterCh)
+	var wg sync.WaitGroup
+	wg.Add(1)
+	var listErr error
+
+	go func(o listPathOptions) {
+		defer wg.Done()
+		o.Limit = 0
+		listErr = es.listMerged(listCtx, o, filterCh)
+		o.debugln("listMerged returned with", listErr)
+	}(*o)
+
+	entries, err = filteredResults()
+	cancelList()
+	wg.Wait()
+	if listErr != nil && !errors.Is(listErr, context.Canceled) {
+		return entries, listErr
+	}
+	entries.reuse = true
+	truncated := entries.len() > o.Limit || err == nil
+	entries.truncate(o.Limit)
+	if !o.Transient && truncated {
+		if o.ID == "" {
+			entries.listID = mustGetUUID()
+		} else {
+			entries.listID = o.ID
+		}
+	}
+	if !truncated {
+		return entries, io.EOF
+	}
+	return entries, nil
+}
+
+// listMerged will list across all sets and return a merged results stream.
+// The result channel is closed when no more results are expected.
+func (es *erasureSingle) listMerged(ctx context.Context, o listPathOptions, results chan<- metaCacheEntry) error {
+	var mu sync.Mutex
+	var wg sync.WaitGroup
+	var listErr error
+	var inputs []chan metaCacheEntry
+
+	innerResults := make(chan metaCacheEntry, 100)
+	inputs = append(inputs, innerResults)
+
+	mu.Lock()
+	listCtx, cancelList := context.WithCancel(ctx)
+	defer cancelList()
+
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		err := es.listPathInner(listCtx, o, innerResults)
+		mu.Lock()
+		defer mu.Unlock()
+		listErr = err
+	}()
+	mu.Unlock()
+
+	// Do lifecycle filtering.
+	if o.Lifecycle != nil {
+		filterIn := make(chan metaCacheEntry, 10)
+		go filterLifeCycle(ctx, o.Bucket, *o.Lifecycle, o.Retention, filterIn, results)
+		// Replace results.
+		results = filterIn
+	}
+
+	// Gather results to a single channel.
+	err := mergeEntryChannels(ctx, inputs, results, func(existing, other *metaCacheEntry) (replace bool) {
+		// Pick object over directory
+		if existing.isDir() && !other.isDir() {
+			return true
+		}
+		if !existing.isDir() && other.isDir() {
+			return false
+		}
+		eMeta, err := existing.xlmeta()
+		if err != nil {
+			return true
+		}
+		oMeta, err := other.xlmeta()
+		if err != nil {
+			return false
+		}
+		// Replace if modtime is newer
+		if !oMeta.latestModtime().Equal(oMeta.latestModtime()) {
+			return oMeta.latestModtime().After(eMeta.latestModtime())
+		}
+		// Use NumVersions as a final tiebreaker.
+		return len(oMeta.versions) > len(eMeta.versions)
+	})
+
+	cancelList()
+	wg.Wait()
+	if err != nil {
+		return err
+	}
+	if listErr != nil {
+		if contextCanceled(ctx) {
+			return nil
+		}
+		if listErr.Error() == io.EOF.Error() {
+			return nil
+		}
+		logger.LogIf(ctx, listErr)
+		return listErr
+	}
+	if contextCanceled(ctx) {
+		return ctx.Err()
+	}
+	return nil
+}
+
 // listMerged will list across all sets and return a merged results stream.
 // The result channel is closed when no more results are expected.
 func (z *erasureServerPools) listMerged(ctx context.Context, o listPathOptions, results chan<- metaCacheEntry) error {
@@ -274,11 +554,11 @@ func (z *erasureServerPools) listMerged(ctx context.Context, o listPathOptions,
 	for _, pool := range z.serverPools {
 		for _, set := range pool.sets {
 			wg.Add(1)
-			results := make(chan metaCacheEntry, 100)
-			inputs = append(inputs, results)
+			innerResults := make(chan metaCacheEntry, 100)
+			inputs = append(inputs, innerResults)
 			go func(i int, set *erasureObjects) {
 				defer wg.Done()
-				err := set.listPath(listCtx, o, results)
+				err := set.listPath(listCtx, o, innerResults)
 				mu.Lock()
 				defer mu.Unlock()
 				if err == nil {
@@ -329,6 +609,7 @@ func (z *erasureServerPools) listMerged(ctx context.Context, o listPathOptions,
 	if err != nil {
 		return err
 	}
+
 	if contextCanceled(ctx) {
 		return ctx.Err()
 	}
@@ -362,6 +643,8 @@ func (z *erasureServerPools) listMerged(ctx context.Context, o listPathOptions,
 // function closes 'out' and exits.
 func filterLifeCycle(ctx context.Context, bucket string, lc lifecycle.Lifecycle, lr lock.Retention, in <-chan metaCacheEntry, out chan<- metaCacheEntry) {
 	defer close(out)
+
+	vcfg, _ := globalBucketVersioningSys.Get(bucket)
 	for {
 		var obj metaCacheEntry
 		var ok bool
@@ -378,7 +661,10 @@ func filterLifeCycle(ctx context.Context, bucket string, lc lifecycle.Lifecycle,
 		if err != nil {
 			continue
 		}
-		objInfo := fi.ToObjectInfo(bucket, obj.name)
+
+		versioned := vcfg != nil && vcfg.Versioned(obj.name)
+
+		objInfo := fi.ToObjectInfo(bucket, obj.name, versioned)
 		action := evalActionFromLifecycle(ctx, lc, lr, objInfo, false)
 		switch action {
 		case lifecycle.DeleteVersionAction, lifecycle.DeleteAction:
@@ -395,6 +681,73 @@ func filterLifeCycle(ctx context.Context, bucket string, lc lifecycle.Lifecycle,
 	}
 }
 
+func (es *erasureSingle) listAndSave(ctx context.Context, o *listPathOptions) (entries metaCacheEntriesSorted, err error) {
+	// Use ID as the object name...
+	o.pool = 0
+	o.set = 0
+	saver := es
+
+	// Disconnect from call above, but cancel on exit.
+	listCtx, cancel := context.WithCancel(GlobalContext)
+	saveCh := make(chan metaCacheEntry, metacacheBlockSize)
+	inCh := make(chan metaCacheEntry, metacacheBlockSize)
+	outCh := make(chan metaCacheEntry, o.Limit)
+
+	filteredResults := o.gatherResults(ctx, outCh)
+
+	mc := o.newMetacache()
+	meta := metaCacheRPC{meta: &mc, cancel: cancel, rpc: globalNotificationSys.restClientFromHash(pathJoin(o.Bucket, o.Prefix)), o: *o}
+
+	// Save listing...
+	go func() {
+		if err := saver.saveMetaCacheStream(listCtx, &meta, saveCh); err != nil {
+			meta.setErr(err.Error())
+		}
+		cancel()
+	}()
+
+	// Do listing...
+	go func(o listPathOptions) {
+		err := es.listMerged(listCtx, o, inCh)
+		if err != nil {
+			meta.setErr(err.Error())
+		}
+		o.debugln("listAndSave: listing", o.ID, "finished with ", err)
+	}(*o)
+
+	// Keep track of when we return since we no longer have to send entries to output.
+	var funcReturned bool
+	var funcReturnedMu sync.Mutex
+	defer func() {
+		funcReturnedMu.Lock()
+		funcReturned = true
+		funcReturnedMu.Unlock()
+	}()
+	// Write listing to results and saver.
+	go func() {
+		var returned bool
+		for entry := range inCh {
+			if !returned {
+				funcReturnedMu.Lock()
+				returned = funcReturned
+				funcReturnedMu.Unlock()
+				outCh <- entry
+				if returned {
+					close(outCh)
+				}
+			}
+			entry.reusable = returned
+			saveCh <- entry
+		}
+		if !returned {
+			close(outCh)
+		}
+		close(saveCh)
+	}()
+
+	return filteredResults()
+}
+
 func (z *erasureServerPools) listAndSave(ctx context.Context, o *listPathOptions) (entries metaCacheEntriesSorted, err error) {
 	// Use ID as the object name...
 	o.pool = z.getAvailablePoolIdx(ctx, minioMetaBucket, o.ID, 10<<20)

cmd/metacache-set.go

@@ -543,6 +543,170 @@ func (er *erasureObjects) streamMetadataParts(ctx context.Context, o listPathOpt
 	}
 }
 
+func (es *erasureSingle) streamMetadataParts(ctx context.Context, o listPathOptions) (entries metaCacheEntriesSorted, err error) {
+	retries := 0
+	rpc := globalNotificationSys.restClientFromHash(pathJoin(o.Bucket, o.Prefix))
+
+	for {
+		if contextCanceled(ctx) {
+			return entries, ctx.Err()
+		}
+
+		// If many failures, check the cache state.
+		if retries > 10 {
+			err := o.checkMetacacheState(ctx, rpc)
+			if err != nil {
+				return entries, fmt.Errorf("remote listing canceled: %w", err)
+			}
+			retries = 1
+		}
+
+		const retryDelay = 250 * time.Millisecond
+		// All operations are performed without locks, so we must be careful and allow for failures.
+		// Read metadata associated with the object from a disk.
+		if retries > 0 {
+			_, err := es.disk.ReadVersion(ctx, minioMetaBucket,
+				o.objectPath(0), "", false)
+			if err != nil {
+				time.Sleep(retryDelay)
+				retries++
+				continue
+			}
+		}
+
+		// Load first part metadata...
+		// Read metadata associated with the object from all disks.
+		fi, metaArr, onlineDisks, err := es.getObjectFileInfo(ctx, minioMetaBucket, o.objectPath(0), ObjectOptions{}, true)
+		if err != nil {
+			switch toObjectErr(err, minioMetaBucket, o.objectPath(0)).(type) {
+			case ObjectNotFound:
+				retries++
+				time.Sleep(retryDelay)
+				continue
+			case InsufficientReadQuorum:
+				retries++
+				time.Sleep(retryDelay)
+				continue
+			default:
+				return entries, fmt.Errorf("reading first part metadata: %w", err)
+			}
+		}
+
+		partN, err := o.findFirstPart(fi)
+		switch {
+		case err == nil:
+		case errors.Is(err, io.ErrUnexpectedEOF):
+			if retries == 10 {
+				err := o.checkMetacacheState(ctx, rpc)
+				if err != nil {
+					return entries, fmt.Errorf("remote listing canceled: %w", err)
+				}
+				retries = -1
+			}
+			retries++
+			time.Sleep(retryDelay)
+			continue
+		case errors.Is(err, io.EOF):
+			return entries, io.EOF
+		}
+
+		// We got a stream to start at.
+		loadedPart := 0
+		for {
+			if contextCanceled(ctx) {
+				return entries, ctx.Err()
+			}
+
+			if partN != loadedPart {
+				if retries > 10 {
+					err := o.checkMetacacheState(ctx, rpc)
+					if err != nil {
+						return entries, fmt.Errorf("waiting for next part %d: %w", partN, err)
+					}
+					retries = 1
+				}
+
+				if retries > 0 {
+					// Load from one disk only
+					_, err := es.disk.ReadVersion(ctx, minioMetaBucket,
+						o.objectPath(partN), "", false)
+					if err != nil {
+						time.Sleep(retryDelay)
+						retries++
+						continue
+					}
+				}
+
+				// Load partN metadata...
+				fi, metaArr, onlineDisks, err = es.getObjectFileInfo(ctx, minioMetaBucket, o.objectPath(partN), ObjectOptions{}, true)
+				if err != nil {
+					time.Sleep(retryDelay)
+					retries++
+					continue
+				}
+				loadedPart = partN
+				bi, err := getMetacacheBlockInfo(fi, partN)
+				logger.LogIf(ctx, err)
+				if err == nil {
+					if bi.pastPrefix(o.Prefix) {
+						return entries, io.EOF
+					}
+				}
+			}
+
+			pr, pw := io.Pipe()
+			go func() {
+				werr := es.getObjectWithFileInfo(ctx, minioMetaBucket, o.objectPath(partN), 0,
+					fi.Size, pw, fi, metaArr, onlineDisks)
+				pw.CloseWithError(werr)
+			}()
+
+			tmp := newMetacacheReader(pr)
+			e, err := tmp.filter(o)
+			pr.CloseWithError(err)
+			entries.o = append(entries.o, e.o...)
+			if o.Limit > 0 && entries.len() > o.Limit {
+				entries.truncate(o.Limit)
+				return entries, nil
+			}
+			if err == nil {
+				// We stopped within the listing, we are done for now...
+				return entries, nil
+			}
+			if err != nil && err.Error() != io.EOF.Error() {
+				switch toObjectErr(err, minioMetaBucket, o.objectPath(partN)).(type) {
+				case ObjectNotFound:
+					retries++
+					time.Sleep(retryDelay)
+					continue
+				case InsufficientReadQuorum:
+					retries++
+					time.Sleep(retryDelay)
+					continue
+				default:
+					logger.LogIf(ctx, err)
+					return entries, err
+				}
+			}
+
+			// We finished at the end of the block.
+			// And should not expect any more results.
+			bi, err := getMetacacheBlockInfo(fi, partN)
+			logger.LogIf(ctx, err)
+			if err != nil || bi.EOS {
+				// We are done and there are no more parts.
+				return entries, io.EOF
+			}
+			if bi.endedPrefix(o.Prefix) {
+				// Nothing more for prefix.
+				return entries, io.EOF
+			}
+			partN++
+			retries = 0
+		}
+	}
+}
+
 // getListQuorum interprets list quorum values and returns appropriate
 // acceptable quorum expected for list operations
 func getListQuorum(quorum string, driveCount int) int {
@@ -562,6 +726,53 @@ func getListQuorum(quorum string, driveCount int) int {
 	return 3
 }
 
+// Will return io.EOF if continuing would not yield more results.
+func (es *erasureSingle) listPathInner(ctx context.Context, o listPathOptions, results chan<- metaCacheEntry) (err error) {
+	defer close(results)
+	o.debugf(color.Green("listPath:")+" with options: %#v", o)
+
+	// How to resolve results.
+	resolver := metadataResolutionParams{
+		dirQuorum: 1,
+		objQuorum: 1,
+		bucket:    o.Bucket,
+	}
+
+	// Maximum versions requested for "latest" object
+	// resolution on versioned buckets, this is to be only
+	// used when o.Versioned is false
+	if !o.Versioned {
+		resolver.requestedVersions = 1
+	}
+
+	ctxDone := ctx.Done()
+	return listPathRaw(ctx, listPathRawOptions{
+		disks:        []StorageAPI{es.disk},
+		bucket:       o.Bucket,
+		path:         o.BaseDir,
+		recursive:    o.Recursive,
+		filterPrefix: o.FilterPrefix,
+		minDisks:     1,
+		forwardTo:    o.Marker,
+		agreed: func(entry metaCacheEntry) {
+			select {
+			case <-ctxDone:
+			case results <- entry:
+			}
+		},
+		partial: func(entries metaCacheEntries, nAgreed int, errs []error) {
+			// Results Disagree :-(
+			entry, ok := entries.resolve(&resolver)
+			if ok {
+				select {
+				case <-ctxDone:
+				case results <- *entry:
+				}
+			}
+		},
+	})
+}
+
 // Will return io.EOF if continuing would not yield more results.
 func (er *erasureObjects) listPath(ctx context.Context, o listPathOptions, results chan<- metaCacheEntry) (err error) {
 	defer close(results)
@@ -654,6 +865,133 @@ func (m *metaCacheRPC) setErr(err string) {
 	*m.meta = meta
 }
 
+func (es *erasureSingle) saveMetaCacheStream(ctx context.Context, mc *metaCacheRPC, entries <-chan metaCacheEntry) (err error) {
+	o := mc.o
+	o.debugf(color.Green("saveMetaCacheStream:")+" with options: %#v", o)
+
+	metaMu := &mc.mu
+	rpc := mc.rpc
+	cancel := mc.cancel
+	defer func() {
+		o.debugln(color.Green("saveMetaCacheStream:")+"err:", err)
+		if err != nil && !errors.Is(err, io.EOF) {
+			go mc.setErr(err.Error())
+			cancel()
+		}
+	}()
+
+	defer cancel()
+	// Save continuous updates
+	go func() {
+		var err error
+		ticker := time.NewTicker(10 * time.Second)
+		defer ticker.Stop()
+		var exit bool
+		for !exit {
+			select {
+			case <-ticker.C:
+			case <-ctx.Done():
+				exit = true
+			}
+			metaMu.Lock()
+			meta := *mc.meta
+			meta, err = o.updateMetacacheListing(meta, rpc)
+			if err == nil && time.Since(meta.lastHandout) > metacacheMaxClientWait {
+				cancel()
+				exit = true
+				meta.status = scanStateError
+				meta.error = fmt.Sprintf("listing canceled since time since last handout was %v ago", time.Since(meta.lastHandout).Round(time.Second))
+				o.debugln(color.Green("saveMetaCacheStream: ") + meta.error)
+				meta, err = o.updateMetacacheListing(meta, rpc)
+			}
+			if err == nil {
+				*mc.meta = meta
+				if meta.status == scanStateError {
+					cancel()
+					exit = true
+				}
+			}
+			metaMu.Unlock()
+		}
+	}()
+
+	const retryDelay = 200 * time.Millisecond
+	const maxTries = 5
+
+	// Keep destination...
+	// Write results to disk.
+	bw := newMetacacheBlockWriter(entries, func(b *metacacheBlock) error {
+		// if the block is 0 bytes and its a first block skip it.
+		// skip only this for Transient caches.
+		if len(b.data) == 0 && b.n == 0 && o.Transient {
+			return nil
+		}
+		o.debugln(color.Green("saveMetaCacheStream:")+" saving block", b.n, "to", o.objectPath(b.n))
+		r, err := hash.NewReader(bytes.NewReader(b.data), int64(len(b.data)), "", "", int64(len(b.data)))
+		logger.LogIf(ctx, err)
+		custom := b.headerKV()
+		_, err = es.putMetacacheObject(ctx, o.objectPath(b.n), NewPutObjReader(r), ObjectOptions{
+			UserDefined: custom,
+		})
+		if err != nil {
+			mc.setErr(err.Error())
+			cancel()
+			return err
+		}
+		if b.n == 0 {
+			return nil
+		}
+		// Update block 0 metadata.
+		var retries int
+		for {
+			meta := b.headerKV()
+			fi := FileInfo{
+				Metadata: make(map[string]string, len(meta)),
+			}
+			for k, v := range meta {
+				fi.Metadata[k] = v
+			}
+			err := es.updateObjectMeta(ctx, minioMetaBucket, o.objectPath(0), fi, es.disk)
+			if err == nil {
+				break
+			}
+			switch err.(type) {
+			case ObjectNotFound:
+				return err
+			case StorageErr:
+				return err
+			case InsufficientReadQuorum:
+			default:
+				logger.LogIf(ctx, err)
+			}
+			if retries >= maxTries {
+				return err
+			}
+			retries++
+			time.Sleep(retryDelay)
+		}
+		return nil
+	})
+
+	// Blocks while consuming entries or an error occurs.
+	err = bw.Close()
+	if err != nil {
+		mc.setErr(err.Error())
+	}
+	metaMu.Lock()
+	defer metaMu.Unlock()
+	if mc.meta.error != "" {
+		return err
+	}
+	// Save success
+	mc.meta.status = scanStateSuccess
+	meta, err := o.updateMetacacheListing(*mc.meta, rpc)
+	if err == nil {
+		*mc.meta = meta
+	}
+	return nil
+}
+
 func (er *erasureObjects) saveMetaCacheStream(ctx context.Context, mc *metaCacheRPC, entries <-chan metaCacheEntry) (err error) {
 	o := mc.o
 	o.debugf(color.Green("saveMetaCacheStream:")+" with options: %#v", o)
@@ -882,7 +1220,11 @@ func listPathRaw(ctx context.Context, opts listPathRawOptions) (err error) {
 			}
 
 			// fallback only when set.
-			for fd := fallback(werr); fd != nil; {
+			for {
+				fd := fallback(werr)
+				if fd == nil {
+					break
+				}
 				// This fallback is only set when
 				// askDisks is less than total
 				// number of disks per set.

cmd/metacache.go

@@ -155,9 +155,9 @@ func (m *metacache) delete(ctx context.Context) {
 		logger.LogIf(ctx, errors.New("metacache.delete: no object layer"))
 		return
 	}
-	ez, ok := objAPI.(*erasureServerPools)
+	ez, ok := objAPI.(renameAllStorager)
 	if !ok {
-		logger.LogIf(ctx, errors.New("metacache.delete: expected objAPI to be *erasureServerPools"))
+		logger.LogIf(ctx, errors.New("metacache.delete: expected objAPI to be 'renameAllStorager'"))
 		return
 	}
 	ez.renameAll(ctx, minioMetaBucket, metacachePrefixForID(m.bucket, m.id))

cmd/metrics-v2.go

@@ -1365,7 +1365,7 @@ func getMinioHealingMetrics() *MetricsGroup {
 	mg := &MetricsGroup{}
 	mg.RegisterRead(func(_ context.Context) (metrics []Metric) {
 		metrics = make([]Metric, 0, 5)
-		if !globalIsErasure {
+		if globalIsGateway {
 			return
 		}
 		bgSeq, exists := globalBackgroundHealState.getHealSequenceByToken(bgHealingUUID)
@@ -1604,8 +1604,9 @@ func getBucketUsageMetrics() *MetricsGroup {
 			Value:       float64(time.Since(dataUsageInfo.LastUpdate)),
 		})
 
+		bucketReplStats := getAllLatestReplicationStats(dataUsageInfo.BucketsUsage)
 		for bucket, usage := range dataUsageInfo.BucketsUsage {
-			stats := getLatestReplicationStats(bucket, usage)
+			stats := bucketReplStats[bucket]
 
 			quota, _ := globalBucketQuotaSys.Get(ctx, bucket)
 
@@ -1816,7 +1817,7 @@ func getClusterStorageMetrics() *MetricsGroup {
 	mg.RegisterRead(func(ctx context.Context) (metrics []Metric) {
 		objLayer := newObjectLayerFn()
 		// Service not initialized yet
-		if objLayer == nil || !globalIsErasure {
+		if objLayer == nil || globalIsGateway {
 			return
 		}
 

cmd/metrics.go

@@ -132,7 +132,7 @@ func nodeHealthMetricsPrometheus(ch chan<- prometheus.Metric) {
 // collects healing specific metrics for MinIO instance in Prometheus specific format
 // and sends to given channel
 func healingMetricsPrometheus(ch chan<- prometheus.Metric) {
-	if !globalIsErasure {
+	if globalIsGateway {
 		return
 	}
 	bgSeq, exists := globalBackgroundHealState.getHealSequenceByToken(bgHealingUUID)

cmd/mrf.go

@@ -190,12 +190,12 @@ func (m *mrfState) healRoutine() {
 	}
 
 	for {
-		idler.Reset(mrfInfoResetInterval)
 		select {
 		case <-m.ctx.Done():
 			return
 		case <-idler.C:
 			m.resetMRFInfoIfNoPendingOps()
+			idler.Reset(mrfInfoResetInterval)
 		case setInfo := <-m.setReconnectEvent:
 			// Get the list of objects related the er.set
 			// to which the connected disk belongs.
@@ -218,21 +218,19 @@ func (m *mrfState) healRoutine() {
 
 			// Heal objects
 			for _, u := range mrfOperations {
-				if _, err := m.objectAPI.HealObject(m.ctx, u.bucket, u.object, u.versionID, mrfHealingOpts); err != nil {
-					// If not deleted, assume they failed.
-					logger.LogIf(m.ctx, err)
-				} else {
-					m.mu.Lock()
-					m.itemsHealed++
-					m.pendingItems--
-					m.bytesHealed += uint64(u.size)
-					m.pendingBytes -= uint64(u.size)
-					m.mu.Unlock()
-				}
-
+				_, err := m.objectAPI.HealObject(m.ctx, u.bucket, u.object, u.versionID, mrfHealingOpts)
 				m.mu.Lock()
+				if err == nil {
+					m.itemsHealed++
+					m.bytesHealed += uint64(u.size)
+				}
+				m.pendingItems--
+				m.pendingBytes -= uint64(u.size)
 				delete(m.pendingOps, u)
 				m.mu.Unlock()
+
+				// Log healing error if any
+				logger.LogIf(m.ctx, err)
 			}
 
 			waitForLowHTTPReq()

cmd/notification.go

@@ -414,6 +414,21 @@ func (sys *NotificationSys) ServerUpdate(ctx context.Context, u *url.URL, sha256
 	return ng.Wait()
 }
 
+// SignalConfigReload reloads requested sub-system on a remote peer dynamically.
+func (sys *NotificationSys) SignalConfigReload(subSys string) []NotificationPeerErr {
+	ng := WithNPeers(len(sys.peerClients))
+	for idx, client := range sys.peerClients {
+		if client == nil {
+			continue
+		}
+		client := client
+		ng.Go(GlobalContext, func() error {
+			return client.SignalService(serviceReloadDynamic, subSys)
+		}, idx, *client.host)
+	}
+	return ng.Wait()
+}
+
 // SignalService - calls signal service RPC call on all peers.
 func (sys *NotificationSys) SignalService(sig serviceSignal) []NotificationPeerErr {
 	ng := WithNPeers(len(sys.peerClients))
@@ -423,7 +438,7 @@ func (sys *NotificationSys) SignalService(sig serviceSignal) []NotificationPeerE
 		}
 		client := client
 		ng.Go(GlobalContext, func() error {
-			return client.SignalService(sig)
+			return client.SignalService(sig, "")
 		}, idx, *client.host)
 	}
 	return ng.Wait()
@@ -586,6 +601,44 @@ func (sys *NotificationSys) DeleteBucketMetadata(ctx context.Context, bucketName
 	}
 }
 
+// GetClusterAllBucketStats - returns bucket stats for all buckets from all remote peers.
+func (sys *NotificationSys) GetClusterAllBucketStats(ctx context.Context) []BucketStatsMap {
+	ng := WithNPeers(len(sys.peerClients))
+	replicationStats := make([]BucketStatsMap, len(sys.peerClients))
+	for index, client := range sys.peerClients {
+		index := index
+		client := client
+		ng.Go(ctx, func() error {
+			if client == nil {
+				return errPeerNotReachable
+			}
+			bsMap, err := client.GetAllBucketStats()
+			if err != nil {
+				return err
+			}
+			replicationStats[index] = bsMap
+			return nil
+		}, index, *client.host)
+	}
+	for _, nErr := range ng.Wait() {
+		reqInfo := (&logger.ReqInfo{}).AppendTags("peerAddress", nErr.Host.String())
+		if nErr.Err != nil {
+			logger.LogIf(logger.SetReqInfo(ctx, reqInfo), nErr.Err)
+		}
+	}
+
+	replicationStatsList := globalReplicationStats.GetAll()
+	bucketStatsMap := make(map[string]BucketStats, len(replicationStatsList))
+	for k, replicationStats := range replicationStatsList {
+		bucketStatsMap[k] = BucketStats{
+			ReplicationStats: replicationStats,
+		}
+	}
+
+	replicationStats = append(replicationStats, BucketStatsMap(bucketStatsMap))
+	return replicationStats
+}
+
 // GetClusterBucketStats - calls GetClusterBucketStats call on all peers for a cluster statistics view.
 func (sys *NotificationSys) GetClusterBucketStats(ctx context.Context, bucketName string) []BucketStats {
 	ng := WithNPeers(len(sys.peerClients))
@@ -1541,7 +1594,7 @@ func (sys *NotificationSys) ServiceFreeze(ctx context.Context, freeze bool) []No
 		}
 		client := client
 		ng.Go(GlobalContext, func() error {
-			return client.SignalService(serviceSig)
+			return client.SignalService(serviceSig, "")
 		}, idx, *client.host)
 	}
 	nerrs := ng.Wait()

cmd/object-api-interface.go

@@ -76,6 +76,8 @@ type ObjectOptions struct {
 	// Mutate set to 'true' if the call is namespace mutation call
 	Mutate        bool
 	WalkAscending bool // return Walk results in ascending order of versions
+
+	PrefixEnabledFn func(prefix string) bool // function which returns true if versioning is enabled on prefix
 }
 
 // ExpirationOptions represents object options for object expiration at objectLayer.

cmd/object-api-listobjects_test.go

@@ -35,9 +35,6 @@ func TestListObjectsVersionedFolders(t *testing.T) {
 }
 
 func testListObjectsVersionedFolders(obj ObjectLayer, instanceType string, t1 TestErrHandler) {
-	if instanceType == FSTestStr {
-		return
-	}
 	t, _ := t1.(*testing.T)
 	testBuckets := []string{
 		// This bucket is used for testing ListObject operations.
@@ -70,7 +67,7 @@ func testListObjectsVersionedFolders(obj ObjectLayer, instanceType string, t1 Te
 		md5Bytes := md5.Sum([]byte(object.content))
 		_, err = obj.PutObject(context.Background(), object.parentBucket, object.name, mustGetPutObjReader(t, bytes.NewBufferString(object.content),
 			int64(len(object.content)), hex.EncodeToString(md5Bytes[:]), ""), ObjectOptions{
-			Versioned:   globalBucketVersioningSys.Enabled(object.parentBucket),
+			Versioned:   globalBucketVersioningSys.PrefixEnabled(object.parentBucket, object.name),
 			UserDefined: object.meta,
 		})
 		if err != nil {
@@ -78,7 +75,7 @@ func testListObjectsVersionedFolders(obj ObjectLayer, instanceType string, t1 Te
 		}
 		if object.addDeleteMarker {
 			oi, err := obj.DeleteObject(context.Background(), object.parentBucket, object.name, ObjectOptions{
-				Versioned: globalBucketVersioningSys.Enabled(object.parentBucket),
+				Versioned: globalBucketVersioningSys.PrefixEnabled(object.parentBucket, object.name),
 			})
 			if err != nil {
 				t.Fatalf("%s : %s", instanceType, err.Error())
@@ -317,9 +314,6 @@ func testListObjects(obj ObjectLayer, instanceType string, t1 TestErrHandler) {
 }
 
 func _testListObjects(obj ObjectLayer, instanceType string, t1 TestErrHandler, versioned bool) {
-	if instanceType == FSTestStr && versioned {
-		return
-	}
 	t, _ := t1.(*testing.T)
 	testBuckets := []string{
 		// This bucket is used for testing ListObject operations.
@@ -380,7 +374,7 @@ func _testListObjects(obj ObjectLayer, instanceType string, t1 TestErrHandler, v
 		_, err = obj.PutObject(context.Background(), object.parentBucket, object.name,
 			mustGetPutObjReader(t, bytes.NewBufferString(object.content),
 				int64(len(object.content)), hex.EncodeToString(md5Bytes[:]), ""), ObjectOptions{
-				Versioned:   globalBucketVersioningSys.Enabled(object.parentBucket),
+				Versioned:   globalBucketVersioningSys.PrefixEnabled(object.parentBucket, object.name),
 				UserDefined: object.meta,
 			})
 		if err != nil {
@@ -1020,10 +1014,6 @@ func TestDeleteObjectVersionMarker(t *testing.T) {
 }
 
 func testDeleteObjectVersion(obj ObjectLayer, instanceType string, t1 TestErrHandler) {
-	if instanceType == FSTestStr {
-		return
-	}
-
 	t, _ := t1.(*testing.T)
 
 	testBuckets := []string{
@@ -1065,16 +1055,16 @@ func testDeleteObjectVersion(obj ObjectLayer, instanceType string, t1 TestErrHan
 		_, err := obj.PutObject(context.Background(), object.parentBucket, object.name,
 			mustGetPutObjReader(t, bytes.NewBufferString(object.content),
 				int64(len(object.content)), hex.EncodeToString(md5Bytes[:]), ""), ObjectOptions{
-				Versioned:        globalBucketVersioningSys.Enabled(object.parentBucket),
-				VersionSuspended: globalBucketVersioningSys.Suspended(object.parentBucket),
+				Versioned:        globalBucketVersioningSys.PrefixEnabled(object.parentBucket, object.name),
+				VersionSuspended: globalBucketVersioningSys.PrefixSuspended(object.parentBucket, object.name),
 				UserDefined:      object.meta,
 			})
 		if err != nil {
 			t.Fatalf("%s : %s", instanceType, err)
 		}
 		obj, err := obj.DeleteObject(context.Background(), object.parentBucket, object.name, ObjectOptions{
-			Versioned:        globalBucketVersioningSys.Enabled(object.parentBucket),
-			VersionSuspended: globalBucketVersioningSys.Suspended(object.parentBucket),
+			Versioned:        globalBucketVersioningSys.PrefixEnabled(object.parentBucket, object.name),
+			VersionSuspended: globalBucketVersioningSys.PrefixSuspended(object.parentBucket, object.name),
 			VersionID:        object.versionID,
 		})
 		if err != nil {
@@ -1101,10 +1091,6 @@ func TestListObjectVersions(t *testing.T) {
 
 // Unit test for ListObjectVersions
 func testListObjectVersions(obj ObjectLayer, instanceType string, t1 TestErrHandler) {
-	if instanceType == FSTestStr {
-		return
-	}
-
 	t, _ := t1.(*testing.T)
 	testBuckets := []string{
 		// This bucket is used for testing ListObject operations.
@@ -1886,16 +1872,14 @@ func testListObjectsContinuation(obj ObjectLayer, instanceType string, t1 TestEr
 
 // Initialize FS backend for the benchmark.
 func initFSObjectsB(disk string, t *testing.B) (obj ObjectLayer) {
-	var err error
-	obj, err = NewFSObjectLayer(disk)
+	obj, _, err := initObjectLayer(context.Background(), mustGetPoolEndpoints(disk))
 	if err != nil {
-		t.Fatal("Unexpected err: ", err)
+		t.Fatal(err)
 	}
 
 	newTestConfig(globalMinioDefaultRegion, obj)
 
 	initAllSubsystems()
-
 	return obj
 }
 

cmd/object-api-multipart_test.go

@@ -1620,39 +1620,35 @@ func testListObjectParts(obj ObjectLayer, instanceType string, t TestErrHandler)
 				t.Errorf("Test %d: %s: Expected Bucket to be \"%s\", but instead found it to be \"%s\"", i+1, instanceType, expectedResult.Bucket, actualResult.Bucket)
 			}
 
-			// ListObjectParts returns empty response always in FS mode
-			if instanceType != FSTestStr {
-				// Asserting IsTruncated.
-				if actualResult.IsTruncated != testCase.expectedResult.IsTruncated {
-					t.Errorf("Test %d: %s: Expected IsTruncated to be \"%v\", but found it to \"%v\"", i+1, instanceType, expectedResult.IsTruncated, actualResult.IsTruncated)
-					continue
+			// Asserting IsTruncated.
+			if actualResult.IsTruncated != testCase.expectedResult.IsTruncated {
+				t.Errorf("Test %d: %s: Expected IsTruncated to be \"%v\", but found it to \"%v\"", i+1, instanceType, expectedResult.IsTruncated, actualResult.IsTruncated)
+				continue
+			}
+			// Asserting NextPartNumberMarker.
+			if actualResult.NextPartNumberMarker != expectedResult.NextPartNumberMarker {
+				t.Errorf("Test %d: %s: Expected NextPartNumberMarker to be \"%d\", but instead found it to be \"%d\"", i+1, instanceType, expectedResult.NextPartNumberMarker, actualResult.NextPartNumberMarker)
+				continue
+			}
+			// Asserting the number of Parts.
+			if len(expectedResult.Parts) != len(actualResult.Parts) {
+				t.Errorf("Test %d: %s: Expected the result to contain info of %d Parts, but found %d instead", i+1, instanceType, len(expectedResult.Parts), len(actualResult.Parts))
+				continue
+			}
+			// Iterating over the partInfos and asserting the fields.
+			for j, actualMetaData := range actualResult.Parts {
+				//  Asserting the PartNumber in the PartInfo.
+				if actualMetaData.PartNumber != expectedResult.Parts[j].PartNumber {
+					t.Errorf("Test %d: %s: Part %d: Expected PartNumber to be \"%d\", but instead found \"%d\"", i+1, instanceType, j+1, expectedResult.Parts[j].PartNumber, actualMetaData.PartNumber)
 				}
-				// Asserting NextPartNumberMarker.
-				if actualResult.NextPartNumberMarker != expectedResult.NextPartNumberMarker {
-					t.Errorf("Test %d: %s: Expected NextPartNumberMarker to be \"%d\", but instead found it to be \"%d\"", i+1, instanceType, expectedResult.NextPartNumberMarker, actualResult.NextPartNumberMarker)
-					continue
+				//  Asserting the Size in the PartInfo.
+				if actualMetaData.Size != expectedResult.Parts[j].Size {
+					t.Errorf("Test %d: %s: Part %d: Expected Part Size to be \"%d\", but instead found \"%d\"", i+1, instanceType, j+1, expectedResult.Parts[j].Size, actualMetaData.Size)
 				}
-				// Asserting the number of Parts.
-				if len(expectedResult.Parts) != len(actualResult.Parts) {
-					t.Errorf("Test %d: %s: Expected the result to contain info of %d Parts, but found %d instead", i+1, instanceType, len(expectedResult.Parts), len(actualResult.Parts))
-					continue
+				//  Asserting the ETag in the PartInfo.
+				if actualMetaData.ETag != expectedResult.Parts[j].ETag {
+					t.Errorf("Test %d: %s: Part %d: Expected Etag to be \"%s\", but instead found \"%s\"", i+1, instanceType, j+1, expectedResult.Parts[j].ETag, actualMetaData.ETag)
 				}
-				// Iterating over the partInfos and asserting the fields.
-				for j, actualMetaData := range actualResult.Parts {
-					//  Asserting the PartNumber in the PartInfo.
-					if actualMetaData.PartNumber != expectedResult.Parts[j].PartNumber {
-						t.Errorf("Test %d: %s: Part %d: Expected PartNumber to be \"%d\", but instead found \"%d\"", i+1, instanceType, j+1, expectedResult.Parts[j].PartNumber, actualMetaData.PartNumber)
-					}
-					//  Asserting the Size in the PartInfo.
-					if actualMetaData.Size != expectedResult.Parts[j].Size {
-						t.Errorf("Test %d: %s: Part %d: Expected Part Size to be \"%d\", but instead found \"%d\"", i+1, instanceType, j+1, expectedResult.Parts[j].Size, actualMetaData.Size)
-					}
-					//  Asserting the ETag in the PartInfo.
-					if actualMetaData.ETag != expectedResult.Parts[j].ETag {
-						t.Errorf("Test %d: %s: Part %d: Expected Etag to be \"%s\", but instead found \"%s\"", i+1, instanceType, j+1, expectedResult.Parts[j].ETag, actualMetaData.ETag)
-					}
-				}
-
 			}
 		}
 	}

cmd/object-api-options.go

@@ -157,17 +157,18 @@ func getOpts(ctx context.Context, r *http.Request, bucket, object string) (Objec
 			}
 		}
 	}
+	opts.Versioned = globalBucketVersioningSys.PrefixEnabled(bucket, object)
+	opts.VersionSuspended = globalBucketVersioningSys.PrefixSuspended(bucket, object)
 	return opts, nil
 }
 
 func delOpts(ctx context.Context, r *http.Request, bucket, object string) (opts ObjectOptions, err error) {
-	versioned := globalBucketVersioningSys.Enabled(bucket)
 	opts, err = getOpts(ctx, r, bucket, object)
 	if err != nil {
 		return opts, err
 	}
-	opts.Versioned = versioned
-	opts.VersionSuspended = globalBucketVersioningSys.Suspended(bucket)
+	opts.Versioned = globalBucketVersioningSys.PrefixEnabled(bucket, object)
+	opts.VersionSuspended = globalBucketVersioningSys.PrefixSuspended(bucket, object)
 	delMarker := strings.TrimSpace(r.Header.Get(xhttp.MinIOSourceDeleteMarker))
 	if delMarker != "" {
 		switch delMarker {
@@ -203,8 +204,9 @@ func delOpts(ctx context.Context, r *http.Request, bucket, object string) (opts
 
 // get ObjectOptions for PUT calls from encryption headers and metadata
 func putOpts(ctx context.Context, r *http.Request, bucket, object string, metadata map[string]string) (opts ObjectOptions, err error) {
-	versioned := globalBucketVersioningSys.Enabled(bucket)
-	versionSuspended := globalBucketVersioningSys.Suspended(bucket)
+	versioned := globalBucketVersioningSys.PrefixEnabled(bucket, object)
+	versionSuspended := globalBucketVersioningSys.PrefixSuspended(bucket, object)
+
 	vid := strings.TrimSpace(r.Form.Get(xhttp.VersionID))
 	if vid != "" && vid != nullVersionID {
 		_, err := uuid.Parse(vid)

cmd/object-api-utils.go

@@ -950,7 +950,7 @@ func compressSelfTest() {
 
 // getDiskInfos returns the disk information for the provided disks.
 // If a disk is nil or an error is returned the result will be nil as well.
-func getDiskInfos(ctx context.Context, disks []StorageAPI) []*DiskInfo {
+func getDiskInfos(ctx context.Context, disks ...StorageAPI) []*DiskInfo {
 	res := make([]*DiskInfo, len(disks))
 	for i, disk := range disks {
 		if disk == nil {