Validate config before saving changes after config reset (#14203)

some upgraded objects might not get listed due
to different quorum ratios across objects.

make sure to list all objects that satisfy the
maximum possible quorum.

.github/ISSUE_TEMPLATE.md

@@ -7,6 +7,9 @@ assignees: ''
 
 ---
 
+## NOTE
+If this case is urgent, please subscribe to [Subnet](https://min.io/pricing) so that our 24/7 support team may help you faster.
+
 <!--- Provide a general summary of the issue in the Title above -->
 
 ## Expected Behavior

.github/ISSUE_TEMPLATE/bug_report.md

@@ -7,6 +7,9 @@ assignees: ''
 
 ---
 
+## NOTE
+If this case is urgent, please subscribe to [Subnet](https://min.io/pricing) so that our 24/7 support team may help you faster.
+
 <!--- Provide a general summary of the issue in the Title above -->
 
 ## Expected Behavior

.github/lock.yml

@@ -1,39 +0,0 @@
-# Configuration for Lock Threads - https://github.com/dessant/lock-threads-app
-
-# Number of days of inactivity before a closed issue or pull request is locked
-daysUntilLock: 365
-
-# Skip issues and pull requests created before a given timestamp. Timestamp must
-# follow ISO 8601 (`YYYY-MM-DD`). Set to `false` to disable
-skipCreatedBefore: false
-
-# Issues and pull requests with these labels will be ignored. Set to `[]` to disable
-exemptLabels: []
-
-# Label to add before locking, such as `outdated`. Set to `false` to disable
-lockLabel: true
-
-# Comment to post before locking. Set to `false` to disable
-lockComment: >-
-
-  This thread has been automatically locked since there has not been
-  any recent activity after it was closed. Please open a new issue for
-  related bugs.
-
-# Assign `resolved` as the reason for locking. Set to `false` to disable
-setLockReason: true
-
-# Limit to only `issues` or `pulls`
-only: issues
-
-# Optionally, specify configuration settings just for `issues` or `pulls`
-# issues:
-#   exemptLabels:
-#     - help-wanted
-#   lockLabel: outdated
-
-# pulls:
-#   daysUntilLock: 30
-
-# Repository to extend settings from
-# _extends: repo

.github/stale.yml

@@ -14,7 +14,7 @@ onlyLabels: []
 exemptLabels:
   - "security"
   - "pending discussion"
-  - "do not close"
+  - "do-not-close"
 
 # Set to true to ignore issues in a project (defaults to false)
 exemptProjects: false

.github/workflows/go-healing.yml

@@ -0,0 +1,48 @@
+name: Healing Functional Tests
+
+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
+
+jobs:
+  build:
+    name: Go ${{ matrix.go-version }} on ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        go-version: [1.17.x]
+        os: [ubuntu-latest]
+    steps:
+      - uses: actions/checkout@v2
+      - uses: actions/setup-go@v2
+        with:
+          go-version: ${{ matrix.go-version }}
+      - uses: actions/cache@v2
+        with:
+          path: |
+            ~/.cache/go-build
+            ~/go/pkg/mod
+          key: ${{ runner.os }}-${{ matrix.go-version }}-go-${{ hashFiles('**/go.sum') }}
+          restore-keys: |
+            ${{ runner.os }}-${{ matrix.go-version }}-go-
+      - name: Build on ${{ matrix.os }}
+        if: matrix.os == 'ubuntu-latest'
+        env:
+          CGO_ENABLED: 0
+          GO111MODULE: on
+          MINIO_KMS_KES_CERT_FILE: /home/runner/work/minio/minio/.github/workflows/root.cert
+          MINIO_KMS_KES_KEY_FILE: /home/runner/work/minio/minio/.github/workflows/root.key
+          MINIO_KMS_KES_ENDPOINT: "https://play.min.io:7373"
+          MINIO_KMS_KES_KEY_NAME: "my-minio-key"
+          MINIO_KMS_AUTO_ENCRYPTION: on
+        run: |
+          sudo sysctl net.ipv6.conf.all.disable_ipv6=0
+          sudo sysctl net.ipv6.conf.default.disable_ipv6=0
+          make verify-healing

.github/workflows/go.yml

@@ -13,7 +13,7 @@ concurrency:
 
 jobs:
   build:
-    name: Go ${{ matrix.go-version }} on ${{ matrix.os }}
+    name: Go ${{ matrix.go-version }} on ${{ matrix.os }} - healing
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
@@ -46,4 +46,3 @@ jobs:
           sudo sysctl net.ipv6.conf.all.disable_ipv6=0
           sudo sysctl net.ipv6.conf.default.disable_ipv6=0
           make verify
-          make verify-healing

.github/workflows/iam-integrations.yaml

@@ -88,4 +88,8 @@ jobs:
       - name: Test LDAP for automatic site replication
         if: matrix.ldap == 'localhost:389'
         run: |
-          make test-site-replication
+          make test-site-replication-ldap
+      - name: Test OIDC for automatic site replication
+        if: matrix.openid == 'http://127.0.0.1:5556/dex'
+        run: |
+          make test-site-replication-oidc

.github/workflows/lock.yml

@@ -0,0 +1,25 @@
+name: 'Lock Threads'
+
+on:
+  schedule:
+    - cron: '0 0 * * *'
+  workflow_dispatch:
+
+permissions:
+  issues: write
+
+concurrency:
+  group: lock
+
+jobs:
+  action:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: dessant/lock-threads@v3
+        with:
+          github-token: ${{ github.token }}
+          issue-inactive-days: '730'
+          exclude-any-issue-labels: 'do-not-close'
+          issue-lock-reason: 'resolved'
+          process-only: 'issues'
+          log-output: true

.github/workflows/replication.yaml

@@ -38,3 +38,9 @@ jobs:
           sudo sysctl net.ipv6.conf.all.disable_ipv6=0
           sudo sysctl net.ipv6.conf.default.disable_ipv6=0
           make test-replication
+      - name: Test MinIO IDP for automatic site replication
+        run: |
+          sudo sysctl net.ipv6.conf.all.disable_ipv6=0
+          sudo sysctl net.ipv6.conf.default.disable_ipv6=0
+          make test-site-replication-minio
+

.gitignore

@@ -25,10 +25,11 @@ node_modules/
 mc.*
 s3-check-md5*
 xl-meta*
-healing-bin*
+healing-*
 inspect*
 200M*
 hash-set
 minio.RELEASE*
 mc
 nancy
+inspects/*

.golangci.yml

@@ -24,6 +24,15 @@ linters:
     - unconvert
     - varcheck
     - gocritic
+    - gofumpt
+
+linters-settings:
+  gofumpt:
+    lang-version: "1.17"
+
+    # Choose whether or not to use the extra rules that are disabled
+    # by default
+    extra-rules: false
 
 issues:
   exclude-use-default: false

COMPLIANCE.md

@@ -0,0 +1,6 @@
+# AGPLv3 Compliance
+We have designed MinIO as an Open Source software for the Open Source software community. This requires applications to consider whether their usage of MinIO is in compliance with the GNU AGPLv3 [license](https://github.com/minio/minio/blob/master/LICENSE).
+
+MinIO cannot make the determination as to whether your application's usage of MinIO is in compliance with the AGPLv3 license requirements. You should instead rely on your own legal counsel or licensing specialists to audit and ensure your application is in compliance with the licenses of MinIO and all other open-source projects with which your application integrates or interacts. We understand that AGPLv3 licensing is complex and nuanced. It is for that reason we strongly encourage using experts in licensing to make any such determinations around compliance instead of relying on apocryphal or anecdotal advice.
+
+[MinIO Commercial Licensing](https://min.io/pricing) is the best option for applications that trigger AGPLv3 obligations (e.g. open sourcing your application). Applications using MinIO - or any other OSS-licensed code - without validating their usage do so at their own risk.

Makefile

@@ -34,13 +34,12 @@ check-gen: ## check for updated autogenerated files
 
 lint: ## runs golangci-lint suite of linters
 	@echo "Running $@ check"
-	@GO111MODULE=on ${GOPATH}/bin/golangci-lint cache clean
-	@GO111MODULE=on ${GOPATH}/bin/golangci-lint run --build-tags kqueue --timeout=10m --config ./.golangci.yml
+	@${GOPATH}/bin/golangci-lint run --build-tags kqueue --timeout=10m --config ./.golangci.yml
 
 check: test
 test: verifiers build ## builds minio, runs linters, tests
 	@echo "Running unit tests"
-	@GO111MODULE=on CGO_ENABLED=0 go test -tags kqueue ./...
+	@CGO_ENABLED=0 go test -tags kqueue ./...
 
 test-upgrade: build
 	@echo "Running minio upgrade tests"
@@ -60,23 +59,32 @@ test-replication: install ## verify multi site replication
 	@echo "Running tests for replicating three sites"
 	@(env bash $(PWD)/docs/bucket/replication/setup_3site_replication.sh)
 
-test-site-replication: install ## verify automatic site replication
-	@echo "Running tests for automatic site replication of IAM"
-	@(env bash $(PWD)/docs/site-replication/run-multi-site.sh)
+test-site-replication-ldap: install ## verify automatic site replication
+	@echo "Running tests for automatic site replication of IAM (with LDAP)"
+	@(env bash $(PWD)/docs/site-replication/run-multi-site-ldap.sh)
+
+test-site-replication-oidc: install ## verify automatic site replication
+	@echo "Running tests for automatic site replication of IAM (with OIDC)"
+	@(env bash $(PWD)/docs/site-replication/run-multi-site-oidc.sh)
+
+test-site-replication-minio: install ## verify automatic site replication
+	@echo "Running tests for automatic site replication of IAM (with MinIO IDP)"
+	@(env bash $(PWD)/docs/site-replication/run-multi-site-minio-idp.sh)
 
 verify: ## verify minio various setups
 	@echo "Verifying build with race"
-	@GO111MODULE=on CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(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"
-	@GO111MODULE=on CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@CGO_ENABLED=1 go build -race -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
 	@(env bash $(PWD)/buildscripts/verify-healing.sh)
+	@(env bash $(PWD)/buildscripts/unaligned-healing.sh)
 
 build: checks ## builds minio to $(PWD)
 	@echo "Building minio binary to './minio'"
-	@GO111MODULE=on CGO_ENABLED=0 go build -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
+	@CGO_ENABLED=0 go build -tags kqueue -trimpath --ldflags "$(LDFLAGS)" -o $(PWD)/minio 1>/dev/null
 
 hotfix-vars:
 	$(eval LDFLAGS := $(shell MINIO_RELEASE="RELEASE" MINIO_HOTFIX="hotfix.$(shell git rev-parse --short HEAD)" go run buildscripts/gen-ldflags.go $(shell git describe --tags --abbrev=0 | \

README.md

@@ -5,7 +5,7 @@
 
 MinIO is a High Performance Object Storage released under GNU Affero General Public License v3.0. It is API compatible with Amazon S3 cloud storage service. Use MinIO to build high performance infrastructure for machine learning, analytics and application data workloads.
 
-This README provides quickstart instructions on running MinIO on baremetal hardware, including container-based installations. For Kubernetes environments, use the [MinIO Kubernetes Operator](https://github.com/minio/operator/blob/master/README.md).
+This README provides quickstart instructions on running MinIO on bare metal hardware, including container-based installations. For Kubernetes environments, use the [MinIO Kubernetes Operator](https://github.com/minio/operator/blob/master/README.md).
 
 # Container Installation
 
@@ -227,23 +227,25 @@ For example: `export MINIO_SERVER_URL="https://minio.example.net"`
 `mc` provides a modern alternative to UNIX commands like ls, cat, cp, mirror, diff etc. It supports filesystems and Amazon S3 compatible cloud storage services. Follow the MinIO Client [Quickstart Guide](https://docs.min.io/docs/minio-client-quickstart-guide) for further instructions.
 
 # Upgrading MinIO
-MinIO server supports rolling upgrades, i.e. you can update one MinIO instance at a time in a distributed cluster. This allows upgrades with no downtime. Upgrades can be done manually by replacing the binary with the latest release and restarting all servers in a rolling fashion. However, we recommend all our users to use [`mc admin update`](https://docs.min.io/docs/minio-admin-complete-guide.html#update) from the client. This will update all the nodes in the cluster simultaneously and restart them, as shown in the following command from the MinIO client (mc):
+Upgrades require zero downtime in MinIO, all upgrades are non-disruptive, all transactions on MinIO are atomic. So upgrading all the servers simultaneously is the recommended way to upgrade MinIO.
 
+> NOTE: requires internet access to update directly from https://dl.min.io, optionally you can host any mirrors at https://my-artifactory.example.com/minio/
+
+- For deployments that installed the MinIO server binary by hand, use [`mc admin update`](https://docs.min.io/minio/baremetal/reference/minio-mc-admin/mc-admin-update.html)
 ```
 mc admin update <minio alias, e.g., myminio>
 ```
 
-> NOTE: some releases might not allow rolling upgrades, this is always called out in the release notes and it is generally advised to read release notes before upgrading. In such a situation `mc admin update` is the recommended upgrading mechanism to upgrade all servers at once.
+- 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/`.
 
-## Important things to remember during MinIO upgrades
+- 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).
 
-- `mc admin update` will only work if the user running MinIO has write access to the parent directory where the binary is located, for example if the current binary is at `/usr/local/bin/minio`, you would need write access to `/usr/local/bin`.
-- `mc admin update` updates and restarts all servers simultaneously, applications would retry and continue their respective operations upon upgrade.
-- `mc admin update` is disabled in kubernetes/container environments, container environments provide their own mechanisms to rollout of updates.
-- In the case of federated setups `mc admin update` should be run against each cluster individually. Avoid updating `mc` to any new releases until all clusters have been successfully updated.
-- If using `kes` as KMS with MinIO, just replace the binary and restart `kes` more information about `kes` can be found [here](https://github.com/minio/kes/wiki)
-- If using Vault as KMS with MinIO, ensure you have followed the Vault upgrade procedure outlined here: https://www.vaultproject.io/docs/upgrading/index.html
-- If using etcd with MinIO for the federation, ensure you have followed the etcd upgrade procedure outlined here: https://github.com/etcd-io/etcd/blob/master/Documentation/upgrades/upgrading-etcd.md
+## 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.
+- `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.**
 
 # Explore Further
 - [MinIO Erasure Code QuickStart Guide](https://docs.min.io/docs/minio-erasure-code-quickstart-guide)
@@ -257,5 +259,6 @@ mc admin update <minio alias, e.g., myminio>
 Please follow MinIO [Contributor's Guide](https://github.com/minio/minio/blob/master/CONTRIBUTING.md)
 
 # License
-MinIO source is licensed under the GNU AGPLv3 license that can be found in the [LICENSE](https://github.com/minio/minio/blob/master/LICENSE) file.
-MinIO [Documentation](https://github.com/minio/minio/tree/master/docs) © 2021 by MinIO, Inc is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).
+- MinIO source is licensed under the GNU AGPLv3 license that can be found in the [LICENSE](https://github.com/minio/minio/blob/master/LICENSE) file.
+- MinIO [Documentation](https://github.com/minio/minio/tree/master/docs) © 2021 by MinIO, Inc is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).
+- [License Compliance](https://github.com/minio/minio/blob/master/COMPLIANCE.md)

buildscripts/race.sh

@@ -2,6 +2,7 @@
 
 set -e
 
-for d in $(go list ./... | grep -v browser); do
-    CGO_ENABLED=1 go test -v -tags kqueue -race --timeout 100m "$d"
+## 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"
 done

buildscripts/unaligned-healing.sh

@@ -0,0 +1,171 @@
+#!/bin/bash -e
+#
+
+set -E
+set -o pipefail
+set -x
+
+if [ ! -x "$PWD/minio" ]; then
+    echo "minio executable binary not found in current directory"
+    exit 1
+fi
+
+WORK_DIR="$PWD/.verify-$RANDOM"
+MINIO_CONFIG_DIR="$WORK_DIR/.minio"
+MINIO_OLD=( "$PWD/minio.RELEASE.2021-11-24T23-19-33Z" --config-dir "$MINIO_CONFIG_DIR" server )
+MINIO=( "$PWD/minio" --config-dir "$MINIO_CONFIG_DIR" server )
+
+function download_old_release() {
+    if [ ! -f minio.RELEASE.2021-11-24T23-19-33Z ]; then
+	curl --silent -O https://dl.minio.io/server/minio/release/linux-amd64/archive/minio.RELEASE.2021-11-24T23-19-33Z
+	chmod a+x minio.RELEASE.2021-11-24T23-19-33Z
+    fi
+}
+
+function start_minio_16drive() {
+    start_port=$1
+
+    export MINIO_ROOT_USER=minio
+    export MINIO_ROOT_PASSWORD=minio123
+    export MC_HOST_minio="http://minio:minio123@127.0.0.1:${start_port}/"
+    unset MINIO_KMS_AUTO_ENCRYPTION # do not auto-encrypt objects
+    export _MINIO_SHARD_DISKTIME_DELTA="5s" # do not change this as its needed for tests
+
+    MC_BUILD_DIR="mc-$RANDOM"
+    if ! git clone --quiet https://github.com/minio/mc "$MC_BUILD_DIR"; then
+	echo "failed to download https://github.com/minio/mc"
+	purge "${MC_BUILD_DIR}"
+	exit 1
+    fi
+
+    (cd "${MC_BUILD_DIR}" && go build -o "$WORK_DIR/mc")
+
+    # remove mc source.
+    purge "${MC_BUILD_DIR}"
+
+    "${MINIO_OLD[@]}" --address ":$start_port" "${WORK_DIR}/xl{1...16}" > "${WORK_DIR}/server1.log" 2>&1 &
+    pid=$!
+    disown $pid
+    sleep 30
+
+    if ! ps -p ${pid} 1>&2 >/dev/null; then
+	echo "server1 log:"
+	cat "${WORK_DIR}/server1.log"
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
+    shred --iterations=1 --size=5241856 - 1>"${WORK_DIR}/unaligned" 2>/dev/null
+    "${WORK_DIR}/mc" mb minio/healing-shard-bucket --quiet
+    "${WORK_DIR}/mc" cp \
+		     "${WORK_DIR}/unaligned" \
+		     minio/healing-shard-bucket/unaligned \
+		     --disable-multipart --quiet
+
+    ## "unaligned" object name gets consistently distributed
+    ## to disks in following distribution order
+    ##
+    ## NOTE: if you change the name make sure to change the
+    ## distribution order present here
+    ##
+    ## [15, 16, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14]
+
+    ## make sure to remove the "last" data shard
+    rm -rf "${WORK_DIR}/xl14/healing-shard-bucket/unaligned"
+    sleep 10
+    ## Heal the shard
+    "${WORK_DIR}/mc" admin heal --quiet --recursive minio/healing-shard-bucket
+    ## then remove any other data shard let's pick first disk
+    ## - 1st data shard.
+    rm -rf "${WORK_DIR}/xl3/healing-shard-bucket/unaligned"
+    sleep 10
+
+    go build ./docs/debugging/s3-check-md5/
+    if ! ./s3-check-md5 \
+	 -debug \
+	 -access-key minio \
+	 -secret-key minio123 \
+	 -endpoint http://127.0.0.1:${start_port}/ 2>&1 | grep CORRUPTED; then
+	echo "server1 log:"
+	cat "${WORK_DIR}/server1.log"
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
+    pkill minio
+    sleep 3
+
+    "${MINIO[@]}" --address ":$start_port" "${WORK_DIR}/xl{1...16}" > "${WORK_DIR}/server1.log" 2>&1 &
+    pid=$!
+    disown $pid
+    sleep 30
+
+    if ! ps -p ${pid} 1>&2 >/dev/null; then
+	echo "server1 log:"
+	cat "${WORK_DIR}/server1.log"
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
+    if ! ./s3-check-md5 \
+	 -debug \
+	 -access-key minio \
+	 -secret-key minio123 \
+	 -endpoint http://127.0.0.1:${start_port}/ 2>&1 | grep INTACT; then
+	echo "server1 log:"
+	cat "${WORK_DIR}/server1.log"
+	echo "FAILED"
+	mkdir -p inspects
+	(cd inspects; "${WORK_DIR}/mc" admin inspect minio/healing-shard-bucket/unaligned/**)
+
+	"${WORK_DIR}/mc" mb play/inspects
+	"${WORK_DIR}/mc" mirror inspects play/inspects
+
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
+    "${WORK_DIR}/mc" admin heal --quiet --recursive minio/healing-shard-bucket
+
+    if ! ./s3-check-md5 \
+	 -debug \
+	 -access-key minio \
+	 -secret-key minio123 \
+	 -endpoint http://127.0.0.1:${start_port}/ 2>&1 | grep INTACT; then
+	echo "server1 log:"
+	cat "${WORK_DIR}/server1.log"
+	echo "FAILED"
+	mkdir -p inspects
+	(cd inspects; "${WORK_DIR}/mc" admin inspect minio/healing-shard-bucket/unaligned/**)
+
+	"${WORK_DIR}/mc" mb play/inspects
+	"${WORK_DIR}/mc" mirror inspects play/inspects
+
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
+    pkill minio
+    sleep 3
+}
+
+function main() {
+    download_old_release
+
+    start_port=$(shuf -i 10000-65000 -n 1)
+
+    start_minio_16drive ${start_port}
+}
+
+function purge()
+{
+    rm -rf "$1"
+}
+
+( main "$@" )
+rv=$?
+purge "$WORK_DIR"
+exit "$rv"

buildscripts/verify-healing.sh

@@ -21,51 +21,70 @@ function start_minio_3_node() {
     start_port=$2
     args=""
     for i in $(seq 1 3); do
-        args="$args http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/1/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/2/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/3/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/4/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/5/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/6/"
+	args="$args http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/1/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/2/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/3/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/4/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/5/ http://127.0.0.1:$[$start_port+$i]${WORK_DIR}/$i/6/"
     done
 
     "${MINIO[@]}" --address ":$[$start_port+1]" $args > "${WORK_DIR}/dist-minio-server1.log" 2>&1 &
-    disown $!
+    pid1=$!
+    disown ${pid1}
 
     "${MINIO[@]}" --address ":$[$start_port+2]" $args > "${WORK_DIR}/dist-minio-server2.log" 2>&1 &
-    disown $!
+    pid2=$!
+    disown $pid2
 
     "${MINIO[@]}" --address ":$[$start_port+3]" $args > "${WORK_DIR}/dist-minio-server3.log" 2>&1 &
-    disown $!
+    pid3=$!
+    disown $pid3
 
     sleep "$1"
-    if [ "$(pgrep -c minio)" -ne 3 ]; then
-        for i in $(seq 1 3); do
-            echo "server$i log:"
-            cat "${WORK_DIR}/dist-minio-server$i.log"
-        done
-        echo "FAILED"
-        purge "$WORK_DIR"
-        exit 1
+
+    if ! ps -p $pid1 1>&2 > /dev/null; then
+	echo "server1 log:"
+	cat "${WORK_DIR}/dist-minio-server1.log"
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
     fi
+
+    if ! ps -p $pid2 1>&2 > /dev/null; then
+	echo "server2 log:"
+	cat "${WORK_DIR}/dist-minio-server2.log"
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
+    if ! ps -p $pid3 1>&2 > /dev/null; then
+	echo "server3 log:"
+	cat "${WORK_DIR}/dist-minio-server3.log"
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
+    fi
+
     if ! pkill minio; then
-        for i in $(seq 1 3); do
-            echo "server$i log:"
-            cat "${WORK_DIR}/dist-minio-server$i.log"
-        done
-        echo "FAILED"
-        purge "$WORK_DIR"
-        exit 1
+	for i in $(seq 1 3); do
+	    echo "server$i log:"
+	    cat "${WORK_DIR}/dist-minio-server$i.log"
+	done
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
     fi
 
     sleep 1;
     if pgrep minio; then
-        # forcibly killing, to proceed further properly.
-        if ! pkill -9 minio; then
-            echo "no minio process running anymore, proceed."
-        fi
+	# forcibly killing, to proceed further properly.
+	if ! pkill -9 minio; then
+	    echo "no minio process running anymore, proceed."
+	fi
     fi
 }
 
 
 function check_online() {
     if grep -q 'Unable to initialize sub-systems' ${WORK_DIR}/dist-minio-*.log; then
-        echo "1"
+	echo "1"
     fi
 }
 
@@ -96,14 +115,14 @@ function perform_test() {
 
     rv=$(check_online)
     if [ "$rv" == "1" ]; then
-        for i in $(seq 1 3); do
-            echo "server$i log:"
-            cat "${WORK_DIR}/dist-minio-server$i.log"
-        done
-        pkill -9 minio
-        echo "FAILED"
-        purge "$WORK_DIR"
-        exit 1
+	for i in $(seq 1 3); do
+	    echo "server$i log:"
+	    cat "${WORK_DIR}/dist-minio-server$i.log"
+	done
+	pkill -9 minio
+	echo "FAILED"
+	purge "$WORK_DIR"
+	exit 1
     fi
 }
 

cmd/admin-bucket-handlers.go

@@ -65,7 +65,8 @@ func (a adminAPIHandlers) PutBucketQuotaConfigHandler(w http.ResponseWriter, r *
 		return
 	}
 
-	if _, err = parseBucketQuota(bucket, data); err != nil {
+	quotaConfig, err := parseBucketQuota(bucket, data)
+	if err != nil {
 		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
 		return
 	}
@@ -75,6 +76,21 @@ func (a adminAPIHandlers) PutBucketQuotaConfigHandler(w http.ResponseWriter, r *
 		return
 	}
 
+	bucketMeta := madmin.SRBucketMeta{
+		Type:   madmin.SRBucketMetaTypeQuotaConfig,
+		Bucket: bucket,
+		Quota:  data,
+	}
+	if quotaConfig.Quota == 0 {
+		bucketMeta.Quota = nil
+	}
+
+	// Call site replication hook.
+	if err = globalSiteReplicationSys.BucketMetaHook(ctx, bucketMeta); err != nil {
+		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+		return
+	}
+
 	// Write success response.
 	writeSuccessResponseHeadersOnly(w)
 }
@@ -155,7 +171,7 @@ func (a adminAPIHandlers) SetRemoteTargetHandler(w http.ResponseWriter, r *http.
 		return
 	}
 	var target madmin.BucketTarget
-	var json = jsoniter.ConfigCompatibleWithStandardLibrary
+	json := jsoniter.ConfigCompatibleWithStandardLibrary
 	if err = json.Unmarshal(reqBytes, &target); err != nil {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminConfigBadJSON, err), r.URL)
 		return

cmd/admin-handler-utils.go

@@ -88,8 +88,26 @@ func toAdminAPIErr(ctx context.Context, err error) APIError {
 		}
 	case SRError:
 		apiErr = errorCodes.ToAPIErrWithErr(e.Code, e.Cause)
+	case decomError:
+		apiErr = APIError{
+			Code:           "XMinioDecommissionNotAllowed",
+			Description:    e.Err,
+			HTTPStatusCode: http.StatusBadRequest,
+		}
 	default:
 		switch {
+		case errors.Is(err, errDecommissionAlreadyRunning):
+			apiErr = APIError{
+				Code:           "XMinioDecommissionNotAllowed",
+				Description:    err.Error(),
+				HTTPStatusCode: http.StatusBadRequest,
+			}
+		case errors.Is(err, errDecommissionComplete):
+			apiErr = APIError{
+				Code:           "XMinioDecommissionNotAllowed",
+				Description:    err.Error(),
+				HTTPStatusCode: http.StatusBadRequest,
+			}
 		case errors.Is(err, errConfigNotFound):
 			apiErr = APIError{
 				Code:           "XMinioConfigError",
@@ -102,6 +120,18 @@ func toAdminAPIErr(ctx context.Context, err error) APIError {
 				Description:    err.Error(),
 				HTTPStatusCode: http.StatusForbidden,
 			}
+		case errors.Is(err, errIAMServiceAccount):
+			apiErr = APIError{
+				Code:           "XMinioIAMServiceAccount",
+				Description:    err.Error(),
+				HTTPStatusCode: http.StatusBadRequest,
+			}
+		case errors.Is(err, errIAMServiceAccountUsed):
+			apiErr = APIError{
+				Code:           "XMinioIAMServiceAccountUsed",
+				Description:    err.Error(),
+				HTTPStatusCode: http.StatusBadRequest,
+			}
 		case errors.Is(err, errIAMNotInitialized):
 			apiErr = APIError{
 				Code:           "XMinioIAMNotInitialized",

cmd/admin-handlers-config-kv.go

@@ -74,6 +74,10 @@ func (a adminAPIHandlers) DelConfigKVHandler(w http.ResponseWriter, r *http.Requ
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
+	if err = validateConfig(cfg); err != nil {
+		writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminConfigBadJSON), err.Error(), r.URL)
+		return
+	}
 
 	if err = saveServerConfig(ctx, objectAPI, cfg); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
@@ -170,7 +174,7 @@ func (a adminAPIHandlers) GetConfigKVHandler(w http.ResponseWriter, r *http.Requ
 
 	cfg := globalServerConfig.Clone()
 	vars := mux.Vars(r)
-	var buf = &bytes.Buffer{}
+	buf := &bytes.Buffer{}
 	cw := config.NewConfigWriteTo(cfg, vars["key"])
 	if _, err := cw.WriteTo(buf); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)

cmd/admin-handlers-pools.go

@@ -0,0 +1,179 @@
+// Copyright (c) 2015-2021 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"
+	"net/http"
+
+	"github.com/gorilla/mux"
+	"github.com/minio/minio/internal/logger"
+	iampolicy "github.com/minio/pkg/iam/policy"
+)
+
+func (a adminAPIHandlers) StartDecommission(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "StartDecommission")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.DecommissionAdminAction)
+	if objectAPI == nil {
+		return
+	}
+
+	// Legacy args style such as non-ellipses style is not supported with this API.
+	if globalEndpoints.Legacy() {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	pools, ok := objectAPI.(*erasureServerPools)
+	if !ok {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	vars := mux.Vars(r)
+	v := vars["pool"]
+
+	idx := globalEndpoints.GetPoolIdx(v)
+	if idx == -1 {
+		// We didn't find any matching pools, invalid input
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, errInvalidArgument), r.URL)
+		return
+	}
+
+	if err := pools.Decommission(r.Context(), idx); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+}
+
+func (a adminAPIHandlers) CancelDecommission(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "CancelDecommission")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.DecommissionAdminAction)
+	if objectAPI == nil {
+		return
+	}
+
+	// Legacy args style such as non-ellipses style is not supported with this API.
+	if globalEndpoints.Legacy() {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	pools, ok := objectAPI.(*erasureServerPools)
+	if !ok {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	vars := mux.Vars(r)
+	v := vars["pool"]
+
+	idx := globalEndpoints.GetPoolIdx(v)
+	if idx == -1 {
+		// We didn't find any matching pools, invalid input
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, errInvalidArgument), r.URL)
+		return
+	}
+
+	if err := pools.DecommissionCancel(ctx, idx); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+}
+
+func (a adminAPIHandlers) StatusPool(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "StatusPool")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.ServerInfoAdminAction, iampolicy.DecommissionAdminAction)
+	if objectAPI == nil {
+		return
+	}
+
+	// Legacy args style such as non-ellipses style is not supported with this API.
+	if globalEndpoints.Legacy() {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	pools, ok := objectAPI.(*erasureServerPools)
+	if !ok {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	vars := mux.Vars(r)
+	v := vars["pool"]
+
+	idx := globalEndpoints.GetPoolIdx(v)
+	if idx == -1 {
+		// We didn't find any matching pools, invalid input
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, errInvalidArgument), r.URL)
+		return
+	}
+
+	status, err := pools.Status(r.Context(), idx)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	logger.LogIf(r.Context(), json.NewEncoder(w).Encode(&status))
+}
+
+func (a adminAPIHandlers) ListPools(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "ListPools")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.ServerInfoAdminAction, iampolicy.DecommissionAdminAction)
+	if objectAPI == nil {
+		return
+	}
+
+	// Legacy args style such as non-ellipses style is not supported with this API.
+	if globalEndpoints.Legacy() {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	pools, ok := objectAPI.(*erasureServerPools)
+	if !ok {
+		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
+		return
+	}
+
+	poolsStatus := make([]PoolStatus, len(globalEndpoints))
+	for idx := range globalEndpoints {
+		status, err := pools.Status(r.Context(), idx)
+		if err != nil {
+			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+			return
+		}
+		poolsStatus[idx] = status
+	}
+
+	logger.LogIf(r.Context(), json.NewEncoder(w).Encode(poolsStatus))
+}

cmd/admin-handlers-site-replication.go

@@ -45,8 +45,7 @@ func (a adminAPIHandlers) SiteReplicationAdd(w http.ResponseWriter, r *http.Requ
 	}
 
 	var sites []madmin.PeerSite
-	err := parseJSONBody(ctx, r.Body, &sites, cred.SecretKey)
-	if err != nil {
+	if err := parseJSONBody(ctx, r.Body, &sites, cred.SecretKey); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
@@ -67,12 +66,12 @@ func (a adminAPIHandlers) SiteReplicationAdd(w http.ResponseWriter, r *http.Requ
 	writeSuccessResponseJSON(w, body)
 }
 
-// SRInternalJoin - PUT /minio/admin/v3/site-replication/join
+// SRPeerJoin - PUT /minio/admin/v3/site-replication/join
 //
 // used internally to tell current cluster to enable SR with
 // the provided peer clusters and service account.
-func (a adminAPIHandlers) SRInternalJoin(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "SRInternalJoin")
+func (a adminAPIHandlers) SRPeerJoin(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SRPeerJoin")
 
 	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
 
@@ -81,22 +80,22 @@ func (a adminAPIHandlers) SRInternalJoin(w http.ResponseWriter, r *http.Request)
 		return
 	}
 
-	var joinArg madmin.SRInternalJoinReq
+	var joinArg madmin.SRPeerJoinReq
 	if err := parseJSONBody(ctx, r.Body, &joinArg, cred.SecretKey); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
 
-	if err := globalSiteReplicationSys.InternalJoinReq(ctx, joinArg); err != nil {
+	if err := globalSiteReplicationSys.PeerJoinReq(ctx, joinArg); err != nil {
 		logger.LogIf(ctx, err)
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
 }
 
-// SRInternalBucketOps - PUT /minio/admin/v3/site-replication/bucket-ops?bucket=x&operation=y
-func (a adminAPIHandlers) SRInternalBucketOps(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "SRInternalBucketOps")
+// SRPeerBucketOps - PUT /minio/admin/v3/site-replication/bucket-ops?bucket=x&operation=y
+func (a adminAPIHandlers) SRPeerBucketOps(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SRPeerBucketOps")
 
 	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
 
@@ -134,12 +133,11 @@ func (a adminAPIHandlers) SRInternalBucketOps(w http.ResponseWriter, r *http.Req
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
-
 }
 
-// SRInternalReplicateIAMItem - PUT /minio/admin/v3/site-replication/iam-item
-func (a adminAPIHandlers) SRInternalReplicateIAMItem(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "SRInternalReplicateIAMItem")
+// SRPeerReplicateIAMItem - PUT /minio/admin/v3/site-replication/iam-item
+func (a adminAPIHandlers) SRPeerReplicateIAMItem(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SRPeerReplicateIAMItem")
 
 	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
 
@@ -179,6 +177,10 @@ func (a adminAPIHandlers) SRInternalReplicateIAMItem(w http.ResponseWriter, r *h
 		err = globalSiteReplicationSys.PeerPolicyMappingHandler(ctx, item.PolicyMapping)
 	case madmin.SRIAMItemSTSAcc:
 		err = globalSiteReplicationSys.PeerSTSAccHandler(ctx, item.STSCredential)
+	case madmin.SRIAMItemIAMUser:
+		err = globalSiteReplicationSys.PeerIAMUserChangeHandler(ctx, item.IAMUser)
+	case madmin.SRIAMItemGroupInfo:
+		err = globalSiteReplicationSys.PeerGroupInfoChangeHandler(ctx, item.GroupInfo)
 	}
 	if err != nil {
 		logger.LogIf(ctx, err)
@@ -187,9 +189,9 @@ func (a adminAPIHandlers) SRInternalReplicateIAMItem(w http.ResponseWriter, r *h
 	}
 }
 
-// SRInternalReplicateBucketItem - PUT /minio/admin/v3/site-replication/bucket-meta
-func (a adminAPIHandlers) SRInternalReplicateBucketItem(w http.ResponseWriter, r *http.Request) {
-	ctx := newContext(r, w, "SRInternalReplicateBucketItem")
+// SRPeerReplicateBucketItem - PUT /minio/admin/v3/site-replication/bucket-meta
+func (a adminAPIHandlers) SRPeerReplicateBucketItem(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SRPeerReplicateBucketItem")
 
 	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
 
@@ -223,6 +225,20 @@ func (a adminAPIHandlers) SRInternalReplicateBucketItem(w http.ResponseWriter, r
 				err = globalSiteReplicationSys.PeerBucketPolicyHandler(ctx, item.Bucket, bktPolicy)
 			}
 		}
+	case madmin.SRBucketMetaTypeQuotaConfig:
+		if item.Quota == nil {
+			err = globalSiteReplicationSys.PeerBucketQuotaConfigHandler(ctx, item.Bucket, nil)
+		} else {
+			quotaConfig, err := parseBucketQuota(item.Bucket, item.Quota)
+			if err != nil {
+				writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+				return
+			}
+			if err = globalSiteReplicationSys.PeerBucketQuotaConfigHandler(ctx, item.Bucket, quotaConfig); err != nil {
+				writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
+				return
+			}
+		}
 	case madmin.SRBucketMetaTypeTags:
 		err = globalSiteReplicationSys.PeerBucketTaggingHandler(ctx, item.Bucket, item.Tags)
 	case madmin.SRBucketMetaTypeObjectLockConfig:
@@ -272,7 +288,7 @@ func (a adminAPIHandlers) SiteReplicationInfo(w http.ResponseWriter, r *http.Req
 	}
 }
 
-func (a adminAPIHandlers) SRInternalGetIDPSettings(w http.ResponseWriter, r *http.Request) {
+func (a adminAPIHandlers) SRPeerGetIDPSettings(w http.ResponseWriter, r *http.Request) {
 	ctx := newContext(r, w, "SiteReplicationGetIDPSettings")
 
 	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
@@ -311,3 +327,104 @@ func parseJSONBody(ctx context.Context, body io.Reader, v interface{}, encryptio
 
 	return json.Unmarshal(data, v)
 }
+
+// SiteReplicationStatus - GET /minio/admin/v3/site-replication/status
+func (a adminAPIHandlers) SiteReplicationStatus(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SiteReplicationStatus")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.SiteReplicationInfoAction)
+	if objectAPI == nil {
+		return
+	}
+
+	info, err := globalSiteReplicationSys.SiteReplicationStatus(ctx, objectAPI)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	if err = json.NewEncoder(w).Encode(info); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+}
+
+// SiteReplicationMetaInfo - GET /minio/admin/v3/site-replication/metainfo
+func (a adminAPIHandlers) SiteReplicationMetaInfo(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SiteReplicationMetaInfo")
+
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.SiteReplicationInfoAction)
+	if objectAPI == nil {
+		return
+	}
+
+	info, err := globalSiteReplicationSys.SiteReplicationMetaInfo(ctx, objectAPI)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	if err = json.NewEncoder(w).Encode(info); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+}
+
+// SiteReplicationEdit - PUT /minio/admin/v3/site-replication/edit
+func (a adminAPIHandlers) SiteReplicationEdit(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SiteReplicationEdit")
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, cred := validateAdminReq(ctx, w, r, iampolicy.SiteReplicationAddAction)
+	if objectAPI == nil {
+		return
+	}
+	var site madmin.PeerInfo
+	err := parseJSONBody(ctx, r.Body, &site, cred.SecretKey)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+	status, err := globalSiteReplicationSys.EditPeerCluster(ctx, site)
+	if err != nil {
+		logger.LogIf(ctx, err)
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+	body, err := json.Marshal(status)
+	if err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	writeSuccessResponseJSON(w, body)
+}
+
+// SRPeerEdit - PUT /minio/admin/v3/site-replication/peer/edit
+//
+// used internally to tell current cluster to update endpoint for peer
+func (a adminAPIHandlers) SRPeerEdit(w http.ResponseWriter, r *http.Request) {
+	ctx := newContext(r, w, "SRPeerEdit")
+	defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
+
+	objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.SiteReplicationAddAction)
+	if objectAPI == nil {
+		return
+	}
+
+	var pi madmin.PeerInfo
+	if err := parseJSONBody(ctx, r.Body, &pi, ""); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	if err := globalSiteReplicationSys.PeerEditReq(ctx, pi); err != nil {
+		logger.LogIf(ctx, err)
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+}

cmd/admin-handlers-users.go

@@ -58,21 +58,24 @@ func (a adminAPIHandlers) RemoveUser(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	if err := globalIAMSys.DeleteUser(ctx, accessKey); err != nil {
+	if err := globalIAMSys.DeleteUser(ctx, accessKey, true); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
 
-	// Notify all other MinIO peers to delete user.
-	for _, nerr := range globalNotificationSys.DeleteUser(accessKey) {
-		if nerr.Err != nil {
-			logger.GetReqInfo(ctx).SetTags("peerAddress", nerr.Host.String())
-			logger.LogIf(ctx, nerr.Err)
-		}
+	if err := globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+		Type: madmin.SRIAMItemIAMUser,
+		IAMUser: &madmin.SRIAMUser{
+			AccessKey:   accessKey,
+			IsDeleteReq: true,
+		},
+	}); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
 	}
 }
 
-// ListUsers - GET /minio/admin/v3/list-users?bucket={bucket}
+// ListBucketUsers - GET /minio/admin/v3/list-users?bucket={bucket}
 func (a adminAPIHandlers) ListBucketUsers(w http.ResponseWriter, r *http.Request) {
 	ctx := newContext(r, w, "ListBucketUsers")
 
@@ -228,11 +231,20 @@ func (a adminAPIHandlers) UpdateGroupMembers(w http.ResponseWriter, r *http.Requ
 	} else {
 		err = globalIAMSys.AddUsersToGroup(ctx, updReq.Group, updReq.Members)
 	}
-
 	if err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
+
+	if err := globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+		Type: madmin.SRIAMItemGroupInfo,
+		GroupInfo: &madmin.SRGroupInfo{
+			UpdateReq: updReq,
+		},
+	}); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
 }
 
 // GetGroup - /minio/admin/v3/group?group=mygroup1
@@ -318,6 +330,20 @@ func (a adminAPIHandlers) SetGroupStatus(w http.ResponseWriter, r *http.Request)
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
+
+	if err := globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+		Type: madmin.SRIAMItemGroupInfo,
+		GroupInfo: &madmin.SRGroupInfo{
+			UpdateReq: madmin.GroupAddRemove{
+				Group:    group,
+				Status:   madmin.GroupStatus(status),
+				IsRemove: false,
+			},
+		},
+	}); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
 }
 
 // SetUserStatus - PUT /minio/admin/v3/set-user-status?accessKey=<access_key>&status=[enabled|disabled]
@@ -335,7 +361,7 @@ func (a adminAPIHandlers) SetUserStatus(w http.ResponseWriter, r *http.Request)
 	accessKey := vars["accessKey"]
 	status := vars["status"]
 
-	// This API is not allowed to lookup accessKey user status
+	// This API is not allowed to lookup master access key user status
 	if accessKey == globalActiveCred.AccessKey {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
 		return
@@ -345,6 +371,20 @@ func (a adminAPIHandlers) SetUserStatus(w http.ResponseWriter, r *http.Request)
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
+
+	if err := globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+		Type: madmin.SRIAMItemIAMUser,
+		IAMUser: &madmin.SRIAMUser{
+			AccessKey:   accessKey,
+			IsDeleteReq: false,
+			UserReq: &madmin.AddOrUpdateUserReq{
+				Status: madmin.AccountStatus(status),
+			},
+		},
+	}); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
 }
 
 // AddUser - PUT /minio/admin/v3/add-user?accessKey=<access_key>
@@ -424,14 +464,26 @@ func (a adminAPIHandlers) AddUser(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	var uinfo madmin.UserInfo
-	if err = json.Unmarshal(configBytes, &uinfo); err != nil {
+	var ureq madmin.AddOrUpdateUserReq
+	if err = json.Unmarshal(configBytes, &ureq); err != nil {
 		logger.LogIf(ctx, err)
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminConfigBadJSON), r.URL)
 		return
 	}
 
-	if err = globalIAMSys.CreateUser(ctx, accessKey, uinfo); err != nil {
+	if err = globalIAMSys.CreateUser(ctx, accessKey, ureq); err != nil {
+		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
+		return
+	}
+
+	if err := globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
+		Type: madmin.SRIAMItemIAMUser,
+		IAMUser: &madmin.SRIAMUser{
+			AccessKey:   accessKey,
+			IsDeleteReq: false,
+			UserReq:     &ureq,
+		},
+	}); err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
@@ -593,14 +645,9 @@ func (a adminAPIHandlers) AddServiceAccount(w http.ResponseWriter, r *http.Reque
 		return
 	}
 
-	// Call hook for cluster-replication.
-	//
-	// FIXME: This wont work in an OpenID situation as the parent credential
-	// may not be present on peer clusters to provide inherited policies.
-	// Also, we should not be replicating root user's service account - as
-	// they are not authenticated by a common external IDP, so we skip when
-	// opts.ldapUser == "".
-	if _, isLDAPAccount := opts.claims[ldapUserN]; isLDAPAccount {
+	// 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{
@@ -621,7 +668,7 @@ func (a adminAPIHandlers) AddServiceAccount(w http.ResponseWriter, r *http.Reque
 		}
 	}
 
-	var createResp = madmin.AddServiceAccountResp{
+	createResp := madmin.AddServiceAccountResp{
 		Credentials: madmin.Credentials{
 			AccessKey: newCred.AccessKey,
 			SecretKey: newCred.SecretKey,
@@ -724,14 +771,8 @@ func (a adminAPIHandlers) UpdateServiceAccount(w http.ResponseWriter, r *http.Re
 		return
 	}
 
-	// Call site replication hook. Only LDAP accounts are supported for
-	// replication operations.
-	svcAccClaims, err := globalIAMSys.GetClaimsForSvcAcc(ctx, accessKey)
-	if err != nil {
-		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-		return
-	}
-	if _, isLDAPAccount := svcAccClaims[ldapUserN]; isLDAPAccount {
+	// Call site replication hook - non-root user accounts are replicated.
+	if svcAccount.ParentUser != globalActiveCred.AccessKey {
 		err = globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
 			Type: madmin.SRIAMItemSvcAcc,
 			SvcAccChange: &madmin.SRSvcAccChange{
@@ -748,6 +789,7 @@ func (a adminAPIHandlers) UpdateServiceAccount(w http.ResponseWriter, r *http.Re
 			return
 		}
 	}
+
 	writeSuccessNoContent(w)
 }
 
@@ -802,18 +844,15 @@ func (a adminAPIHandlers) InfoServiceAccount(w http.ResponseWriter, r *http.Requ
 
 	var svcAccountPolicy iampolicy.Policy
 
-	impliedPolicy := policy == nil
-
-	// If policy is empty, check for policy of the parent user
-	if !impliedPolicy {
-		svcAccountPolicy = svcAccountPolicy.Merge(*policy)
+	if policy != nil {
+		svcAccountPolicy = *policy
 	} else {
 		policiesNames, err := globalIAMSys.PolicyDBGet(svcAccount.ParentUser, false)
 		if err != nil {
 			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 			return
 		}
-		svcAccountPolicy = svcAccountPolicy.Merge(globalIAMSys.GetCombinedPolicy(policiesNames...))
+		svcAccountPolicy = globalIAMSys.GetCombinedPolicy(policiesNames...)
 	}
 
 	policyJSON, err := json.MarshalIndent(svcAccountPolicy, "", " ")
@@ -822,10 +861,10 @@ func (a adminAPIHandlers) InfoServiceAccount(w http.ResponseWriter, r *http.Requ
 		return
 	}
 
-	var infoResp = madmin.InfoServiceAccountResp{
+	infoResp := madmin.InfoServiceAccountResp{
 		ParentUser:    svcAccount.ParentUser,
 		AccountStatus: svcAccount.Status,
-		ImpliedPolicy: impliedPolicy,
+		ImpliedPolicy: policy == nil,
 		Policy:        string(policyJSON),
 	}
 
@@ -899,7 +938,7 @@ func (a adminAPIHandlers) ListServiceAccounts(w http.ResponseWriter, r *http.Req
 		serviceAccountsNames = append(serviceAccountsNames, svc.AccessKey)
 	}
 
-	var listResp = madmin.ListServiceAccountsResp{
+	listResp := madmin.ListServiceAccountsResp{
 		Accounts: serviceAccountsNames,
 	}
 
@@ -971,28 +1010,14 @@ func (a adminAPIHandlers) DeleteServiceAccount(w http.ResponseWriter, r *http.Re
 		}
 	}
 
-	// Save svc acc claims before deletion (for site replication hook).
-	svcAccClaims, err := globalIAMSys.GetClaimsForSvcAcc(ctx, serviceAccount)
-	if err != nil && err != errNoSuchServiceAccount {
-		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
-		return
-	}
-
-	err = globalIAMSys.DeleteServiceAccount(ctx, serviceAccount)
+	err = globalIAMSys.DeleteServiceAccount(ctx, serviceAccount, true)
 	if err != nil {
 		writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
 		return
 	}
-	for _, nerr := range globalNotificationSys.DeleteServiceAccount(serviceAccount) {
-		if nerr.Err != nil {
-			logger.GetReqInfo(ctx).SetTags("peerAddress", nerr.Host.String())
-			logger.LogIf(ctx, nerr.Err)
-		}
-	}
 
-	// Call site replication hook. Only LDAP accounts are supported for
-	// replication operations.
-	if _, isLDAPAccount := svcAccClaims[ldapUserN]; isLDAPAccount {
+	// Call site replication hook - non-root user accounts are replicated.
+	if svcAccount.ParentUser != globalActiveCred.AccessKey {
 		err = globalSiteReplicationSys.IAMChangeHook(ctx, madmin.SRIAMItem{
 			Type: madmin.SRIAMItemSvcAcc,
 			SvcAccChange: &madmin.SRSvcAccChange{
@@ -1072,11 +1097,7 @@ func (a adminAPIHandlers) AccountInfoHandler(w http.ResponseWriter, r *http.Requ
 	var err error
 	if !globalIsGateway {
 		// Load the latest calculated data usage
-		dataUsageInfo, err = loadDataUsageFromBackend(ctx, objectAPI)
-		if err != nil {
-			// log the error, continue with the accounting response
-			logger.LogIf(ctx, err)
-		}
+		dataUsageInfo, _ = loadDataUsageFromBackend(ctx, objectAPI)
 	}
 
 	// If etcd, dns federation configured list buckets from etcd.
@@ -1145,11 +1166,7 @@ func (a adminAPIHandlers) AccountInfoHandler(w http.ResponseWriter, r *http.Requ
 			// Fetch the prefix usage of the current bucket
 			var prefixUsage map[string]uint64
 			if enablePrefixUsage {
-				if pu, err := loadPrefixUsageFromBackend(ctx, objectAPI, bucket.Name); err == nil {
-					prefixUsage = pu
-				} else {
-					logger.LogIf(ctx, err)
-				}
+				prefixUsage, _ = loadPrefixUsageFromBackend(ctx, objectAPI, bucket.Name)
 			}
 
 			lcfg, _ := globalBucketObjectLockSys.Get(bucket.Name)
@@ -1265,7 +1282,7 @@ func (a adminAPIHandlers) ListBucketPolicies(w http.ResponseWriter, r *http.Requ
 		return
 	}
 
-	var newPolicies = make(map[string]iampolicy.Policy)
+	newPolicies := make(map[string]iampolicy.Policy)
 	for name, p := range policies {
 		_, err = json.Marshal(p)
 		if err != nil {
@@ -1297,7 +1314,7 @@ func (a adminAPIHandlers) ListCannedPolicies(w http.ResponseWriter, r *http.Requ
 		return
 	}
 
-	var newPolicies = make(map[string]iampolicy.Policy)
+	newPolicies := make(map[string]iampolicy.Policy)
 	for name, p := range policies {
 		_, err = json.Marshal(p)
 		if err != nil {

cmd/admin-handlers-users_test.go

@@ -18,8 +18,15 @@
 package cmd
 
 import (
+	"bytes"
 	"context"
+	"crypto/sha256"
+	"encoding/hex"
+	"encoding/json"
 	"fmt"
+	"io/ioutil"
+	"net/http"
+	"net/url"
 	"os"
 	"strings"
 	"testing"
@@ -29,7 +36,9 @@ import (
 	minio "github.com/minio/minio-go/v7"
 	"github.com/minio/minio-go/v7/pkg/credentials"
 	cr "github.com/minio/minio-go/v7/pkg/credentials"
+	"github.com/minio/minio-go/v7/pkg/s3utils"
 	"github.com/minio/minio-go/v7/pkg/set"
+	"github.com/minio/minio-go/v7/pkg/signer"
 	"github.com/minio/minio/internal/auth"
 )
 
@@ -177,6 +186,7 @@ func TestIAMInternalIDPServerSuite(t *testing.T) {
 
 				suite.SetUpSuite(c)
 				suite.TestUserCreate(c)
+				suite.TestUserPolicyEscalationBug(c)
 				suite.TestPolicyCreate(c)
 				suite.TestCannedPolicies(c)
 				suite.TestGroupAddRemove(c)
@@ -222,6 +232,24 @@ func (s *TestSuiteIAM) TestUserCreate(c *check) {
 		c.Fatalf("user could not create bucket: %v", err)
 	}
 
+	// 3.10. Check that user's password can be updated.
+	_, newSecretKey := mustGenerateCredentials(c)
+	err = s.adm.SetUser(ctx, accessKey, newSecretKey, madmin.AccountEnabled)
+	if err != nil {
+		c.Fatalf("Unable to update user's secret key: %v", err)
+	}
+	// 3.10.1 Check that old password no longer works.
+	err = client.MakeBucket(ctx, getRandomBucketName(), minio.MakeBucketOptions{})
+	if err == nil {
+		c.Fatalf("user was unexpectedly able to create bucket with bad password!")
+	}
+	// 3.10.2 Check that new password works.
+	client = s.getUserClient(c, accessKey, newSecretKey, "")
+	err = client.MakeBucket(ctx, getRandomBucketName(), minio.MakeBucketOptions{})
+	if err != nil {
+		c.Fatalf("user could not create bucket: %v", err)
+	}
+
 	// 4. Check that user can be disabled and verify it.
 	err = s.adm.SetUserStatus(ctx, accessKey, madmin.AccountDisabled)
 	if err != nil {
@@ -260,6 +288,108 @@ func (s *TestSuiteIAM) TestUserCreate(c *check) {
 	}
 }
 
+func (s *TestSuiteIAM) TestUserPolicyEscalationBug(c *check) {
+	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
+	defer cancel()
+
+	bucket := getRandomBucketName()
+	err := s.client.MakeBucket(ctx, bucket, minio.MakeBucketOptions{})
+	if err != nil {
+		c.Fatalf("bucket creat error: %v", err)
+	}
+
+	// 2. Create a user, associate policy and verify access
+	accessKey, secretKey := mustGenerateCredentials(c)
+	err = s.adm.SetUser(ctx, accessKey, secretKey, madmin.AccountEnabled)
+	if err != nil {
+		c.Fatalf("Unable to set user: %v", err)
+	}
+	// 2.1 check that user does not have any access to the bucket
+	uClient := s.getUserClient(c, accessKey, secretKey, "")
+	c.mustNotListObjects(ctx, uClient, bucket)
+
+	// 2.2 create and associate policy to user
+	policy := "mypolicy-test-user-update"
+	policyBytes := []byte(fmt.Sprintf(`{
+ "Version": "2012-10-17",
+ "Statement": [
+  {
+   "Effect": "Allow",
+   "Action": [
+    "s3:PutObject",
+    "s3:GetObject",
+    "s3:ListBucket"
+   ],
+   "Resource": [
+    "arn:aws:s3:::%s/*"
+   ]
+  }
+ ]
+}`, bucket))
+	err = s.adm.AddCannedPolicy(ctx, policy, policyBytes)
+	if err != nil {
+		c.Fatalf("policy add error: %v", err)
+	}
+	err = s.adm.SetPolicy(ctx, policy, accessKey, false)
+	if err != nil {
+		c.Fatalf("Unable to set policy: %v", err)
+	}
+	// 2.3 check user has access to bucket
+	c.mustListObjects(ctx, uClient, bucket)
+	// 2.3 check that user cannot delete the bucket
+	err = uClient.RemoveBucket(ctx, bucket)
+	if err == nil || err.Error() != "Access Denied." {
+		c.Fatalf("bucket was deleted unexpectedly or got unexpected err: %v", err)
+	}
+
+	// 3. Craft a request to update the user's permissions
+	ep := s.adm.GetEndpointURL()
+	urlValue := url.Values{}
+	urlValue.Add("accessKey", accessKey)
+	u, err := url.Parse(fmt.Sprintf("%s://%s/minio/admin/v3/add-user?%s", ep.Scheme, ep.Host, s3utils.QueryEncode(urlValue)))
+	if err != nil {
+		c.Fatalf("unexpected url parse err: %v", err)
+	}
+	req, err := http.NewRequestWithContext(ctx, http.MethodPut, u.String(), nil)
+	if err != nil {
+		c.Fatalf("unexpected new request err: %v", err)
+	}
+	reqBodyArg := madmin.UserInfo{
+		SecretKey:  secretKey,
+		PolicyName: "consoleAdmin",
+		Status:     madmin.AccountEnabled,
+	}
+	buf, err := json.Marshal(reqBodyArg)
+	if err != nil {
+		c.Fatalf("unexpected json encode err: %v", err)
+	}
+	buf, err = madmin.EncryptData(secretKey, buf)
+	if err != nil {
+		c.Fatalf("unexpected encryption err: %v", err)
+	}
+
+	req.ContentLength = int64(len(buf))
+	sum := sha256.Sum256(buf)
+	req.Header.Set("X-Amz-Content-Sha256", hex.EncodeToString(sum[:]))
+	req.Body = ioutil.NopCloser(bytes.NewReader(buf))
+	req = signer.SignV4(*req, accessKey, secretKey, "", "")
+
+	// 3.1 Execute the request.
+	resp, err := s.TestSuiteCommon.client.Do(req)
+	if err != nil {
+		c.Fatalf("unexpected request err: %v", err)
+	}
+	if resp.StatusCode != 200 {
+		c.Fatalf("got unexpected response: %#v\n", resp)
+	}
+
+	// 3.2 check that user cannot delete the bucket
+	err = uClient.RemoveBucket(ctx, bucket)
+	if err == nil || err.Error() != "Access Denied." {
+		c.Fatalf("User was able to escalate privileges (Err=%v)!", err)
+	}
+}
+
 func (s *TestSuiteIAM) TestAddServiceAccountPerms(c *check) {
 	ctx, cancel := context.WithTimeout(context.Background(), testDefaultTimeout)
 	defer cancel()
@@ -523,7 +653,6 @@ func (s *TestSuiteIAM) TestCannedPolicies(c *check) {
 	if !strings.Contains(infoStr, `"s3:PutObject"`) || !strings.Contains(infoStr, ":"+bucket+"/") {
 		c.Fatalf("policy contains unexpected content!")
 	}
-
 }
 
 func (s *TestSuiteIAM) TestGroupAddRemove(c *check) {

cmd/admin-handlers.go

@@ -333,7 +333,6 @@ func (a adminAPIHandlers) StorageInfoHandler(w http.ResponseWriter, r *http.Requ
 	// Reply with storage information (across nodes in a
 	// distributed setup) as json.
 	writeSuccessResponseJSON(w, jsonBytes)
-
 }
 
 // DataUsageInfoHandler - GET /minio/admin/v3/datausage
@@ -979,8 +978,7 @@ func (a adminAPIHandlers) SpeedtestHandler(w http.ResponseWriter, r *http.Reques
 	}
 
 	deleteBucket := func() {
-		loc := pathJoin(minioMetaSpeedTestBucket, minioMetaSpeedTestBucketPrefix)
-		objectAPI.DeleteBucket(context.Background(), loc, DeleteBucketOptions{
+		objectAPI.DeleteBucket(context.Background(), pathJoin(minioMetaBucket, "speedtest"), DeleteBucketOptions{
 			Force:      true,
 			NoRecreate: true,
 		})
@@ -1332,7 +1330,7 @@ func (a adminAPIHandlers) KMSKeyStatusHandler(w http.ResponseWriter, r *http.Req
 	if keyID == "" {
 		keyID = stat.DefaultKey
 	}
-	var response = madmin.KMSKeyStatus{
+	response := madmin.KMSKeyStatus{
 		KeyID: keyID,
 	}
 
@@ -1816,7 +1814,6 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 			anonNetwork[anonEndpoint] = status
 		}
 		return anonNetwork
-
 	}
 
 	anonymizeDrives := func(drives []madmin.Disk) []madmin.Disk {
@@ -1916,7 +1913,6 @@ func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Reque
 			return
 		}
 	}
-
 }
 
 func getTLSInfo() madmin.TLSInfo {
@@ -2042,7 +2038,6 @@ func assignPoolNumbers(servers []madmin.ServerProperties) {
 }
 
 func fetchLambdaInfo() []map[string][]madmin.TargetIDStatus {
-
 	lambdaMap := make(map[string][]madmin.TargetIDStatus)
 
 	for _, tgt := range globalConfigTargetList.Targets() {
@@ -2284,7 +2279,7 @@ func (a adminAPIHandlers) InspectDataHandler(w http.ResponseWriter, r *http.Requ
 		}
 		if si.Mode == 0 {
 			// Not, set it to default.
-			si.Mode = 0600
+			si.Mode = 0o600
 		}
 		header, zerr := zip.FileInfoHeader(dummyFileInfo{
 			name:    filename,

cmd/admin-handlers_test.go

@@ -236,8 +236,8 @@ func TestServiceRestartHandler(t *testing.T) {
 
 // buildAdminRequest - helper function to build an admin API request.
 func buildAdminRequest(queryVal url.Values, method, path string,
-	contentLength int64, bodySeeker io.ReadSeeker) (*http.Request, error) {
-
+	contentLength int64, bodySeeker io.ReadSeeker) (*http.Request, error,
+) {
 	req, err := newTestRequest(method,
 		adminPathPrefix+adminAPIVersionPrefix+path+"?"+queryVal.Encode(),
 		contentLength, bodySeeker)
@@ -380,5 +380,4 @@ func TestExtractHealInitParams(t *testing.T) {
 			}
 		}
 	}
-
 }

cmd/admin-heal-ops.go

@@ -278,8 +278,8 @@ func (ahs *allHealState) stopHealSequence(path string) ([]byte, APIError) {
 // background routine to clean up heal results after the
 // aforementioned duration.
 func (ahs *allHealState) LaunchNewHealSequence(h *healSequence, objAPI ObjectLayer) (
-	respBytes []byte, apiErr APIError, errMsg string) {
-
+	respBytes []byte, apiErr APIError, errMsg string,
+) {
 	if h.forceStarted {
 		_, apiErr = ahs.stopHealSequence(pathJoin(h.bucket, h.object))
 		if apiErr.Code != "" {
@@ -338,8 +338,8 @@ func (ahs *allHealState) LaunchNewHealSequence(h *healSequence, objAPI ObjectLay
 // representation. The clientToken helps ensure there aren't
 // conflicting clients fetching status.
 func (ahs *allHealState) PopHealStatusJSON(hpath string,
-	clientToken string) ([]byte, APIErrorCode) {
-
+	clientToken string) ([]byte, APIErrorCode,
+) {
 	// fetch heal state for given path
 	h, exists := ahs.getHealSequence(hpath)
 	if !exists {
@@ -453,8 +453,8 @@ type healSequence struct {
 // NewHealSequence - creates healSettings, assumes bucket and
 // objPrefix are already validated.
 func newHealSequence(ctx context.Context, bucket, objPrefix, clientAddr string,
-	hs madmin.HealOpts, forceStart bool) *healSequence {
-
+	hs madmin.HealOpts, forceStart bool,
+) *healSequence {
 	reqInfo := &logger.ReqInfo{RemoteHost: clientAddr, API: "Heal", BucketName: bucket}
 	reqInfo.AppendTags("prefix", objPrefix)
 	ctx, cancel := context.WithCancel(logger.SetReqInfo(ctx, reqInfo))
@@ -710,6 +710,9 @@ func (h *healSequence) queueHealTask(source healSource, healType madmin.HealItem
 
 	select {
 	case globalBackgroundHealRoutine.tasks <- task:
+		if serverDebugLog {
+			logger.Info("Task in the queue: %#v", task)
+		}
 	case <-h.ctx.Done():
 		return nil
 	}
@@ -885,6 +888,7 @@ func (h *healSequence) healObject(bucket, object, versionID string) error {
 		bucket:    bucket,
 		object:    object,
 		versionID: versionID,
+		opts:      &h.settings,
 	}, madmin.HealItemObject)
 
 	// Wait and proceed if there are active requests

cmd/admin-router.go

@@ -38,7 +38,6 @@ type adminAPIHandlers struct{}
 
 // registerAdminRouter - Add handler functions for each service REST API routes.
 func registerAdminRouter(router *mux.Router, enableConfigOps bool) {
-
 	adminAPI := adminAPIHandlers{}
 	// Admin router
 	adminRouter := router.PathPrefix(adminPathPrefix).Subrouter()
@@ -75,8 +74,14 @@ func registerAdminRouter(router *mux.Router, enableConfigOps bool) {
 			adminRouter.Methods(http.MethodPost).Path(adminVersion + "/heal/").HandlerFunc(gz(httpTraceAll(adminAPI.HealHandler)))
 			adminRouter.Methods(http.MethodPost).Path(adminVersion + "/heal/{bucket}").HandlerFunc(gz(httpTraceAll(adminAPI.HealHandler)))
 			adminRouter.Methods(http.MethodPost).Path(adminVersion + "/heal/{bucket}/{prefix:.*}").HandlerFunc(gz(httpTraceAll(adminAPI.HealHandler)))
-
 			adminRouter.Methods(http.MethodPost).Path(adminVersion + "/background-heal/status").HandlerFunc(gz(httpTraceAll(adminAPI.BackgroundHealStatusHandler)))
+
+			// Pool operations
+			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/pools/list").HandlerFunc(gz(httpTraceAll(adminAPI.ListPools)))
+			adminRouter.Methods(http.MethodGet).Path(adminVersion+"/pools/status").HandlerFunc(gz(httpTraceAll(adminAPI.StatusPool))).Queries("pool", "{pool:.*}")
+
+			adminRouter.Methods(http.MethodPost).Path(adminVersion+"/pools/decommission").HandlerFunc(gz(httpTraceAll(adminAPI.StartDecommission))).Queries("pool", "{pool:.*}")
+			adminRouter.Methods(http.MethodPost).Path(adminVersion+"/pools/cancel").HandlerFunc(gz(httpTraceAll(adminAPI.CancelDecommission))).Queries("pool", "{pool:.*}")
 		}
 
 		// Profiling operations
@@ -194,11 +199,16 @@ func registerAdminRouter(router *mux.Router, enableConfigOps bool) {
 			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/add").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationAdd)))
 			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/disable").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationDisable)))
 			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/info").HandlerFunc(gz(httpTraceHdrs(adminAPI.SiteReplicationInfo)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/join").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRInternalJoin)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion+"/site-replication/peer/bucket-ops").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRInternalBucketOps))).Queries("bucket", "{bucket:.*}").Queries("operation", "{operation:.*}")
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/iam-item").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRInternalReplicateIAMItem)))
-			adminRouter.Methods(http.MethodPut).Path(adminVersion + "/site-replication/peer/bucket-meta").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRInternalReplicateBucketItem)))
-			adminRouter.Methods(http.MethodGet).Path(adminVersion + "/site-replication/peer/idp-settings").HandlerFunc(gz(httpTraceHdrs(adminAPI.SRInternalGetIDPSettings)))
+			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)))
 		}
 
 		if globalIsDistErasure {

cmd/api-datatypes.go

@@ -48,10 +48,15 @@ func (t DeleteMarkerMTime) MarshalXML(e *xml.Encoder, startElement xml.StartElem
 	return e.EncodeElement(t.Time.Format(time.RFC3339), startElement)
 }
 
-// ObjectToDelete carries key name for the object to delete.
-type ObjectToDelete struct {
+// ObjectV object version key/versionId
+type ObjectV struct {
 	ObjectName string `xml:"Key"`
 	VersionID  string `xml:"VersionId"`
+}
+
+// ObjectToDelete carries key name for the object to delete.
+type ObjectToDelete struct {
+	ObjectV
 	// Replication status of DeleteMarker
 	DeleteMarkerReplicationStatus string `xml:"DeleteMarkerReplicationStatus"`
 	// Status of versioned delete (of object or DeleteMarker)

cmd/api-errors.go

@@ -82,6 +82,7 @@ const (
 	ErrIncompleteBody
 	ErrInternalError
 	ErrInvalidAccessKeyID
+	ErrAccessKeyDisabled
 	ErrInvalidBucketName
 	ErrInvalidDigest
 	ErrInvalidRange
@@ -514,6 +515,11 @@ var errorCodes = errorCodeMap{
 		Description:    "The Access Key Id you provided does not exist in our records.",
 		HTTPStatusCode: http.StatusForbidden,
 	},
+	ErrAccessKeyDisabled: {
+		Code:           "InvalidAccessKeyId",
+		Description:    "Your account is disabled; please contact your administrator.",
+		HTTPStatusCode: http.StatusForbidden,
+	},
 	ErrInvalidBucketName: {
 		Code:           "InvalidBucketName",
 		Description:    "The specified bucket is not valid.",
@@ -681,7 +687,7 @@ var errorCodes = errorCodeMap{
 	},
 	ErrAllAccessDisabled: {
 		Code:           "AllAccessDisabled",
-		Description:    "All access to this bucket has been disabled.",
+		Description:    "All access to this resource has been disabled.",
 		HTTPStatusCode: http.StatusForbidden,
 	},
 	ErrMalformedPolicy: {
@@ -2119,7 +2125,7 @@ func toAPIError(ctx context.Context, err error) APIError {
 		return noError
 	}
 
-	var apiErr = errorCodes.ToAPIErr(toAPIErrorCode(ctx, err))
+	apiErr := errorCodes.ToAPIErr(toAPIErrorCode(ctx, err))
 	e, ok := err.(dns.ErrInvalidBucketName)
 	if ok {
 		code := toAPIErrorCode(ctx, e)
@@ -2232,7 +2238,6 @@ func toAPIError(ctx context.Context, err error) APIError {
 			// since S3 only sends one Error XML response.
 			if len(e.Errors) >= 1 {
 				apiErr.Code = e.Errors[0].Reason
-
 			}
 		case azblob.StorageError:
 			apiErr = APIError{

cmd/api-headers_test.go

@@ -23,7 +23,7 @@ import (
 
 func TestNewRequestID(t *testing.T) {
 	// Ensure that it returns an alphanumeric result of length 16.
-	var id = mustGetRequestID(UTCNow())
+	id := mustGetRequestID(UTCNow())
 
 	if len(id) != 16 {
 		t.Fail()

cmd/api-response.go

@@ -268,7 +268,6 @@ type StringMap map[string]string
 
 // MarshalXML - StringMap marshals into XML.
 func (s StringMap) MarshalXML(e *xml.Encoder, start xml.StartElement) error {
-
 	tokens := []xml.Token{start}
 
 	for key, value := range s {
@@ -417,8 +416,8 @@ func getObjectLocation(r *http.Request, domains []string, bucket, object string)
 // serialized to match XML and JSON API spec output.
 func generateListBucketsResponse(buckets []BucketInfo) ListBucketsResponse {
 	listbuckets := make([]Bucket, 0, len(buckets))
-	var data = ListBucketsResponse{}
-	var owner = Owner{
+	data := ListBucketsResponse{}
+	owner := Owner{
 		ID:          globalMinioDefaultOwnerID,
 		DisplayName: "minio",
 	}
@@ -439,14 +438,14 @@ func generateListBucketsResponse(buckets []BucketInfo) ListBucketsResponse {
 // generates an ListBucketVersions response for the said bucket with other enumerated options.
 func generateListVersionsResponse(bucket, prefix, marker, versionIDMarker, delimiter, encodingType string, maxKeys int, resp ListObjectVersionsInfo) ListVersionsResponse {
 	versions := make([]ObjectVersion, 0, len(resp.Objects))
-	var owner = Owner{
+	owner := Owner{
 		ID:          globalMinioDefaultOwnerID,
 		DisplayName: "minio",
 	}
-	var data = ListVersionsResponse{}
+	data := ListVersionsResponse{}
 
 	for _, object := range resp.Objects {
-		var content = ObjectVersion{}
+		content := ObjectVersion{}
 		if object.Name == "" {
 			continue
 		}
@@ -486,7 +485,7 @@ func generateListVersionsResponse(bucket, prefix, marker, versionIDMarker, delim
 
 	prefixes := make([]CommonPrefix, 0, len(resp.Prefixes))
 	for _, prefix := range resp.Prefixes {
-		var prefixItem = CommonPrefix{}
+		prefixItem := CommonPrefix{}
 		prefixItem.Prefix = s3EncodeName(prefix, encodingType)
 		prefixes = append(prefixes, prefixItem)
 	}
@@ -497,14 +496,14 @@ func generateListVersionsResponse(bucket, prefix, marker, versionIDMarker, delim
 // generates an ListObjectsV1 response for the said bucket with other enumerated options.
 func generateListObjectsV1Response(bucket, prefix, marker, delimiter, encodingType string, maxKeys int, resp ListObjectsInfo) ListObjectsResponse {
 	contents := make([]Object, 0, len(resp.Objects))
-	var owner = Owner{
+	owner := Owner{
 		ID:          globalMinioDefaultOwnerID,
 		DisplayName: "minio",
 	}
-	var data = ListObjectsResponse{}
+	data := ListObjectsResponse{}
 
 	for _, object := range resp.Objects {
-		var content = Object{}
+		content := Object{}
 		if object.Name == "" {
 			continue
 		}
@@ -535,7 +534,7 @@ func generateListObjectsV1Response(bucket, prefix, marker, delimiter, encodingTy
 
 	prefixes := make([]CommonPrefix, 0, len(resp.Prefixes))
 	for _, prefix := range resp.Prefixes {
-		var prefixItem = CommonPrefix{}
+		prefixItem := CommonPrefix{}
 		prefixItem.Prefix = s3EncodeName(prefix, encodingType)
 		prefixes = append(prefixes, prefixItem)
 	}
@@ -546,14 +545,14 @@ func generateListObjectsV1Response(bucket, prefix, marker, delimiter, encodingTy
 // generates an ListObjectsV2 response for the said bucket with other enumerated options.
 func generateListObjectsV2Response(bucket, prefix, token, nextToken, startAfter, delimiter, encodingType string, fetchOwner, isTruncated bool, maxKeys int, objects []ObjectInfo, prefixes []string, metadata bool) ListObjectsV2Response {
 	contents := make([]Object, 0, len(objects))
-	var owner = Owner{
+	owner := Owner{
 		ID:          globalMinioDefaultOwnerID,
 		DisplayName: "minio",
 	}
-	var data = ListObjectsV2Response{}
+	data := ListObjectsV2Response{}
 
 	for _, object := range objects {
-		var content = Object{}
+		content := Object{}
 		if object.Name == "" {
 			continue
 		}
@@ -608,7 +607,7 @@ func generateListObjectsV2Response(bucket, prefix, token, nextToken, startAfter,
 
 	commonPrefixes := make([]CommonPrefix, 0, len(prefixes))
 	for _, prefix := range prefixes {
-		var prefixItem = CommonPrefix{}
+		prefixItem := CommonPrefix{}
 		prefixItem.Prefix = s3EncodeName(prefix, encodingType)
 		commonPrefixes = append(commonPrefixes, prefixItem)
 	}
@@ -821,8 +820,8 @@ func writeErrorResponseJSON(ctx context.Context, w http.ResponseWriter, err APIE
 // but accepts the error message directly (this allows messages to be
 // dynamically generated.)
 func writeCustomErrorResponseJSON(ctx context.Context, w http.ResponseWriter, err APIError,
-	errBody string, reqURL *url.URL) {
-
+	errBody string, reqURL *url.URL,
+) {
 	reqInfo := logger.GetReqInfo(ctx)
 	errorResponse := APIErrorResponse{
 		Code:       err.Code,

cmd/api-router.go

@@ -484,7 +484,6 @@ func registerAPIRouter(router *mux.Router) {
 	// If none of the routes match add default error handler routes
 	apiRouter.NotFoundHandler = collectAPIStats("notfound", httpTraceAll(errorResponseHandler))
 	apiRouter.MethodNotAllowedHandler = collectAPIStats("methodnotallowed", httpTraceAll(methodNotAllowedHandler("S3")))
-
 }
 
 // corsHandler handler for CORS (Cross Origin Resource Sharing)

cmd/api-utils_test.go

@@ -44,7 +44,6 @@ func TestS3EncodeName(t *testing.T) {
 			if testCase.expectedOutput != outputText {
 				t.Errorf("Expected `%s`, got `%s`", testCase.expectedOutput, outputText)
 			}
-
 		})
 	}
 }

cmd/background-heal-ops.go

@@ -115,7 +115,6 @@ func newHealRoutine() *healRoutine {
 		tasks:   make(chan healTask),
 		workers: workers,
 	}
-
 }
 
 // healDiskFormat - heals format.json, return value indicates if a

cmd/background-newdisks-heal-ops.go

@@ -289,23 +289,22 @@ func initAutoHeal(ctx context.Context, objAPI ObjectLayer) {
 }
 
 func getLocalDisksToHeal() (disksToHeal Endpoints) {
-	for _, ep := range globalEndpoints {
-		for _, endpoint := range ep.Endpoints {
-			if !endpoint.IsLocal {
-				continue
-			}
-			// Try to connect to the current endpoint
-			// and reformat if the current disk is not formatted
-			disk, _, err := connectEndpoint(endpoint)
-			if errors.Is(err, errUnformattedDisk) {
-				disksToHeal = append(disksToHeal, endpoint)
-			} else if err == nil && disk != nil && disk.Healing() != nil {
-				disksToHeal = append(disksToHeal, disk.Endpoint())
-			}
+	for _, disk := range globalLocalDrives {
+		_, err := disk.GetDiskID()
+		if errors.Is(err, errUnformattedDisk) {
+			disksToHeal = append(disksToHeal, disk.Endpoint())
+			continue
+		}
+		if disk.Healing() != nil {
+			disksToHeal = append(disksToHeal, disk.Endpoint())
 		}
 	}
+	if len(disksToHeal) == globalEndpoints.NEndpoints() {
+		// When all disks == all command line endpoints
+		// this is a fresh setup, no need to trigger healing.
+		return Endpoints{}
+	}
 	return disksToHeal
-
 }
 
 // monitorLocalDisksAndHeal - ensures that detected new disks are healed

cmd/bitrot.go

@@ -119,8 +119,10 @@ func newBitrotReader(disk StorageAPI, data []byte, bucket string, filePath strin
 // Close all the readers.
 func closeBitrotReaders(rs []io.ReaderAt) {
 	for _, r := range rs {
-		if br, ok := r.(io.Closer); ok {
-			br.Close()
+		if r != nil {
+			if br, ok := r.(io.Closer); ok {
+				br.Close()
+			}
 		}
 	}
 }
@@ -128,8 +130,10 @@ func closeBitrotReaders(rs []io.ReaderAt) {
 // Close all the writers.
 func closeBitrotWriters(ws []io.Writer) {
 	for _, w := range ws {
-		if bw, ok := w.(io.Closer); ok {
-			bw.Close()
+		if w != nil {
+			if bw, ok := w.(io.Closer); ok {
+				bw.Close()
+			}
 		}
 	}
 }
@@ -212,7 +216,7 @@ func bitrotVerify(r io.Reader, wantSize, partSize int64, algo BitrotAlgorithm, w
 // bitrotSelfTest tries to catch any issue in the bitrot implementation
 // early instead of silently corrupting data.
 func bitrotSelfTest() {
-	var checksums = map[BitrotAlgorithm]string{
+	checksums := map[BitrotAlgorithm]string{
 		SHA256:          "a7677ff19e0182e4d52e3a3db727804abc82a5818749336369552e54b838b004",
 		BLAKE2b512:      "e519b7d84b1c3c917985f544773a35cf265dcab10948be3550320d156bab612124a5ae2ae5a8c73c0eea360f68b0e28136f26e858756dbfe7375a7389f26c669",
 		HighwayHash256:  "39c0407ed3f01b18d22c85db4aeff11e060ca5f43131b0126731ca197cd42313",

cmd/bootstrap-peer-server.go

@@ -86,14 +86,29 @@ func (s1 ServerSystemConfig) Diff(s2 ServerSystemConfig) error {
 		}
 	}
 	if !reflect.DeepEqual(s1.MinioEnv, s2.MinioEnv) {
-		return fmt.Errorf("Expected same MINIO_ environment variables and values")
+		var missing []string
+		var mismatching []string
+		for k, v := range s1.MinioEnv {
+			ev, ok := s2.MinioEnv[k]
+			if !ok {
+				missing = append(missing, k)
+			} else if v != ev {
+				mismatching = append(mismatching, k)
+			}
+		}
+		if len(mismatching) > 0 {
+			return fmt.Errorf(`Expected same MINIO_ environment variables and values across all servers: Missing environment values: %s / Mismatch environment values: %s`, missing, mismatching)
+		}
+		return fmt.Errorf(`Expected same MINIO_ environment variables and values across all servers: Missing environment values: %s`, missing)
 	}
 	return nil
 }
 
 var skipEnvs = map[string]struct{}{
-	"MINIO_OPTS":        {},
-	"MINIO_CERT_PASSWD": {},
+	"MINIO_OPTS":         {},
+	"MINIO_CERT_PASSWD":  {},
+	"MINIO_SERVER_DEBUG": {},
+	"MINIO_DSYNC_TRACE":  {},
 }
 
 func getServerSystemCfg() ServerSystemConfig {

cmd/bucket-handlers.go

@@ -422,11 +422,16 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 		return
 	}
 
+	objects := make([]ObjectV, len(deleteObjectsReq.Objects))
 	// Convert object name delete objects if it has `/` in the beginning.
 	for i := range deleteObjectsReq.Objects {
 		deleteObjectsReq.Objects[i].ObjectName = trimLeadingSlash(deleteObjectsReq.Objects[i].ObjectName)
+		objects[i] = deleteObjectsReq.Objects[i].ObjectV
 	}
 
+	// Make sure to update context to print ObjectNames for multi objects.
+	ctx = updateReqContext(ctx, objects...)
+
 	// Call checkRequestAuthType to populate ReqInfo.AccessKey before GetBucketInfo()
 	// Ignore errors here to preserve the S3 error behavior of GetBucketInfo()
 	checkRequestAuthType(ctx, r, policy.DeleteObjectAction, bucket, "")
@@ -449,7 +454,7 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 		return
 	}
 
-	var objectsToDelete = map[ObjectToDelete]int{}
+	objectsToDelete := map[ObjectToDelete]int{}
 	getObjectInfoFn := objectAPI.GetObjectInfo
 	if api.CacheAPI() != nil {
 		getObjectInfoFn = api.CacheAPI().GetObjectInfo
@@ -527,8 +532,10 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 
 		if replicateDeletes {
 			dsc = checkReplicateDelete(ctx, bucket, ObjectToDelete{
-				ObjectName: object.ObjectName,
-				VersionID:  object.VersionID,
+				ObjectV: ObjectV{
+					ObjectName: object.ObjectName,
+					VersionID:  object.VersionID,
+				},
 			}, goi, opts, gerr)
 			if dsc.ReplicateAny() {
 				if object.VersionID != "" {
@@ -569,6 +576,9 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 		return
 	}
 
+	// Disable timeouts and cancellation
+	ctx = bgContext(ctx)
+
 	deleteList := toNames(objectsToDelete)
 	dObjects, errs := deleteObjectsFn(ctx, bucket, deleteList, ObjectOptions{
 		Versioned:        versioned,
@@ -581,8 +591,10 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 		// created during DeleteMarker creation when client didn't
 		// specify a versionID.
 		objToDel := ObjectToDelete{
-			ObjectName:                    dObjects[i].ObjectName,
-			VersionID:                     dObjects[i].VersionID,
+			ObjectV: ObjectV{
+				ObjectName: dObjects[i].ObjectName,
+				VersionID:  dObjects[i].VersionID,
+			},
 			VersionPurgeStatus:            dObjects[i].VersionPurgeStatus(),
 			VersionPurgeStatuses:          dObjects[i].ReplicationState.VersionPurgeStatusInternal,
 			DeleteMarkerReplicationStatus: dObjects[i].ReplicationState.ReplicationStatusInternal,
@@ -606,8 +618,8 @@ func (api objectAPIHandlers) DeleteMultipleObjectsHandler(w http.ResponseWriter,
 	}
 
 	// Generate response
-	var deleteErrors = make([]DeleteError, 0, len(deleteObjectsReq.Objects))
-	var deletedObjects = make([]DeletedObject, 0, len(deleteObjectsReq.Objects))
+	deleteErrors := make([]DeleteError, 0, len(deleteObjectsReq.Objects))
+	deletedObjects := make([]DeletedObject, 0, len(deleteObjectsReq.Objects))
 	for _, deleteResult := range deleteResults {
 		if deleteResult.errInfo.Code != "" {
 			deleteErrors = append(deleteErrors, deleteResult.errInfo)
@@ -939,6 +951,7 @@ func (api objectAPIHandlers) PostPolicyBucketHandler(w http.ResponseWriter, r *h
 	// explicit permissions for the user.
 	if !globalIAMSys.IsAllowed(iampolicy.Args{
 		AccountName:     cred.AccessKey,
+		Groups:          cred.Groups,
 		Action:          iampolicy.PutObjectAction,
 		ConditionValues: getConditionValues(r, "", cred.AccessKey, cred.Claims),
 		BucketName:      bucket,
@@ -1806,7 +1819,8 @@ func (api objectAPIHandlers) ResetBucketReplicationStateHandler(w http.ResponseW
 	tgtArns := config.FilterTargetArns(
 		replication.ObjectOpts{
 			OpType:    replication.ResyncReplicationType,
-			TargetArn: arn})
+			TargetArn: arn,
+		})
 
 	if len(tgtArns) == 0 {
 		writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrBadRequest, InvalidArgument{

cmd/bucket-handlers_test.go

@@ -81,8 +81,8 @@ func TestGetBucketLocationHandler(t *testing.T) {
 }
 
 func testGetBucketLocationHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	// test cases with sample input and expected output.
 	testCases := []struct {
 		bucketName string
@@ -163,7 +163,6 @@ func testGetBucketLocationHandler(obj ObjectLayer, instanceType, bucketName stri
 		recV2 := httptest.NewRecorder()
 		// construct HTTP request for PUT bucket policy endpoint.
 		reqV2, err := newTestSignedRequestV2(http.MethodGet, getBucketLocationURL("", testCase.bucketName), 0, nil, testCase.accessKey, testCase.secretKey, nil)
-
 		if err != nil {
 			t.Fatalf("Test %d: %s: Failed to create HTTP request for PutBucketPolicyHandler: <ERROR> %v", i+1, instanceType, err)
 		}
@@ -210,7 +209,6 @@ func testGetBucketLocationHandler(obj ObjectLayer, instanceType, bucketName stri
 
 	nilBucket := "dummy-bucket"
 	nilReq, err := newTestRequest(http.MethodGet, getBucketLocationURL("", nilBucket), 0, nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}
@@ -225,8 +223,8 @@ func TestHeadBucketHandler(t *testing.T) {
 }
 
 func testHeadBucketHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	// test cases with sample input and expected output.
 	testCases := []struct {
 		bucketName string
@@ -282,7 +280,6 @@ func testHeadBucketHandler(obj ObjectLayer, instanceType, bucketName string, api
 		recV2 := httptest.NewRecorder()
 		// construct HTTP request for PUT bucket policy endpoint.
 		reqV2, err := newTestSignedRequestV2(http.MethodHead, getHEADBucketURL("", testCase.bucketName), 0, nil, testCase.accessKey, testCase.secretKey, nil)
-
 		if err != nil {
 			t.Fatalf("Test %d: %s: Failed to create HTTP request for PutBucketPolicyHandler: <ERROR> %v", i+1, instanceType, err)
 		}
@@ -297,7 +294,6 @@ func testHeadBucketHandler(obj ObjectLayer, instanceType, bucketName string, api
 
 	// Test for Anonymous/unsigned http request.
 	anonReq, err := newTestRequest(http.MethodHead, getHEADBucketURL("", bucketName), 0, nil)
-
 	if err != nil {
 		t.Fatalf("MinIO %s: Failed to create an anonymous request for bucket \"%s\": <ERROR> %v",
 			instanceType, bucketName, err)
@@ -315,7 +311,6 @@ func testHeadBucketHandler(obj ObjectLayer, instanceType, bucketName string, api
 
 	nilBucket := "dummy-bucket"
 	nilReq, err := newTestRequest(http.MethodHead, getHEADBucketURL("", nilBucket), 0, nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}
@@ -331,8 +326,8 @@ func TestListMultipartUploadsHandler(t *testing.T) {
 
 // testListMultipartUploadsHandler - Tests validate listing of multipart uploads.
 func testListMultipartUploadsHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	// Collection of non-exhaustive ListMultipartUploads test cases, valid errors
 	// and success responses.
 	testCases := []struct {
@@ -552,7 +547,6 @@ func testListMultipartUploadsHandler(obj ObjectLayer, instanceType, bucketName s
 		testCases[6].uploadIDMarker, testCases[6].delimiter, testCases[6].maxUploads)
 
 	nilReq, err := newTestRequest(http.MethodGet, url, 0, nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}
@@ -568,8 +562,8 @@ func TestListBucketsHandler(t *testing.T) {
 
 // testListBucketsHandler - Tests validate listing of buckets.
 func testListBucketsHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	testCases := []struct {
 		bucketName         string
 		accessKey          string
@@ -615,7 +609,6 @@ func testListBucketsHandler(obj ObjectLayer, instanceType, bucketName string, ap
 
 		// verify response for V2 signed HTTP request.
 		reqV2, err := newTestSignedRequestV2(http.MethodGet, getListBucketURL(""), 0, nil, testCase.accessKey, testCase.secretKey, nil)
-
 		if err != nil {
 			t.Fatalf("Test %d: %s: Failed to create HTTP request for PutBucketPolicyHandler: <ERROR> %v", i+1, instanceType, err)
 		}
@@ -630,7 +623,6 @@ func testListBucketsHandler(obj ObjectLayer, instanceType, bucketName string, ap
 	// Test for Anonymous/unsigned http request.
 	// ListBucketsHandler doesn't support bucket policies, setting the policies shouldn't make a difference.
 	anonReq, err := newTestRequest(http.MethodGet, getListBucketURL(""), 0, nil)
-
 	if err != nil {
 		t.Fatalf("MinIO %s: Failed to create an anonymous request.", instanceType)
 	}
@@ -646,7 +638,6 @@ func testListBucketsHandler(obj ObjectLayer, instanceType, bucketName string, ap
 	// The only aim is to generate an HTTP request in a way that the relevant/registered end point is evoked/called.
 
 	nilReq, err := newTestRequest(http.MethodGet, getListBucketURL(""), 0, nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}
@@ -661,8 +652,8 @@ func TestAPIDeleteMultipleObjectsHandler(t *testing.T) {
 }
 
 func testAPIDeleteMultipleObjectsHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	var err error
 
 	contentBytes := []byte("hello")
@@ -707,7 +698,9 @@ func testAPIDeleteMultipleObjectsHandler(obj ObjectLayer, instanceType, bucketNa
 	getObjectToDeleteList := func(objectNames []string) (objectList []ObjectToDelete) {
 		for _, objectName := range objectNames {
 			objectList = append(objectList, ObjectToDelete{
-				ObjectName: objectName,
+				ObjectV: ObjectV{
+					ObjectName: objectName,
+				},
 			})
 		}
 
@@ -726,10 +719,21 @@ func testAPIDeleteMultipleObjectsHandler(obj ObjectLayer, instanceType, bucketNa
 		return deleteErrorList
 	}
 
+	objects := []ObjectToDelete{}
+	objects = append(objects, ObjectToDelete{
+		ObjectV: ObjectV{
+			ObjectName: "private/object",
+		},
+	})
+	objects = append(objects, ObjectToDelete{
+		ObjectV: ObjectV{
+			ObjectName: "public/object",
+		},
+	})
 	requestList := []DeleteObjectsRequest{
 		{Quiet: false, Objects: getObjectToDeleteList(objectNames[:5])},
 		{Quiet: true, Objects: getObjectToDeleteList(objectNames[5:])},
-		{Quiet: false, Objects: []ObjectToDelete{{ObjectName: "private/object"}, {ObjectName: "public/object"}}},
+		{Quiet: false, Objects: objects},
 	}
 
 	// generate multi objects delete response.

cmd/bucket-lifecycle-handlers_test.go

@@ -150,8 +150,8 @@ func TestBucketLifecycle(t *testing.T) {
 // Simple tests of bucket lifecycle: PUT, GET, DELETE.
 // Tests are related and the order is important.
 func testBucketLifecycleHandlers(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	creds auth.Credentials, t *testing.T) {
-
+	creds auth.Credentials, t *testing.T,
+) {
 	// test cases with sample input and expected output.
 	testCases := []struct {
 		method     string
@@ -266,8 +266,8 @@ func testBucketLifecycle(obj ObjectLayer, instanceType, bucketName string, apiRo
 		lifecycleResponse  []byte
 		errorResponse      APIErrorResponse
 		shouldPass         bool
-	}) {
-
+	},
+) {
 	for i, testCase := range testCases {
 		// initialize httptest Recorder, this records any mutations to response writer inside the handler.
 		rec := httptest.NewRecorder()

cmd/bucket-lifecycle.go

@@ -165,6 +165,9 @@ type transitionState struct {
 	killCh     chan struct{}
 
 	activeTasks int32
+
+	lastDayMu    sync.RWMutex
+	lastDayStats map[string]*lastDayTierStats
 }
 
 func (t *transitionState) queueTransitionTask(oi ObjectInfo) {
@@ -178,9 +181,7 @@ func (t *transitionState) queueTransitionTask(oi ObjectInfo) {
 	}
 }
 
-var (
-	globalTransitionState *transitionState
-)
+var globalTransitionState *transitionState
 
 func newTransitionState(ctx context.Context, objAPI ObjectLayer) *transitionState {
 	return &transitionState{
@@ -188,6 +189,7 @@ func newTransitionState(ctx context.Context, objAPI ObjectLayer) *transitionStat
 		ctx:          ctx,
 		objAPI:       objAPI,
 		killCh:       make(chan struct{}),
+		lastDayStats: make(map[string]*lastDayTierStats),
 	}
 }
 
@@ -215,14 +217,46 @@ func (t *transitionState) worker(ctx context.Context, objectAPI ObjectLayer) {
 				return
 			}
 			atomic.AddInt32(&t.activeTasks, 1)
-			if err := transitionObject(ctx, objectAPI, oi); err != nil {
+			var tier string
+			var err error
+			if tier, err = transitionObject(ctx, objectAPI, oi); err != nil {
 				logger.LogIf(ctx, fmt.Errorf("Transition failed for %s/%s version:%s with %w", oi.Bucket, oi.Name, oi.VersionID, err))
 			}
 			atomic.AddInt32(&t.activeTasks, -1)
+
+			ts := tierStats{
+				TotalSize:   uint64(oi.Size),
+				NumVersions: 1,
+			}
+			if oi.IsLatest {
+				ts.NumObjects = 1
+			}
+			t.addLastDayStats(tier, ts)
 		}
 	}
 }
 
+func (t *transitionState) addLastDayStats(tier string, ts tierStats) {
+	t.lastDayMu.Lock()
+	defer t.lastDayMu.Unlock()
+
+	if _, ok := t.lastDayStats[tier]; !ok {
+		t.lastDayStats[tier] = &lastDayTierStats{}
+	}
+	t.lastDayStats[tier].addStats(ts)
+}
+
+func (t *transitionState) getDailyAllTierStats() dailyAllTierStats {
+	t.lastDayMu.RLock()
+	defer t.lastDayMu.RUnlock()
+
+	res := make(dailyAllTierStats, len(t.lastDayStats))
+	for tier, st := range t.lastDayStats {
+		res[tier] = st.clone()
+	}
+	return res
+}
+
 // UpdateWorkers at the end of this function leaves n goroutines waiting for
 // transition tasks
 func (t *transitionState) UpdateWorkers(n int) {
@@ -367,15 +401,16 @@ func genTransitionObjName(bucket string) (string, error) {
 // storage specified by the transition ARN, the metadata is left behind on source cluster and original content
 // is moved to the transition tier. Note that in the case of encrypted objects, entire encrypted stream is moved
 // to the transition tier without decrypting or re-encrypting.
-func transitionObject(ctx context.Context, objectAPI ObjectLayer, oi ObjectInfo) error {
+func transitionObject(ctx context.Context, objectAPI ObjectLayer, oi ObjectInfo) (string, error) {
 	lc, err := globalLifecycleSys.Get(oi.Bucket)
 	if err != nil {
-		return err
+		return "", err
 	}
+	tier := lc.TransitionTier(oi.ToLifecycleOpts())
 	opts := ObjectOptions{
 		Transition: TransitionOptions{
 			Status: lifecycle.TransitionPending,
-			Tier:   lc.TransitionTier(oi.ToLifecycleOpts()),
+			Tier:   tier,
 			ETag:   oi.ETag,
 		},
 		VersionID:        oi.VersionID,
@@ -383,7 +418,7 @@ func transitionObject(ctx context.Context, objectAPI ObjectLayer, oi ObjectInfo)
 		VersionSuspended: globalBucketVersioningSys.Suspended(oi.Bucket),
 		MTime:            oi.ModTime,
 	}
-	return objectAPI.TransitionObject(ctx, oi.Bucket, oi.Name, opts)
+	return tier, objectAPI.TransitionObject(ctx, oi.Bucket, oi.Name, opts)
 }
 
 // getTransitionedObjectReader returns a reader from the transitioned tier.
@@ -466,9 +501,7 @@ func (sp *SelectParameters) IsEmpty() bool {
 	return sp == nil
 }
 
-var (
-	selectParamsXMLName = "SelectParameters"
-)
+var selectParamsXMLName = "SelectParameters"
 
 // UnmarshalXML - decodes XML data.
 func (sp *SelectParameters) UnmarshalXML(d *xml.Decoder, start xml.StartElement) error {

cmd/bucket-metadata.go

@@ -120,7 +120,7 @@ func (b *BucketMetadata) Load(ctx context.Context, api ObjectLayer, name string)
 		logger.LogIf(ctx, errors.New("bucket name cannot be empty"))
 		return errors.New("bucket name cannot be empty")
 	}
-	configFile := path.Join(bucketConfigPrefix, name, bucketMetadataFile)
+	configFile := path.Join(bucketMetaPrefix, name, bucketMetadataFile)
 	data, err := readConfig(ctx, api, configFile)
 	if err != nil {
 		return err
@@ -277,7 +277,7 @@ func (b *BucketMetadata) convertLegacyConfigs(ctx context.Context, objectAPI Obj
 	}
 
 	for _, legacyFile := range legacyConfigs {
-		configFile := path.Join(bucketConfigPrefix, b.Name, legacyFile)
+		configFile := path.Join(bucketMetaPrefix, b.Name, legacyFile)
 
 		configData, err := readConfig(ctx, objectAPI, configFile)
 		if err != nil {
@@ -338,7 +338,7 @@ func (b *BucketMetadata) convertLegacyConfigs(ctx context.Context, objectAPI Obj
 	}
 
 	for legacyFile := range configs {
-		configFile := path.Join(bucketConfigPrefix, b.Name, legacyFile)
+		configFile := path.Join(bucketMetaPrefix, b.Name, legacyFile)
 		if err := deleteConfig(ctx, objectAPI, configFile); err != nil && !errors.Is(err, errConfigNotFound) {
 			logger.LogIf(ctx, err)
 		}
@@ -365,7 +365,7 @@ func (b *BucketMetadata) Save(ctx context.Context, api ObjectLayer) error {
 		return err
 	}
 
-	configFile := path.Join(bucketConfigPrefix, b.Name, bucketMetadataFile)
+	configFile := path.Join(bucketMetaPrefix, b.Name, bucketMetadataFile)
 	return saveConfig(ctx, api, configFile, data)
 }
 
@@ -377,7 +377,7 @@ func deleteBucketMetadata(ctx context.Context, obj objectDeleter, bucket string)
 		bucketMetadataFile,
 	}
 	for _, metaFile := range metadataFiles {
-		configFile := path.Join(bucketConfigPrefix, bucket, metaFile)
+		configFile := path.Join(bucketMetaPrefix, bucket, metaFile)
 		if err := deleteConfig(ctx, obj, configFile); err != nil && err != errConfigNotFound {
 			return err
 		}

cmd/bucket-notification-handlers.go

@@ -30,7 +30,6 @@ import (
 )
 
 const (
-	bucketConfigPrefix       = "buckets"
 	bucketNotificationConfig = "notification.xml"
 )
 

cmd/bucket-object-lock.go

@@ -58,6 +58,10 @@ func (sys *BucketObjectLockSys) Get(bucketName string) (r objectlock.Retention,
 // enforceRetentionForDeletion checks if it is appropriate to remove an
 // object according to locking configuration when this is lifecycle/ bucket quota asking.
 func enforceRetentionForDeletion(ctx context.Context, objInfo ObjectInfo) (locked bool) {
+	if objInfo.DeleteMarker {
+		return false
+	}
+
 	lhold := objectlock.GetObjectLegalHoldMeta(objInfo.UserDefined)
 	if lhold.Status.Valid() && lhold.Status == objectlock.LegalHoldOn {
 		return true

cmd/bucket-policy-handlers_test.go

@@ -37,60 +37,72 @@ import (
 func getAnonReadOnlyBucketPolicy(bucketName string) *policy.Policy {
 	return &policy.Policy{
 		Version: policy.DefaultVersion,
-		Statements: []policy.Statement{policy.NewStatement(
-			policy.Allow,
-			policy.NewPrincipal("*"),
-			policy.NewActionSet(policy.GetBucketLocationAction, policy.ListBucketAction),
-			policy.NewResourceSet(policy.NewResource(bucketName, "")),
-			condition.NewFunctions(),
-		)},
+		Statements: []policy.Statement{
+			policy.NewStatement(
+				"",
+				policy.Allow,
+				policy.NewPrincipal("*"),
+				policy.NewActionSet(policy.GetBucketLocationAction, policy.ListBucketAction),
+				policy.NewResourceSet(policy.NewResource(bucketName, "")),
+				condition.NewFunctions(),
+			),
+		},
 	}
 }
 
 func getAnonWriteOnlyBucketPolicy(bucketName string) *policy.Policy {
 	return &policy.Policy{
 		Version: policy.DefaultVersion,
-		Statements: []policy.Statement{policy.NewStatement(
-			policy.Allow,
-			policy.NewPrincipal("*"),
-			policy.NewActionSet(
-				policy.GetBucketLocationAction,
-				policy.ListBucketMultipartUploadsAction,
+		Statements: []policy.Statement{
+			policy.NewStatement(
+				"",
+				policy.Allow,
+				policy.NewPrincipal("*"),
+				policy.NewActionSet(
+					policy.GetBucketLocationAction,
+					policy.ListBucketMultipartUploadsAction,
+				),
+				policy.NewResourceSet(policy.NewResource(bucketName, "")),
+				condition.NewFunctions(),
 			),
-			policy.NewResourceSet(policy.NewResource(bucketName, "")),
-			condition.NewFunctions(),
-		)},
+		},
 	}
 }
 
 func getAnonReadOnlyObjectPolicy(bucketName, prefix string) *policy.Policy {
 	return &policy.Policy{
 		Version: policy.DefaultVersion,
-		Statements: []policy.Statement{policy.NewStatement(
-			policy.Allow,
-			policy.NewPrincipal("*"),
-			policy.NewActionSet(policy.GetObjectAction),
-			policy.NewResourceSet(policy.NewResource(bucketName, prefix)),
-			condition.NewFunctions(),
-		)},
+		Statements: []policy.Statement{
+			policy.NewStatement(
+				"",
+				policy.Allow,
+				policy.NewPrincipal("*"),
+				policy.NewActionSet(policy.GetObjectAction),
+				policy.NewResourceSet(policy.NewResource(bucketName, prefix)),
+				condition.NewFunctions(),
+			),
+		},
 	}
 }
 
 func getAnonWriteOnlyObjectPolicy(bucketName, prefix string) *policy.Policy {
 	return &policy.Policy{
 		Version: policy.DefaultVersion,
-		Statements: []policy.Statement{policy.NewStatement(
-			policy.Allow,
-			policy.NewPrincipal("*"),
-			policy.NewActionSet(
-				policy.AbortMultipartUploadAction,
-				policy.DeleteObjectAction,
-				policy.ListMultipartUploadPartsAction,
-				policy.PutObjectAction,
+		Statements: []policy.Statement{
+			policy.NewStatement(
+				"",
+				policy.Allow,
+				policy.NewPrincipal("*"),
+				policy.NewActionSet(
+					policy.AbortMultipartUploadAction,
+					policy.DeleteObjectAction,
+					policy.ListMultipartUploadPartsAction,
+					policy.PutObjectAction,
+				),
+				policy.NewResourceSet(policy.NewResource(bucketName, prefix)),
+				condition.NewFunctions(),
 			),
-			policy.NewResourceSet(policy.NewResource(bucketName, prefix)),
-			condition.NewFunctions(),
-		)},
+		},
 	}
 }
 
@@ -105,7 +117,7 @@ func testCreateBucket(obj ObjectLayer, instanceType, bucketName string, apiRoute
 	bucketName1 := fmt.Sprintf("%s-1", bucketName)
 
 	const n = 100
-	var start = make(chan struct{})
+	start := make(chan struct{})
 	var ok, errs int
 	var wg sync.WaitGroup
 	var mu sync.Mutex
@@ -147,8 +159,8 @@ func TestPutBucketPolicyHandler(t *testing.T) {
 
 // testPutBucketPolicyHandler - Test for Bucket policy end point.
 func testPutBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	bucketName1 := fmt.Sprintf("%s-1", bucketName)
 	if err := obj.MakeBucketWithLocation(GlobalContext, bucketName1, BucketOptions{}); err != nil {
 		t.Fatal(err)
@@ -333,7 +345,6 @@ func testPutBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 	// create unsigned HTTP request for PutBucketPolicyHandler.
 	anonReq, err := newTestRequest(http.MethodPut, getPutPolicyURL("", bucketName),
 		int64(len(bucketPolicyStr)), bytes.NewReader([]byte(bucketPolicyStr)))
-
 	if err != nil {
 		t.Fatalf("MinIO %s: Failed to create an anonymous request for bucket \"%s\": <ERROR> %v",
 			instanceType, bucketName, err)
@@ -352,14 +363,12 @@ func testPutBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 
 	nilReq, err := newTestSignedRequestV4(http.MethodPut, getPutPolicyURL("", nilBucket),
 		0, nil, "", "", nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}
 	// execute the object layer set to `nil` test.
 	// `ExecObjectLayerAPINilTest` manages the operation.
 	ExecObjectLayerAPINilTest(t, nilBucket, "", instanceType, apiRouter, nilReq)
-
 }
 
 // Wrapper for calling Get Bucket Policy HTTP handler tests for both Erasure multiple disks and single node setup.
@@ -465,7 +474,6 @@ func testGetBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 		// construct HTTP request for PUT bucket policy endpoint.
 		reqV4, err := newTestSignedRequestV4(http.MethodGet, getGetPolicyURL("", testCase.bucketName),
 			0, nil, testCase.accessKey, testCase.secretKey, nil)
-
 		if err != nil {
 			t.Fatalf("Test %d: Failed to create HTTP request for GetBucketPolicyHandler: <ERROR> %v", i+1, err)
 		}
@@ -540,7 +548,6 @@ func testGetBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 	// Bucket policy related functions doesn't support anonymous requests, setting policies shouldn't make a difference.
 	// create unsigned HTTP request for PutBucketPolicyHandler.
 	anonReq, err := newTestRequest(http.MethodGet, getPutPolicyURL("", bucketName), 0, nil)
-
 	if err != nil {
 		t.Fatalf("MinIO %s: Failed to create an anonymous request for bucket \"%s\": <ERROR> %v",
 			instanceType, bucketName, err)
@@ -559,7 +566,6 @@ func testGetBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string
 
 	nilReq, err := newTestSignedRequestV4(http.MethodGet, getGetPolicyURL("", nilBucket),
 		0, nil, "", "", nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}
@@ -575,8 +581,8 @@ func TestDeleteBucketPolicyHandler(t *testing.T) {
 
 // testDeleteBucketPolicyHandler - Test for Delete bucket policy end point.
 func testDeleteBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName string, apiRouter http.Handler,
-	credentials auth.Credentials, t *testing.T) {
-
+	credentials auth.Credentials, t *testing.T,
+) {
 	// template for constructing HTTP request body for PUT bucket policy.
 	bucketPolicyTemplate := `{
     "Version": "2012-10-17",
@@ -743,7 +749,6 @@ func testDeleteBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName str
 	// Bucket policy related functions doesn't support anonymous requests, setting policies shouldn't make a difference.
 	// create unsigned HTTP request for PutBucketPolicyHandler.
 	anonReq, err := newTestRequest(http.MethodDelete, getPutPolicyURL("", bucketName), 0, nil)
-
 	if err != nil {
 		t.Fatalf("MinIO %s: Failed to create an anonymous request for bucket \"%s\": <ERROR> %v",
 			instanceType, bucketName, err)
@@ -762,7 +767,6 @@ func testDeleteBucketPolicyHandler(obj ObjectLayer, instanceType, bucketName str
 
 	nilReq, err := newTestSignedRequestV4(http.MethodDelete, getDeletePolicyURL("", nilBucket),
 		0, nil, "", "", nil)
-
 	if err != nil {
 		t.Errorf("MinIO %s: Failed to create HTTP request for testing the response when object Layer is set to `nil`.", instanceType)
 	}

cmd/bucket-policy.go

@@ -200,7 +200,7 @@ func PolicyToBucketAccessPolicy(bucketPolicy *policy.Policy) (*miniogopolicy.Buc
 	}
 
 	var policyInfo miniogopolicy.BucketAccessPolicy
-	var json = jsoniter.ConfigCompatibleWithStandardLibrary
+	json := jsoniter.ConfigCompatibleWithStandardLibrary
 	if err = json.Unmarshal(data, &policyInfo); err != nil {
 		// This should not happen because data is valid to JSON data.
 		return nil, err
@@ -218,7 +218,7 @@ func BucketAccessPolicyToPolicy(policyInfo *miniogopolicy.BucketAccessPolicy) (*
 	}
 
 	var bucketPolicy policy.Policy
-	var json = jsoniter.ConfigCompatibleWithStandardLibrary
+	json := jsoniter.ConfigCompatibleWithStandardLibrary
 	if err = json.Unmarshal(data, &bucketPolicy); err != nil {
 		// This should not happen because data is valid to JSON data.
 		return nil, err

cmd/bucket-quota.go

@@ -154,7 +154,7 @@ func enforceFIFOQuotaBucket(ctx context.Context, objectAPI ObjectLayer, bucket s
 	// ModTime to find the oldest objects in bucket to delete. In
 	// the context of bucket quota enforcement - number of hits are
 	// irrelevant.
-	scorer, err := newFileScorer(toFree, time.Now().Unix(), 1)
+	scorer, err := newFileScorer(toFree, UTCNow().Unix(), 1)
 	if err != nil {
 		logger.LogIf(ctx, err)
 		return
@@ -162,11 +162,6 @@ func enforceFIFOQuotaBucket(ctx context.Context, objectAPI ObjectLayer, bucket s
 
 	rcfg, _ := globalBucketObjectLockSys.Get(bucket)
 	for obj := range objInfoCh {
-		if obj.DeleteMarker {
-			// Delete markers are automatically added for FIFO purge.
-			scorer.addFileWithObjInfo(obj, 1)
-			continue
-		}
 		// skip objects currently under retention
 		if rcfg.LockEnabled && enforceRetentionForDeletion(ctx, obj) {
 			continue
@@ -178,6 +173,7 @@ func enforceFIFOQuotaBucket(ctx context.Context, objectAPI ObjectLayer, bucket s
 	if scorer.seenBytes <= cfg.Quota {
 		return
 	}
+
 	// Calculate how much we want to delete now.
 	toFreeNow := scorer.seenBytes - cfg.Quota
 	// We were less over quota than we thought. Adjust so we delete less.
@@ -193,8 +189,10 @@ func enforceFIFOQuotaBucket(ctx context.Context, objectAPI ObjectLayer, bucket s
 	numKeys := len(scorer.fileObjInfos())
 	for i, obj := range scorer.fileObjInfos() {
 		objects = append(objects, ObjectToDelete{
-			ObjectName: obj.Name,
-			VersionID:  obj.VersionID,
+			ObjectV: ObjectV{
+				ObjectName: obj.Name,
+				VersionID:  obj.VersionID,
+			},
 		})
 		if len(objects) < maxDeleteList && (i < numKeys-1) {
 			// skip deletion until maxDeleteList or end of slice
@@ -209,9 +207,9 @@ func enforceFIFOQuotaBucket(ctx context.Context, objectAPI ObjectLayer, bucket s
 		_, deleteErrs := objectAPI.DeleteObjects(ctx, bucket, objects, ObjectOptions{
 			Versioned: versioned,
 		})
-		for i := range deleteErrs {
-			if deleteErrs[i] != nil {
-				logger.LogIf(ctx, deleteErrs[i])
+		for j := range deleteErrs {
+			if deleteErrs[j] != nil {
+				logger.LogIf(ctx, deleteErrs[j])
 				continue
 			}
 

cmd/bucket-replication-stats.go

@@ -55,7 +55,6 @@ func (r *ReplicationStats) Delete(bucket string) {
 	r.ulock.Lock()
 	defer r.ulock.Unlock()
 	delete(r.UsageCache, bucket)
-
 }
 
 // UpdateReplicaStat updates in-memory replica statistics with new values.

cmd/bucket-replication-utils.go

@@ -502,8 +502,10 @@ func getHealReplicateObjectInfo(objInfo ObjectInfo, rcfg replicationConfig) Repl
 	var tgtStatuses map[string]replication.StatusType
 	if oi.DeleteMarker || !oi.VersionPurgeStatus.Empty() {
 		dsc = checkReplicateDelete(GlobalContext, oi.Bucket, ObjectToDelete{
-			ObjectName: oi.Name,
-			VersionID:  oi.VersionID,
+			ObjectV: ObjectV{
+				ObjectName: oi.Name,
+				VersionID:  oi.VersionID,
+			},
 		}, oi, ObjectOptions{}, nil)
 	} else {
 		dsc = mustReplicate(GlobalContext, oi.Bucket, oi.Name, getMustReplicateOptions(ObjectInfo{

cmd/bucket-replication-utils_test.go

@@ -77,7 +77,8 @@ var replicatedInfosTests = []struct {
 				ReplicationStatus:     replication.Failed,
 				OpType:                replication.ObjectReplicationType,
 				ReplicationAction:     replicateAll,
-			}},
+			},
+		},
 		expectedCompletedSize:             249,
 		expectedReplicationStatusInternal: "arn1=COMPLETED;arn2=FAILED;",
 		expectedReplicationStatus:         replication.Failed,
@@ -102,7 +103,8 @@ var replicatedInfosTests = []struct {
 				ReplicationStatus:     replication.Failed,
 				OpType:                replication.ObjectReplicationType,
 				ReplicationAction:     replicateAll,
-			}},
+			},
+		},
 		expectedCompletedSize:             0,
 		expectedReplicationStatusInternal: "arn1=PENDING;arn2=FAILED;",
 		expectedReplicationStatus:         replication.Failed,
@@ -182,7 +184,6 @@ var parseReplicationDecisionTest = []struct {
 func TestParseReplicateDecision(t *testing.T) {
 	for i, test := range parseReplicationDecisionTest {
 		dsc, err := parseReplicateDecision(test.expDsc.String())
-
 		if err != nil {
 			if test.expErr != err {
 				t.Errorf("Test%d (%s): Expected parse error got %t , want %t", i+1, test.name, err, test.expErr)

cmd/bucket-replication.go

@@ -28,6 +28,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/dustin/go-humanize"
 	"github.com/minio/madmin-go"
 	"github.com/minio/minio-go/v7"
 	miniogo "github.com/minio/minio-go/v7"
@@ -139,6 +140,7 @@ func (o mustReplicateOptions) ReplicationStatus() (s replication.StatusType) {
 	}
 	return s
 }
+
 func (o mustReplicateOptions) isExistingObjectReplication() bool {
 	return o.opType == replication.ExistingObjectReplicationType
 }
@@ -146,6 +148,7 @@ func (o mustReplicateOptions) isExistingObjectReplication() bool {
 func (o mustReplicateOptions) isMetadataReplication() bool {
 	return o.opType == replication.MetadataReplicationType
 }
+
 func getMustReplicateOptions(o ObjectInfo, op replication.Type, opts ObjectOptions) mustReplicateOptions {
 	if !op.Valid() {
 		op = replication.ObjectReplicationType
@@ -441,7 +444,7 @@ func replicateDelete(ctx context.Context, dobj DeletedObjectReplicationInfo, obj
 		}
 	}
 
-	var eventName = event.ObjectReplicationComplete
+	eventName := event.ObjectReplicationComplete
 	if replicationStatus == replication.Failed {
 		eventName = event.ObjectReplicationFailed
 	}
@@ -523,7 +526,8 @@ func replicateDeleteToTarget(ctx context.Context, dobj DeletedObjectReplicationI
 			VersionID: versionID,
 			Internal: miniogo.AdvancedGetOptions{
 				ReplicationProxyRequest: "false",
-			}}); isErrMethodNotAllowed(ErrorRespToObjectError(err, dobj.Bucket, dobj.ObjectName)) {
+			},
+		}); isErrMethodNotAllowed(ErrorRespToObjectError(err, dobj.Bucket, dobj.ObjectName)) {
 			if dobj.VersionID == "" {
 				rinfo.ReplicationStatus = replication.Completed
 				return
@@ -766,7 +770,7 @@ func getReplicationAction(oi1 ObjectInfo, oi2 minio.ObjectInfo, opType replicati
 	}
 
 	t, _ := tags.ParseObjectTags(oi1.UserTags)
-	if !reflect.DeepEqual(oi2.UserTags, t.ToMap()) {
+	if !reflect.DeepEqual(oi2.UserTags, t.ToMap()) || (oi2.UserTagCount != len(t.ToMap())) {
 		return replicateMetadata
 	}
 
@@ -857,8 +861,9 @@ func replicateObject(ctx context.Context, ri ReplicateObjectInfo, objectAPI Obje
 		return
 	}
 	tgtArns := cfg.FilterTargetArns(replication.ObjectOpts{
-		Name: object,
-		SSEC: crypto.SSEC.IsEncrypted(objInfo.UserDefined),
+		Name:     object,
+		SSEC:     crypto.SSEC.IsEncrypted(objInfo.UserDefined),
+		UserTags: objInfo.UserTags,
 	})
 	// Lock the object name before starting replication.
 	// Use separate lock that doesn't collide with regular objects.
@@ -902,7 +907,7 @@ func replicateObject(ctx context.Context, ri ReplicateObjectInfo, objectAPI Obje
 	// FIXME: add support for missing replication events
 	// - event.ObjectReplicationMissedThreshold
 	// - event.ObjectReplicationReplicatedAfterThreshold
-	var eventName = event.ObjectReplicationComplete
+	eventName := event.ObjectReplicationComplete
 	if rinfos.ReplicationStatus() == replication.Failed {
 		eventName = event.ObjectReplicationFailed
 	}
@@ -1058,7 +1063,8 @@ func replicateObjectToTarget(ctx context.Context, ri ReplicateObjectInfo, object
 		VersionID: objInfo.VersionID,
 		Internal: miniogo.AdvancedGetOptions{
 			ReplicationProxyRequest: "false",
-		}})
+		},
+	})
 	if cerr == nil {
 		rAction = getReplicationAction(objInfo, oi, ri.OpType)
 		rinfo.ReplicationStatus = replication.Completed
@@ -1117,7 +1123,8 @@ func replicateObjectToTarget(ctx context.Context, ri ReplicateObjectInfo, object
 			Internal: miniogo.AdvancedPutOptions{
 				SourceVersionID:    objInfo.VersionID,
 				ReplicationRequest: true, // always set this to distinguish between `mc mirror` replication and serverside
-			}}
+			},
+		}
 		if _, err = c.CopyObject(ctx, tgt.Bucket, object, tgt.Bucket, object, getCopyObjMetadata(objInfo, tgt.StorageClass), srcOpts, dstOpts); err != nil {
 			rinfo.ReplicationStatus = replication.Failed
 			logger.LogIf(ctx, fmt.Errorf("Unable to replicate metadata for object %s/%s(%s): %s", bucket, objInfo.Name, objInfo.VersionID, err))
@@ -1179,9 +1186,17 @@ func replicateObjectWithMultipart(ctx context.Context, c *miniogo.Core, bucket,
 	defer func() {
 		if err != nil {
 			// block and abort remote upload upon failure.
-			if aerr := c.AbortMultipartUpload(ctx, bucket, object, uploadID); aerr != nil {
-				aerr = fmt.Errorf("Unable to cleanup failed multipart replication %s on remote %s/%s: %w", uploadID, bucket, object, aerr)
-				logger.LogIf(ctx, aerr)
+			attempts := 1
+			for attempts <= 3 {
+				aerr := c.AbortMultipartUpload(ctx, bucket, object, uploadID)
+				if aerr == nil {
+					return
+				}
+				logger.LogIf(ctx,
+					fmt.Errorf("Trying %s: Unable to cleanup failed multipart replication %s on remote %s/%s: %w - this may consume space on remote cluster",
+						humanize.Ordinal(attempts), uploadID, bucket, object, aerr))
+				attempts++
+				time.Sleep(time.Second)
 			}
 		}
 	}()
@@ -1213,7 +1228,8 @@ func replicateObjectWithMultipart(ctx context.Context, c *miniogo.Core, bucket,
 			SourceMTime: objInfo.ModTime,
 			// always set this to distinguish between `mc mirror` replication and serverside
 			ReplicationRequest: true,
-		}})
+		},
+	})
 	return err
 }
 
@@ -1357,7 +1373,6 @@ func (p *ReplicationPool) AddWorker() {
 			return
 		}
 	}
-
 }
 
 // AddExistingObjectReplicateWorker adds a worker to queue existing objects that need to be sync'd
@@ -1671,6 +1686,7 @@ type replicationConfig struct {
 func (c replicationConfig) Empty() bool {
 	return c.Config == nil
 }
+
 func (c replicationConfig) Replicate(opts replication.ObjectOpts) bool {
 	return c.Config.Replicate(opts)
 }
@@ -1694,7 +1710,8 @@ func (c replicationConfig) Resync(ctx context.Context, oi ObjectInfo, dsc *Repli
 			DeleteMarker:   oi.DeleteMarker,
 			VersionID:      oi.VersionID,
 			OpType:         replication.DeleteReplicationType,
-			ExistingObject: true}
+			ExistingObject: true,
+		}
 
 		tgtArns := c.Config.FilterTargetArns(opts)
 		// indicates no matching target with Existing object replication enabled.

cmd/bucket-replication_test.go

@@ -75,7 +75,8 @@ var replicationConfigTests = []struct {
 	},
 	{ // 4. existing object replication enabled, versioning enabled; no reset in progress
 		name: "existing object replication enabled, versioning enabled; no reset in progress",
-		info: ObjectInfo{Size: 100,
+		info: ObjectInfo{
+			Size:              100,
 			ReplicationStatus: replication.Completed,
 			VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
 		},
@@ -93,174 +94,192 @@ func TestReplicationResync(t *testing.T) {
 	}
 }
 
-var start = UTCNow().AddDate(0, 0, -1)
-var replicationConfigTests2 = []struct {
-	info         ObjectInfo
-	name         string
-	rcfg         replicationConfig
-	dsc          ReplicateDecision
-	tgtStatuses  map[string]replication.StatusType
-	expectedSync bool
-}{
-	{ // Cases 1-4: existing object replication enabled, versioning enabled, no reset - replication status varies
-		// 1: Pending replication
-		name: "existing object replication on object in Pending replication status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:PENDING;",
-			ReplicationStatus:         replication.Pending,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
-		},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn: "arn1",
-		}}}},
-		dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		expectedSync: true,
-	},
-
-	{ // 2. replication status Failed
-		name: "existing object replication on object in Failed replication status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:FAILED",
-			ReplicationStatus:         replication.Failed,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
-		},
-		dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn: "arn1",
-		}}}},
-		expectedSync: true,
-	},
-	{ // 3. replication status unset
-		name: "existing object replication on pre-existing unreplicated object",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatus: replication.StatusType(""),
-			VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
-		},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn: "arn1",
-		}}}},
-		dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		expectedSync: true,
-	},
-	{ // 4. replication status Complete
-		name: "existing object replication on object in Completed replication status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:COMPLETED",
-			ReplicationStatus:         replication.Completed,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
-		},
-		dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", false, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn: "arn1",
-		}}}},
-		expectedSync: false,
-	},
-	{ // 5. existing object replication enabled, versioning enabled, replication status Pending & reset ID present
-		name: "existing object replication with reset in progress and object in Pending status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:PENDING;",
-			ReplicationStatus:         replication.Pending,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
-			UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
-		},
-		expectedSync: true,
-		dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn:             "arn1",
-			ResetID:         "xyz",
-			ResetBeforeDate: UTCNow(),
-		}}},
-		},
-	},
-	{ // 6. existing object replication enabled, versioning enabled, replication status Failed & reset ID present
-		name: "existing object replication with reset in progress and object in Failed status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:FAILED;",
-			ReplicationStatus:         replication.Failed,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
-			UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
-		},
-		dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn:             "arn1",
-			ResetID:         "xyz",
-			ResetBeforeDate: UTCNow(),
-		}}},
-		},
-		expectedSync: true,
-	},
-	{ // 7. existing object replication enabled, versioning enabled, replication status unset & reset ID present
-		name: "existing object replication with reset in progress and object never replicated before",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatus: replication.StatusType(""),
-			VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
-			UserDefined:       map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
-		},
-		dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn:             "arn1",
-			ResetID:         "xyz",
-			ResetBeforeDate: UTCNow(),
-		}}},
+var (
+	start                   = UTCNow().AddDate(0, 0, -1)
+	replicationConfigTests2 = []struct {
+		info         ObjectInfo
+		name         string
+		rcfg         replicationConfig
+		dsc          ReplicateDecision
+		tgtStatuses  map[string]replication.StatusType
+		expectedSync bool
+	}{
+		{ // Cases 1-4: existing object replication enabled, versioning enabled, no reset - replication status varies
+			// 1: Pending replication
+			name: "existing object replication on object in Pending replication status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:PENDING;",
+				ReplicationStatus:         replication.Pending,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
+			},
+			rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+				Arn: "arn1",
+			}}}},
+			dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			expectedSync: true,
 		},
 
-		expectedSync: true,
-	},
+		{ // 2. replication status Failed
+			name: "existing object replication on object in Failed replication status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:FAILED",
+				ReplicationStatus:         replication.Failed,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
+			},
+			dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+				Arn: "arn1",
+			}}}},
+			expectedSync: true,
+		},
+		{ // 3. replication status unset
+			name: "existing object replication on pre-existing unreplicated object",
+			info: ObjectInfo{
+				Size:              100,
+				ReplicationStatus: replication.StatusType(""),
+				VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
+			},
+			rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+				Arn: "arn1",
+			}}}},
+			dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			expectedSync: true,
+		},
+		{ // 4. replication status Complete
+			name: "existing object replication on object in Completed replication status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:COMPLETED",
+				ReplicationStatus:         replication.Completed,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
+			},
+			dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", false, false)}},
+			rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+				Arn: "arn1",
+			}}}},
+			expectedSync: false,
+		},
+		{ // 5. existing object replication enabled, versioning enabled, replication status Pending & reset ID present
+			name: "existing object replication with reset in progress and object in Pending status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:PENDING;",
+				ReplicationStatus:         replication.Pending,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
+				UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
+			},
+			expectedSync: true,
+			dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{
+				remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+					Arn:             "arn1",
+					ResetID:         "xyz",
+					ResetBeforeDate: UTCNow(),
+				}}},
+			},
+		},
+		{ // 6. existing object replication enabled, versioning enabled, replication status Failed & reset ID present
+			name: "existing object replication with reset in progress and object in Failed status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:FAILED;",
+				ReplicationStatus:         replication.Failed,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
+				UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
+			},
+			dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{
+				remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+					Arn:             "arn1",
+					ResetID:         "xyz",
+					ResetBeforeDate: UTCNow(),
+				}}},
+			},
+			expectedSync: true,
+		},
+		{ // 7. existing object replication enabled, versioning enabled, replication status unset & reset ID present
+			name: "existing object replication with reset in progress and object never replicated before",
+			info: ObjectInfo{
+				Size:              100,
+				ReplicationStatus: replication.StatusType(""),
+				VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
+				UserDefined:       map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
+			},
+			dsc: ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{
+				remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+					Arn:             "arn1",
+					ResetID:         "xyz",
+					ResetBeforeDate: UTCNow(),
+				}}},
+			},
 
-	{ // 8. existing object replication enabled, versioning enabled, replication status Complete & reset ID present
-		name: "existing object replication enabled - reset in progress for an object in Completed status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:COMPLETED;",
-			ReplicationStatus:         replication.Completed,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df8",
-			UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
+			expectedSync: true,
 		},
-		expectedSync: true,
-		dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn:             "arn1",
-			ResetID:         "xyz",
-			ResetBeforeDate: UTCNow(),
-		}}},
-		},
-	},
-	{ // 9. existing object replication enabled, versioning enabled, replication status Pending & reset ID different
-		name: "existing object replication enabled, newer reset in progress on object in Pending replication status",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:PENDING;",
 
-			ReplicationStatus: replication.Pending,
-			VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
-			UserDefined:       map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;%s", UTCNow().AddDate(0, 0, -1).Format(http.TimeFormat), "abc")},
-			ModTime:           UTCNow().AddDate(0, 0, -2),
+		{ // 8. existing object replication enabled, versioning enabled, replication status Complete & reset ID present
+			name: "existing object replication enabled - reset in progress for an object in Completed status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:COMPLETED;",
+				ReplicationStatus:         replication.Completed,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df8",
+				UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;abc", UTCNow().AddDate(0, -1, 0).String())},
+			},
+			expectedSync: true,
+			dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{
+				remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+					Arn:             "arn1",
+					ResetID:         "xyz",
+					ResetBeforeDate: UTCNow(),
+				}}},
+			},
 		},
-		expectedSync: true,
-		dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn:             "arn1",
-			ResetID:         "xyz",
-			ResetBeforeDate: UTCNow(),
-		}}},
+		{ // 9. existing object replication enabled, versioning enabled, replication status Pending & reset ID different
+			name: "existing object replication enabled, newer reset in progress on object in Pending replication status",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:PENDING;",
+
+				ReplicationStatus: replication.Pending,
+				VersionID:         "a3348c34-c352-4498-82f0-1098e8b34df9",
+				UserDefined:       map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;%s", UTCNow().AddDate(0, 0, -1).Format(http.TimeFormat), "abc")},
+				ModTime:           UTCNow().AddDate(0, 0, -2),
+			},
+			expectedSync: true,
+			dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{
+				remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+					Arn:             "arn1",
+					ResetID:         "xyz",
+					ResetBeforeDate: UTCNow(),
+				}}},
+			},
 		},
-	},
-	{ // 10. existing object replication enabled, versioning enabled, replication status Complete & reset done
-		name: "reset done on object in Completed Status - ineligbile for re-replication",
-		info: ObjectInfo{Size: 100,
-			ReplicationStatusInternal: "arn1:COMPLETED;",
-			ReplicationStatus:         replication.Completed,
-			VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
-			UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;%s", start.Format(http.TimeFormat), "xyz")},
+		{ // 10. existing object replication enabled, versioning enabled, replication status Complete & reset done
+			name: "reset done on object in Completed Status - ineligbile for re-replication",
+			info: ObjectInfo{
+				Size:                      100,
+				ReplicationStatusInternal: "arn1:COMPLETED;",
+				ReplicationStatus:         replication.Completed,
+				VersionID:                 "a3348c34-c352-4498-82f0-1098e8b34df9",
+				UserDefined:               map[string]string{xhttp.MinIOReplicationResetStatus: fmt.Sprintf("%s;%s", start.Format(http.TimeFormat), "xyz")},
+			},
+			expectedSync: false,
+			dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
+			rcfg: replicationConfig{
+				remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
+					Arn:             "arn1",
+					ResetID:         "xyz",
+					ResetBeforeDate: start,
+				}}},
+			},
 		},
-		expectedSync: false,
-		dsc:          ReplicateDecision{targetsMap: map[string]replicateTargetDecision{"arn1": newReplicateTargetDecision("arn1", true, false)}},
-		rcfg: replicationConfig{remotes: &madmin.BucketTargets{Targets: []madmin.BucketTarget{{
-			Arn:             "arn1",
-			ResetID:         "xyz",
-			ResetBeforeDate: start,
-		}}},
-		},
-	},
-}
+	}
+)
 
 func TestReplicationResyncwrapper(t *testing.T) {
 	for i, test := range replicationConfigTests2 {

cmd/bucket-stats.go

@@ -38,7 +38,7 @@ func (rl ReplicationLatency) merge(other ReplicationLatency) (newReplLatency Rep
 // Get upload latency of each object size range
 func (rl ReplicationLatency) getUploadLatency() (ret map[string]uint64) {
 	ret = make(map[string]uint64)
-	avg := rl.UploadHistogram.GetAvg()
+	avg := rl.UploadHistogram.GetAvgData()
 	for k, v := range avg {
 		// Convert nanoseconds to milliseconds
 		ret[sizeTagToString(k)] = v.avg() / uint64(time.Millisecond)

cmd/bucket-targets.go

@@ -332,8 +332,10 @@ func (sys *BucketTargetSys) set(bucket BucketInfo, meta BucketMetadata) {
 }
 
 // getRemoteTargetInstanceTransport contains a singleton roundtripper.
-var getRemoteTargetInstanceTransport http.RoundTripper
-var getRemoteTargetInstanceTransportOnce sync.Once
+var (
+	getRemoteTargetInstanceTransport     http.RoundTripper
+	getRemoteTargetInstanceTransportOnce sync.Once
+)
 
 // Returns a minio-go Client configured to access remote host described in replication target config.
 func (sys *BucketTargetSys) getRemoteTargetClient(tcfg *madmin.BucketTarget) (*TargetClient, error) {
@@ -414,7 +416,7 @@ func parseBucketTargetConfig(bucket string, cdata, cmetadata []byte) (*madmin.Bu
 		return nil, nil
 	}
 	data = cdata
-	var json = jsoniter.ConfigCompatibleWithStandardLibrary
+	json := jsoniter.ConfigCompatibleWithStandardLibrary
 	if len(cmetadata) != 0 {
 		if err := json.Unmarshal(cmetadata, &meta); err != nil {
 			return nil, err

cmd/bucket-versioning-handler.go

@@ -144,5 +144,4 @@ func (api objectAPIHandlers) GetBucketVersioningHandler(w http.ResponseWriter, r
 
 	// Write bucket versioning configuration to client
 	writeSuccessResponseXML(w, configData)
-
 }

cmd/common-main.go

@@ -19,6 +19,7 @@ package cmd
 
 import (
 	"bufio"
+	"bytes"
 	"context"
 	"crypto/tls"
 	"crypto/x509"
@@ -39,6 +40,7 @@ import (
 	"syscall"
 	"time"
 
+	"github.com/dustin/go-humanize"
 	fcolor "github.com/fatih/color"
 	"github.com/go-openapi/loads"
 	"github.com/inconshreveable/mousetrap"
@@ -48,6 +50,7 @@ import (
 	"github.com/minio/console/restapi"
 	"github.com/minio/console/restapi/operations"
 	"github.com/minio/kes"
+	"github.com/minio/madmin-go"
 	"github.com/minio/minio-go/v7"
 	"github.com/minio/minio-go/v7/pkg/credentials"
 	"github.com/minio/minio-go/v7/pkg/set"
@@ -67,7 +70,11 @@ import (
 
 // serverDebugLog will enable debug printing
 var serverDebugLog = env.Get("_MINIO_SERVER_DEBUG", config.EnableOff) == config.EnableOn
-var defaultAWSCredProvider []credentials.Provider
+
+var (
+	shardDiskTimeDelta     time.Duration
+	defaultAWSCredProvider []credentials.Provider
+)
 
 func init() {
 	if runtime.GOOS == "windows" {
@@ -127,6 +134,8 @@ func init() {
 	console.SetColor("Debug", fcolor.New())
 
 	gob.Register(StorageErr(""))
+	gob.Register(madmin.TimeInfo{})
+	gob.Register(map[string]interface{}{})
 
 	defaultAWSCredProvider = []credentials.Provider{
 		&credentials.IAM{
@@ -136,6 +145,12 @@ func init() {
 		},
 	}
 
+	var err error
+	shardDiskTimeDelta, err = time.ParseDuration(env.Get("_MINIO_SHARD_DISKTIME_DELTA", "1m"))
+	if err != nil {
+		shardDiskTimeDelta = 1 * time.Minute
+	}
+
 	// All minio-go API operations shall be performed only once,
 	// another way to look at this is we are turning off retries.
 	minio.MaxRetry = 1
@@ -354,7 +369,6 @@ func newConfigDirFromCtx(ctx *cli.Context, option string, getDefaultDir func() s
 }
 
 func handleCommonCmdArgs(ctx *cli.Context) {
-
 	// Get "json" flag from command line argument and
 	// enable json and quite modes if json flag is turned on.
 	globalCLIContext.JSON = ctx.IsSet("json") || ctx.GlobalIsSet("json")
@@ -449,7 +463,8 @@ func (e envKV) String() string {
 }
 
 func parsEnvEntry(envEntry string) (envKV, error) {
-	if strings.TrimSpace(envEntry) == "" {
+	envEntry = strings.TrimSpace(envEntry)
+	if envEntry == "" {
 		// Skip all empty lines
 		return envKV{
 			Skip: true,
@@ -522,7 +537,7 @@ func readFromSecret(sp string) (string, error) {
 		}
 		return "", err
 	}
-	return string(credBuf), nil
+	return string(bytes.TrimSpace(credBuf)), nil
 }
 
 func loadEnvVarsFromFiles() {
@@ -638,6 +653,16 @@ func handleCommonEnvVars() {
 		logger.Fatal(config.ErrInvalidFSOSyncValue(err), "Invalid MINIO_FS_OSYNC value in environment variable")
 	}
 
+	if rootDiskSize := env.Get(config.EnvRootDiskThresholdSize, ""); rootDiskSize != "" {
+		size, err := humanize.ParseBytes(rootDiskSize)
+		if err != nil {
+			logger.Fatal(err, fmt.Sprintf("Invalid %s value in environment variable", config.EnvRootDiskThresholdSize))
+		}
+		globalRootDiskThreshold = size
+	}
+
+	globalIsCICD = env.Get("MINIO_CI_CD", "") != ""
+
 	domains := env.Get(config.EnvDomain, "")
 	if len(domains) != 0 {
 		for _, domainName := range strings.Split(domains, config.ValueSeparator) {
@@ -660,11 +685,11 @@ func handleCommonEnvVars() {
 	publicIPs := env.Get(config.EnvPublicIPs, "")
 	if len(publicIPs) != 0 {
 		minioEndpoints := strings.Split(publicIPs, config.ValueSeparator)
-		var domainIPs = set.NewStringSet()
+		domainIPs := set.NewStringSet()
 		for _, endpoint := range minioEndpoints {
 			if net.ParseIP(endpoint) == nil {
 				// Checking if the IP is a DNS entry.
-				addrs, err := net.LookupHost(endpoint)
+				addrs, err := globalDNSCache.LookupHost(GlobalContext, endpoint)
 				if err != nil {
 					logger.FatalIf(err, "Unable to initialize MinIO server with [%s] invalid entry found in MINIO_PUBLIC_IPS", endpoint)
 				}
@@ -777,7 +802,7 @@ func handleCommonEnvVars() {
 			logger.Fatal(err, fmt.Sprintf("Unable to load X.509 root CAs for KES from %q", env.Get(config.EnvKESServerCA, globalCertsCADir.Get())))
 		}
 
-		var defaultKeyID = env.Get(config.EnvKESKeyName, "")
+		defaultKeyID := env.Get(config.EnvKESKeyName, "")
 		KMS, err := kms.NewWithConfig(kms.Config{
 			Endpoints:    endpoints,
 			DefaultKeyID: defaultKeyID,
@@ -897,3 +922,32 @@ func contextCanceled(ctx context.Context) bool {
 		return false
 	}
 }
+
+// bgContext returns a context that can be used for async operations.
+// Cancellation/timeouts are removed, so parent cancellations/timeout will
+// not propagate from parent.
+// Context values are preserved.
+// This can be used for goroutines that live beyond the parent context.
+func bgContext(parent context.Context) context.Context {
+	return bgCtx{parent: parent}
+}
+
+type bgCtx struct {
+	parent context.Context
+}
+
+func (a bgCtx) Done() <-chan struct{} {
+	return nil
+}
+
+func (a bgCtx) Err() error {
+	return nil
+}
+
+func (a bgCtx) Deadline() (deadline time.Time, ok bool) {
+	return time.Time{}, false
+}
+
+func (a bgCtx) Value(key interface{}) interface{} {
+	return a.parent.Value(key)
+}

cmd/common-main_test.go

@@ -24,13 +24,57 @@ import (
 	"testing"
 )
 
+func Test_readFromSecret(t *testing.T) {
+	testCases := []struct {
+		content       string
+		expectedErr   bool
+		expectedValue string
+	}{
+		{
+			"value\n",
+			false,
+			"value",
+		},
+		{
+			" \t\n Hello, Gophers \n\t\r\n",
+			false,
+			"Hello, Gophers",
+		},
+	}
+
+	for _, testCase := range testCases {
+		testCase := testCase
+		t.Run("", func(t *testing.T) {
+			tmpfile, err := ioutil.TempFile("", "testfile")
+			if err != nil {
+				t.Error(err)
+			}
+			tmpfile.WriteString(testCase.content)
+			tmpfile.Sync()
+			tmpfile.Close()
+
+			value, err := readFromSecret(tmpfile.Name())
+			if err != nil && !testCase.expectedErr {
+				t.Error(err)
+			}
+			if err == nil && testCase.expectedErr {
+				t.Error(errors.New("expected error, found success"))
+			}
+			if value != testCase.expectedValue {
+				t.Errorf("Expected %s, got %s", testCase.expectedValue, value)
+			}
+		})
+	}
+}
+
 func Test_minioEnvironFromFile(t *testing.T) {
 	testCases := []struct {
 		content      string
 		expectedErr  bool
 		expectedEkvs []envKV
 	}{
-		{`
+		{
+			`
 export MINIO_ROOT_USER=minio
 export MINIO_ROOT_PASSWORD=minio123`,
 			false,
@@ -46,7 +90,8 @@ export MINIO_ROOT_PASSWORD=minio123`,
 			},
 		},
 		// Value with double quotes
-		{`export MINIO_ROOT_USER="minio"`,
+		{
+			`export MINIO_ROOT_USER="minio"`,
 			false,
 			[]envKV{
 				{
@@ -56,7 +101,8 @@ export MINIO_ROOT_PASSWORD=minio123`,
 			},
 		},
 		// Value with single quotes
-		{`export MINIO_ROOT_USER='minio'`,
+		{
+			`export MINIO_ROOT_USER='minio'`,
 			false,
 			[]envKV{
 				{
@@ -65,7 +111,8 @@ export MINIO_ROOT_PASSWORD=minio123`,
 				},
 			},
 		},
-		{`
+		{
+			`
 MINIO_ROOT_USER=minio
 MINIO_ROOT_PASSWORD=minio123`,
 			false,
@@ -80,7 +127,8 @@ MINIO_ROOT_PASSWORD=minio123`,
 				},
 			},
 		},
-		{`
+		{
+			`
 export MINIO_ROOT_USERminio
 export MINIO_ROOT_PASSWORD=minio123`,
 			true,

cmd/config-common.go

@@ -29,8 +29,8 @@ import (
 
 var errConfigNotFound = errors.New("config file not found")
 
-func readConfigWithMetadata(ctx context.Context, objAPI ObjectLayer, configFile string) ([]byte, ObjectInfo, error) {
-	r, err := objAPI.GetObjectNInfo(ctx, minioMetaBucket, configFile, nil, http.Header{}, readLock, ObjectOptions{})
+func readConfigWithMetadata(ctx context.Context, store objectIO, configFile string) ([]byte, ObjectInfo, error) {
+	r, err := store.GetObjectNInfo(ctx, minioMetaBucket, configFile, nil, http.Header{}, readLock, ObjectOptions{})
 	if err != nil {
 		// Treat object not found as config not found.
 		if isErrObjectNotFound(err) {
@@ -51,8 +51,8 @@ func readConfigWithMetadata(ctx context.Context, objAPI ObjectLayer, configFile
 	return buf, r.ObjInfo, nil
 }
 
-func readConfig(ctx context.Context, objAPI ObjectLayer, configFile string) ([]byte, error) {
-	buf, _, err := readConfigWithMetadata(ctx, objAPI, configFile)
+func readConfig(ctx context.Context, store objectIO, configFile string) ([]byte, error) {
+	buf, _, err := readConfigWithMetadata(ctx, store, configFile)
 	return buf, err
 }
 
@@ -70,13 +70,13 @@ func deleteConfig(ctx context.Context, objAPI objectDeleter, configFile string)
 	return err
 }
 
-func saveConfig(ctx context.Context, objAPI ObjectLayer, configFile string, data []byte) error {
+func saveConfig(ctx context.Context, store objectIO, configFile string, data []byte) error {
 	hashReader, err := hash.NewReader(bytes.NewReader(data), int64(len(data)), "", getSHA256Hash(data), int64(len(data)))
 	if err != nil {
 		return err
 	}
 
-	_, err = objAPI.PutObject(ctx, minioMetaBucket, configFile, NewPutObjReader(hashReader), ObjectOptions{MaxParity: true})
+	_, err = store.PutObject(ctx, minioMetaBucket, configFile, NewPutObjReader(hashReader), ObjectOptions{MaxParity: true})
 	return err
 }
 

cmd/config-current.go

@@ -50,7 +50,7 @@ import (
 )
 
 func initHelp() {
-	var kvs = map[string]config.KVS{
+	kvs := map[string]config.KVS{
 		config.EtcdSubSys:           etcd.DefaultKVS,
 		config.CacheSubSys:          cache.DefaultKVS,
 		config.CompressionSubSys:    compress.DefaultKVS,
@@ -78,7 +78,7 @@ func initHelp() {
 	config.RegisterDefaultKVS(kvs)
 
 	// Captures help for each sub-system
-	var helpSubSys = config.HelpKVS{
+	helpSubSys := config.HelpKVS{
 		config.HelpKV{
 			Key:         config.SiteSubSys,
 			Description: "label the server and its location",
@@ -205,7 +205,7 @@ func initHelp() {
 		}
 	}
 
-	var helpMap = map[string]config.HelpKVS{
+	helpMap := map[string]config.HelpKVS{
 		"":                          helpSubSys, // Help for all sub-systems.
 		config.SiteSubSys:           config.SiteHelp,
 		config.RegionSubSys:         config.RegionHelp,

cmd/config-dir.go

@@ -84,7 +84,7 @@ func (dir *ConfigDir) Get() string {
 
 // Attempts to create all directories, ignores any permission denied errors.
 func mkdirAllIgnorePerm(path string) error {
-	err := os.MkdirAll(path, 0700)
+	err := os.MkdirAll(path, 0o700)
 	if err != nil {
 		// It is possible in kubernetes like deployments this directory
 		// is already mounted and is not writable, ignore any write errors.

cmd/config-migrate.go

@@ -2445,12 +2445,12 @@ func migrateConfigToMinioSys(objAPI ObjectLayer) (err error) {
 		return err
 	} // if errConfigNotFound proceed to migrate..
 
-	var configFiles = []string{
+	configFiles := []string{
 		getConfigFile(),
 		getConfigFile() + ".deprecated",
 		configFile,
 	}
-	var config = &serverConfigV27{}
+	config := &serverConfigV27{}
 	for _, cfgFile := range configFiles {
 		if _, err = Load(cfgFile, config); err != nil {
 			if !osIsNotExist(err) && !osIsPermission(err) {

cmd/config-migrate_test.go

@@ -51,7 +51,7 @@ func TestServerConfigMigrateV1(t *testing.T) {
 	// Create a V1 config json file and store it
 	configJSON := "{ \"version\":\"1\", \"accessKeyId\":\"abcde\", \"secretAccessKey\":\"abcdefgh\"}"
 	configPath := rootPath + "/fsUsers.json"
-	if err := ioutil.WriteFile(configPath, []byte(configJSON), 0644); err != nil {
+	if err := ioutil.WriteFile(configPath, []byte(configJSON), 0o644); err != nil {
 		t.Fatal("Unexpected error: ", err)
 	}
 
@@ -181,7 +181,7 @@ func TestServerConfigMigrateV2toV33(t *testing.T) {
 	configPath := rootPath + SlashSeparator + minioConfigFile
 
 	// Create a corrupted config file
-	if err := ioutil.WriteFile(configPath, []byte("{ \"version\":\"2\","), 0644); err != nil {
+	if err := ioutil.WriteFile(configPath, []byte("{ \"version\":\"2\","), 0o644); err != nil {
 		t.Fatal("Unexpected error: ", err)
 	}
 	// Fire a migrateConfig()
@@ -194,7 +194,7 @@ func TestServerConfigMigrateV2toV33(t *testing.T) {
 
 	// Create a V2 config json file and store it
 	configJSON := "{ \"version\":\"2\", \"credentials\": {\"accessKeyId\":\"" + accessKey + "\", \"secretAccessKey\":\"" + secretKey + "\", \"region\":\"us-east-1\"}, \"mongoLogger\":{\"addr\":\"127.0.0.1:3543\", \"db\":\"foodb\", \"collection\":\"foo\"}, \"syslogLogger\":{\"network\":\"127.0.0.1:543\", \"addr\":\"addr\"}, \"fileLogger\":{\"filename\":\"log.out\"}}"
-	if err := ioutil.WriteFile(configPath, []byte(configJSON), 0644); err != nil {
+	if err := ioutil.WriteFile(configPath, []byte(configJSON), 0o644); err != nil {
 		t.Fatal("Unexpected error: ", err)
 	}
 
@@ -244,7 +244,7 @@ func TestServerConfigMigrateFaultyConfig(t *testing.T) {
 	configPath := rootPath + SlashSeparator + minioConfigFile
 
 	// Create a corrupted config file
-	if err := ioutil.WriteFile(configPath, []byte("{ \"version\":\"2\", \"test\":"), 0644); err != nil {
+	if err := ioutil.WriteFile(configPath, []byte("{ \"version\":\"2\", \"test\":"), 0o644); err != nil {
 		t.Fatal("Unexpected error: ", err)
 	}
 
@@ -343,7 +343,7 @@ func TestServerConfigMigrateCorruptedConfig(t *testing.T) {
 	for i := 3; i <= 17; i++ {
 		// Create a corrupted config file
 		if err = ioutil.WriteFile(configPath, []byte(fmt.Sprintf("{ \"version\":\"%d\", \"credential\": { \"accessKey\": 1 } }", i)),
-			0644); err != nil {
+			0o644); err != nil {
 			t.Fatal("Unexpected error: ", err)
 		}
 
@@ -354,7 +354,7 @@ func TestServerConfigMigrateCorruptedConfig(t *testing.T) {
 	}
 
 	// Create a corrupted config file for version '2'.
-	if err = ioutil.WriteFile(configPath, []byte("{ \"version\":\"2\", \"credentials\": { \"accessKeyId\": 1 } }"), 0644); err != nil {
+	if err = ioutil.WriteFile(configPath, []byte("{ \"version\":\"2\", \"credentials\": { \"accessKeyId\": 1 } }"), 0o644); err != nil {
 		t.Fatal("Unexpected error: ", err)
 	}
 

cmd/config.go

@@ -35,8 +35,8 @@ import (
 
 const (
 	minioConfigPrefix = "config"
-
-	kvPrefix = ".kv"
+	minioConfigBucket = minioMetaBucket + SlashSeparator + minioConfigPrefix
+	kvPrefix          = ".kv"
 
 	// Captures all the previous SetKV operations and allows rollback.
 	minioConfigHistoryPrefix = minioConfigPrefix + "/history"
@@ -46,8 +46,8 @@ const (
 )
 
 func listServerConfigHistory(ctx context.Context, objAPI ObjectLayer, withData bool, count int) (
-	[]madmin.ConfigHistoryEntry, error) {
-
+	[]madmin.ConfigHistoryEntry, error,
+) {
 	var configHistory []madmin.ConfigHistoryEntry
 
 	// List all kvs
@@ -140,7 +140,7 @@ func saveServerConfig(ctx context.Context, objAPI ObjectLayer, cfg interface{})
 		return err
 	}
 
-	var configFile = path.Join(minioConfigPrefix, minioConfigFile)
+	configFile := path.Join(minioConfigPrefix, minioConfigFile)
 	if GlobalKMS != nil {
 		data, err = config.EncryptBytes(GlobalKMS, data, kms.Context{
 			minioMetaBucket: path.Join(minioMetaBucket, configFile),
@@ -153,7 +153,7 @@ func saveServerConfig(ctx context.Context, objAPI ObjectLayer, cfg interface{})
 }
 
 func readServerConfig(ctx context.Context, objAPI ObjectLayer) (config.Config, error) {
-	var srvCfg = config.New()
+	srvCfg := config.New()
 	configFile := path.Join(minioConfigPrefix, minioConfigFile)
 	data, err := readConfig(ctx, objAPI, configFile)
 	if err != nil {
@@ -174,7 +174,7 @@ func readServerConfig(ctx context.Context, objAPI ObjectLayer) (config.Config, e
 		}
 	}
 
-	var json = jsoniter.ConfigCompatibleWithStandardLibrary
+	json := jsoniter.ConfigCompatibleWithStandardLibrary
 	if err = json.Unmarshal(data, &srvCfg); err != nil {
 		return nil, err
 	}

cmd/data-scanner.go

@@ -22,10 +22,10 @@ import (
 	"context"
 	"encoding/binary"
 	"errors"
+	"fmt"
 	"io/fs"
 	"math"
 	"math/rand"
-	"net/http"
 	"os"
 	"path"
 	"strings"
@@ -34,13 +34,13 @@ import (
 	"time"
 
 	"github.com/bits-and-blooms/bloom/v3"
+	"github.com/dustin/go-humanize"
 	"github.com/minio/madmin-go"
 	"github.com/minio/minio/internal/bucket/lifecycle"
 	"github.com/minio/minio/internal/bucket/replication"
 	"github.com/minio/minio/internal/color"
 	"github.com/minio/minio/internal/config/heal"
 	"github.com/minio/minio/internal/event"
-	"github.com/minio/minio/internal/hash"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/pkg/console"
 )
@@ -72,14 +72,16 @@ var (
 // initDataScanner will start the scanner in the background.
 func initDataScanner(ctx context.Context, objAPI ObjectLayer) {
 	go func() {
+		r := rand.New(rand.NewSource(time.Now().UnixNano()))
 		// Run the data scanner in a loop
 		for {
-			select {
-			case <-ctx.Done():
-				return
-			default:
-				runDataScanner(ctx, objAPI)
+			runDataScanner(ctx, objAPI)
+			duration := time.Duration(r.Float64() * float64(scannerCycle.Get()))
+			if duration < time.Second {
+				// Make sure to sleep atleast a second to avoid high CPU ticks.
+				duration = time.Second
 			}
+			time.Sleep(duration)
 		}
 	}()
 }
@@ -106,36 +108,26 @@ func (s *safeDuration) Get() time.Duration {
 // There should only ever be one scanner running per cluster.
 func runDataScanner(pctx context.Context, objAPI ObjectLayer) {
 	// Make sure only 1 scanner is running on the cluster.
-	locker := objAPI.NewNSLock(minioMetaBucket, "runDataScanner.lock")
-	var ctx context.Context
-	r := rand.New(rand.NewSource(time.Now().UnixNano()))
-	for {
-		lkctx, err := locker.GetLock(pctx, dataScannerLeaderLockTimeout)
-		if err != nil {
-			time.Sleep(time.Duration(r.Float64() * float64(scannerCycle.Get())))
-			continue
+	locker := objAPI.NewNSLock(minioMetaBucket, "scanner/runDataScanner.lock")
+	lkctx, err := locker.GetLock(pctx, dataScannerLeaderLockTimeout)
+	if err != nil {
+		if intDataUpdateTracker.debug {
+			logger.LogIf(pctx, err)
 		}
-		ctx = lkctx.Context()
-		defer lkctx.Cancel()
-		break
-		// No unlock for "leader" lock.
+		return
 	}
+	ctx := lkctx.Context()
+	defer lkctx.Cancel()
+	// No unlock for "leader" lock.
 
 	// Load current bloom cycle
 	nextBloomCycle := intDataUpdateTracker.current() + 1
 
-	br, err := objAPI.GetObjectNInfo(ctx, dataUsageBucket, dataUsageBloomName, nil, http.Header{}, readLock, ObjectOptions{})
-	if err != nil {
-		if !isErrObjectNotFound(err) && !isErrBucketNotFound(err) {
+	buf, _ := readConfig(ctx, objAPI, dataUsageBloomNamePath)
+	if len(buf) >= 8 {
+		if err = binary.Read(bytes.NewReader(buf), binary.LittleEndian, &nextBloomCycle); err != nil {
 			logger.LogIf(ctx, err)
 		}
-	} else {
-		if br.ObjInfo.Size == 8 {
-			if err = binary.Read(br, binary.LittleEndian, &nextBloomCycle); err != nil {
-				logger.LogIf(ctx, err)
-			}
-		}
-		br.Close()
 	}
 
 	scannerTimer := time.NewTimer(scannerCycle.Get())
@@ -165,14 +157,7 @@ func runDataScanner(pctx context.Context, objAPI ObjectLayer) {
 				nextBloomCycle++
 				var tmp [8]byte
 				binary.LittleEndian.PutUint64(tmp[:], nextBloomCycle)
-				r, err := hash.NewReader(bytes.NewReader(tmp[:]), int64(len(tmp)), "", "", int64(len(tmp)))
-				if err != nil {
-					logger.LogIf(ctx, err)
-					continue
-				}
-
-				_, err = objAPI.PutObject(ctx, dataUsageBucket, dataUsageBloomName, NewPutObjReader(r), ObjectOptions{})
-				if !isErrBucketNotFound(err) {
+				if err = saveConfig(ctx, objAPI, dataUsageBloomNamePath, tmp[:]); err != nil {
 					logger.LogIf(ctx, err)
 				}
 			}
@@ -198,7 +183,8 @@ type folderScanner struct {
 	healFolderInclude     uint32 // Include a clean folder one in n cycles.
 	healObjectSelect      uint32 // Do a heal check on an object once every n cycles. Must divide into healFolderInclude
 
-	disks []StorageAPI
+	disks       []StorageAPI
+	disksQuorum int
 
 	// If set updates will be sent regularly to this channel.
 	// Will not be closed when returned.
@@ -264,7 +250,7 @@ var globalScannerStats scannerStats
 // The returned cache will always be valid, but may not be updated from the existing.
 // Before each operation sleepDuration is called which can be used to temporarily halt the scanner.
 // If the supplied context is canceled the function will return at the first chance.
-func scanDataFolder(ctx context.Context, basePath string, cache dataUsageCache, getSize getSizeFn) (dataUsageCache, error) {
+func scanDataFolder(ctx context.Context, poolIdx, setIdx int, basePath string, cache dataUsageCache, getSize getSizeFn) (dataUsageCache, error) {
 	t := UTCNow()
 
 	logPrefix := color.Green("data-usage: ")
@@ -277,7 +263,6 @@ func scanDataFolder(ctx context.Context, basePath string, cache dataUsageCache,
 		defer func() {
 			console.Debugf(logPrefix+" Scanner time: %v %s\n", time.Since(t), logSuffix)
 		}()
-
 	}
 
 	switch cache.Info.Name {
@@ -298,13 +283,15 @@ func scanDataFolder(ctx context.Context, basePath string, cache dataUsageCache,
 	}
 
 	// Add disks for set healing.
-	if len(cache.Disks) > 0 {
+	if poolIdx >= 0 && setIdx >= 0 {
 		objAPI, ok := newObjectLayerFn().(*erasureServerPools)
 		if ok {
-			s.disks = objAPI.GetDisksID(cache.Disks...)
-			if len(s.disks) != len(cache.Disks) {
-				console.Debugf(logPrefix+"Missing disks, want %d, found %d. Cannot heal. %s\n", len(cache.Disks), len(s.disks), logSuffix)
-				s.disks = s.disks[:0]
+			if poolIdx < len(objAPI.serverPools) && setIdx < len(objAPI.serverPools[poolIdx].sets) {
+				// Pass the disks belonging to the set.
+				s.disks = objAPI.serverPools[poolIdx].sets[setIdx].getDisks()
+				s.disksQuorum = len(s.disks) / 2
+			} else {
+				logger.LogIf(ctx, fmt.Errorf("Matching pool %s, set %s not found", humanize.Ordinal(poolIdx+1), humanize.Ordinal(setIdx+1)))
 			}
 		}
 	}
@@ -641,7 +628,7 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 		}
 
 		objAPI, ok := newObjectLayerFn().(*erasureServerPools)
-		if !ok || len(f.disks) == 0 {
+		if !ok || len(f.disks) == 0 || f.disksQuorum == 0 {
 			break
 		}
 
@@ -663,8 +650,8 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 		// This means that the next run will not look for it.
 		// How to resolve results.
 		resolver := metadataResolutionParams{
-			dirQuorum: getReadQuorum(len(f.disks)),
-			objQuorum: getReadQuorum(len(f.disks)),
+			dirQuorum: f.disksQuorum,
+			objQuorum: f.disksQuorum,
 			bucket:    "",
 			strict:    false,
 		}
@@ -694,8 +681,7 @@ func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, int
 				path:           prefix,
 				recursive:      true,
 				reportNotFound: true,
-				minDisks:       len(f.disks), // We want full consistency.
-				// Weird, maybe transient error.
+				minDisks:       f.disksQuorum,
 				agreed: func(entry metaCacheEntry) {
 					if f.dataUsageScannerDebug {
 						console.Debugf(healObjectsPrefix+" got agreement: %v\n", entry.name)
@@ -875,8 +861,10 @@ func (i *scannerItem) transformMetaDir() {
 	i.objectName = split[len(split)-1]
 }
 
-var applyActionsLogPrefix = color.Green("applyActions:")
-var applyVersionActionsLogPrefix = color.Green("applyVersionActions:")
+var (
+	applyActionsLogPrefix        = color.Green("applyActions:")
+	applyVersionActionsLogPrefix = color.Green("applyVersionActions:")
+)
 
 func (i *scannerItem) applyHealing(ctx context.Context, o ObjectLayer, oi ObjectInfo) (size int64) {
 	if i.debug {
@@ -979,7 +967,6 @@ func (i *scannerItem) applyTierObjSweep(ctx context.Context, o ObjectLayer, oi O
 	if ignoreNotFoundErr(err) != nil {
 		logger.LogIf(ctx, err)
 	}
-
 }
 
 // applyNewerNoncurrentVersionLimit removes noncurrent versions older than the most recent NewerNoncurrentVersions configured.
@@ -1026,8 +1013,10 @@ func (i *scannerItem) applyNewerNoncurrentVersionLimit(ctx context.Context, _ Ob
 		}
 
 		toDel = append(toDel, ObjectToDelete{
-			ObjectName: fi.Name,
-			VersionID:  fi.VersionID,
+			ObjectV: ObjectV{
+				ObjectName: fi.Name,
+				VersionID:  fi.VersionID,
+			},
 		})
 	}
 
@@ -1100,7 +1089,6 @@ func applyTransitionRule(obj ObjectInfo) bool {
 	}
 	globalTransitionState.queueTransitionTask(obj)
 	return true
-
 }
 
 func applyExpiryOnTransitionedObject(ctx context.Context, objLayer ObjectLayer, obj ObjectInfo, restoredObject bool) bool {

cmd/data-update-tracker.go

@@ -204,14 +204,14 @@ func (d *dataUpdateTracker) latestWithDir(dir string) uint64 {
 // All of these will exit when the context is canceled.
 func (d *dataUpdateTracker) start(ctx context.Context, drives ...string) {
 	if len(drives) == 0 {
-		logger.LogIf(ctx, errors.New("dataUpdateTracker.start: No drives specified"))
+		logger.LogIf(ctx, errors.New("dataUpdateTracker.start: No local drives specified"))
 		return
 	}
 	d.load(ctx, drives...)
 	go d.startCollector(ctx)
 	// startSaver will unlock.
 	d.mu.Lock()
-	go d.startSaver(ctx, dataUpdateTrackerSaveInterval, drives)
+	go d.startSaver(ctx, dataUpdateTrackerSaveInterval, drives...)
 }
 
 // load will attempt to load data tracking information from the supplied drives.
@@ -221,7 +221,7 @@ func (d *dataUpdateTracker) start(ctx context.Context, drives ...string) {
 // If object is shared the caller should lock it.
 func (d *dataUpdateTracker) load(ctx context.Context, drives ...string) {
 	if len(drives) == 0 {
-		logger.LogIf(ctx, errors.New("dataUpdateTracker.load: No drives specified"))
+		logger.LogIf(ctx, errors.New("dataUpdateTracker.load: No local drives specified"))
 		return
 	}
 	for _, drive := range drives {
@@ -246,7 +246,11 @@ func (d *dataUpdateTracker) load(ctx context.Context, drives ...string) {
 // startSaver will start a saver that will write d to all supplied drives at specific intervals.
 // 'd' must be write locked when started and will be unlocked.
 // The saver will save and exit when supplied context is closed.
-func (d *dataUpdateTracker) startSaver(ctx context.Context, interval time.Duration, drives []string) {
+func (d *dataUpdateTracker) startSaver(ctx context.Context, interval time.Duration, drives ...string) {
+	if len(drives) == 0 {
+		return
+	}
+
 	saveNow := d.save
 	exited := make(chan struct{})
 	d.saveExited = exited

cmd/data-update-tracker_test.go

@@ -105,7 +105,7 @@ func TestDataUpdateTracker(t *testing.T) {
 	defer cancel()
 	dut.start(ctx, tmpDir)
 
-	var tests = []struct {
+	tests := []struct {
 		in    string
 		check []string // if not empty, check against these instead.
 		exist bool

cmd/data-usage-cache.go

@@ -225,7 +225,6 @@ type dataUsageEntryV6 struct {
 type dataUsageCache struct {
 	Info  dataUsageCacheInfo
 	Cache map[string]dataUsageEntry
-	Disks []string
 }
 
 //msgp:encode ignore dataUsageCacheV2 dataUsageCacheV3 dataUsageCacheV4 dataUsageCacheV5 dataUsageCacheV6
@@ -234,35 +233,30 @@ type dataUsageCache struct {
 // dataUsageCacheV2 contains a cache of data usage entries version 2.
 type dataUsageCacheV2 struct {
 	Info  dataUsageCacheInfo
-	Disks []string
 	Cache map[string]dataUsageEntryV2
 }
 
 // dataUsageCacheV3 contains a cache of data usage entries version 3.
 type dataUsageCacheV3 struct {
 	Info  dataUsageCacheInfo
-	Disks []string
 	Cache map[string]dataUsageEntryV3
 }
 
 // dataUsageCacheV4 contains a cache of data usage entries version 4.
 type dataUsageCacheV4 struct {
 	Info  dataUsageCacheInfo
-	Disks []string
 	Cache map[string]dataUsageEntryV4
 }
 
 // dataUsageCacheV5 contains a cache of data usage entries version 5.
 type dataUsageCacheV5 struct {
 	Info  dataUsageCacheInfo
-	Disks []string
 	Cache map[string]dataUsageEntryV5
 }
 
 // dataUsageCacheV6 contains a cache of data usage entries version 6.
 type dataUsageCacheV6 struct {
 	Info  dataUsageCacheInfo
-	Disks []string
 	Cache map[string]dataUsageEntryV6
 }
 
@@ -624,7 +618,7 @@ func (d *dataUsageCache) reduceChildrenOf(path dataUsageHash, limit int, compact
 	// Appears to be printed with _MINIO_SERVER_DEBUG=off
 	// console.Debugf(" %d children found, compacting %v\n", total, path)
 
-	var leaves = make([]struct {
+	leaves := make([]struct {
 		objects uint64
 		path    dataUsageHash
 	}, total)
@@ -774,7 +768,7 @@ func (d *dataUsageCache) tiersUsageInfo(buckets []BucketInfo) *allTierStats {
 // bucketsUsageInfo returns the buckets usage info as a map, with
 // key as bucket name
 func (d *dataUsageCache) bucketsUsageInfo(buckets []BucketInfo) map[string]BucketUsageInfo {
-	var dst = make(map[string]BucketUsageInfo, len(buckets))
+	dst := make(map[string]BucketUsageInfo, len(buckets))
 	for _, bucket := range buckets {
 		e := d.find(bucket.Name)
 		if e == nil {
@@ -797,7 +791,6 @@ func (d *dataUsageCache) bucketsUsageInfo(buckets []BucketInfo) map[string]Bucke
 					ReplicationPendingCount: stat.PendingCount,
 					ReplicationFailedCount:  stat.FailedCount,
 				}
-
 			}
 		}
 		dst[bucket.Name] = bui
@@ -922,6 +915,7 @@ func (d *dataUsageCache) load(ctx context.Context, store objectIO, name string)
 	// Abandon if more than 5 minutes, so we don't hold up scanner.
 	ctx, cancel := context.WithTimeout(ctx, 5*time.Minute)
 	defer cancel()
+
 	r, err := store.GetObjectNInfo(ctx, dataUsageBucket, name, nil, http.Header{}, readLock, ObjectOptions{})
 	if err != nil {
 		switch err.(type) {
@@ -1037,7 +1031,6 @@ func (d *dataUsageCache) deserialize(r io.Reader) error {
 			return err
 		}
 		d.Info = dold.Info
-		d.Disks = dold.Disks
 		d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
 		for k, v := range dold.Cache {
 			d.Cache[k] = dataUsageEntry{
@@ -1061,7 +1054,6 @@ func (d *dataUsageCache) deserialize(r io.Reader) error {
 			return err
 		}
 		d.Info = dold.Info
-		d.Disks = dold.Disks
 		d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
 		for k, v := range dold.Cache {
 			due := dataUsageEntry{
@@ -1101,7 +1093,6 @@ func (d *dataUsageCache) deserialize(r io.Reader) error {
 			return err
 		}
 		d.Info = dold.Info
-		d.Disks = dold.Disks
 		d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
 		for k, v := range dold.Cache {
 			due := dataUsageEntry{
@@ -1153,7 +1144,6 @@ func (d *dataUsageCache) deserialize(r io.Reader) error {
 			return err
 		}
 		d.Info = dold.Info
-		d.Disks = dold.Disks
 		d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
 		for k, v := range dold.Cache {
 			due := dataUsageEntry{
@@ -1205,7 +1195,6 @@ func (d *dataUsageCache) deserialize(r io.Reader) error {
 			return err
 		}
 		d.Info = dold.Info
-		d.Disks = dold.Disks
 		d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
 		for k, v := range dold.Cache {
 			var replicationStats *replicationAllStats

cmd/data-usage-cache_gen.go

@@ -338,25 +338,6 @@ func (z *dataUsageCache) DecodeMsg(dc *msgp.Reader) (err error) {
 				}
 				z.Cache[za0001] = za0002
 			}
-		case "Disks":
-			var zb0003 uint32
-			zb0003, err = dc.ReadArrayHeader()
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0003) {
-				z.Disks = (z.Disks)[:zb0003]
-			} else {
-				z.Disks = make([]string, zb0003)
-			}
-			for za0003 := range z.Disks {
-				z.Disks[za0003], err = dc.ReadString()
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0003)
-					return
-				}
-			}
 		default:
 			err = dc.Skip()
 			if err != nil {
@@ -370,9 +351,9 @@ func (z *dataUsageCache) DecodeMsg(dc *msgp.Reader) (err error) {
 
 // EncodeMsg implements msgp.Encodable
 func (z *dataUsageCache) EncodeMsg(en *msgp.Writer) (err error) {
-	// map header, size 3
+	// map header, size 2
 	// write "Info"
-	err = en.Append(0x83, 0xa4, 0x49, 0x6e, 0x66, 0x6f)
+	err = en.Append(0x82, 0xa4, 0x49, 0x6e, 0x66, 0x6f)
 	if err != nil {
 		return
 	}
@@ -403,32 +384,15 @@ func (z *dataUsageCache) EncodeMsg(en *msgp.Writer) (err error) {
 			return
 		}
 	}
-	// write "Disks"
-	err = en.Append(0xa5, 0x44, 0x69, 0x73, 0x6b, 0x73)
-	if err != nil {
-		return
-	}
-	err = en.WriteArrayHeader(uint32(len(z.Disks)))
-	if err != nil {
-		err = msgp.WrapError(err, "Disks")
-		return
-	}
-	for za0003 := range z.Disks {
-		err = en.WriteString(z.Disks[za0003])
-		if err != nil {
-			err = msgp.WrapError(err, "Disks", za0003)
-			return
-		}
-	}
 	return
 }
 
 // MarshalMsg implements msgp.Marshaler
 func (z *dataUsageCache) MarshalMsg(b []byte) (o []byte, err error) {
 	o = msgp.Require(b, z.Msgsize())
-	// map header, size 3
+	// map header, size 2
 	// string "Info"
-	o = append(o, 0x83, 0xa4, 0x49, 0x6e, 0x66, 0x6f)
+	o = append(o, 0x82, 0xa4, 0x49, 0x6e, 0x66, 0x6f)
 	o, err = z.Info.MarshalMsg(o)
 	if err != nil {
 		err = msgp.WrapError(err, "Info")
@@ -445,12 +409,6 @@ func (z *dataUsageCache) MarshalMsg(b []byte) (o []byte, err error) {
 			return
 		}
 	}
-	// string "Disks"
-	o = append(o, 0xa5, 0x44, 0x69, 0x73, 0x6b, 0x73)
-	o = msgp.AppendArrayHeader(o, uint32(len(z.Disks)))
-	for za0003 := range z.Disks {
-		o = msgp.AppendString(o, z.Disks[za0003])
-	}
 	return
 }
 
@@ -508,25 +466,6 @@ func (z *dataUsageCache) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				}
 				z.Cache[za0001] = za0002
 			}
-		case "Disks":
-			var zb0003 uint32
-			zb0003, bts, err = msgp.ReadArrayHeaderBytes(bts)
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0003) {
-				z.Disks = (z.Disks)[:zb0003]
-			} else {
-				z.Disks = make([]string, zb0003)
-			}
-			for za0003 := range z.Disks {
-				z.Disks[za0003], bts, err = msgp.ReadStringBytes(bts)
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0003)
-					return
-				}
-			}
 		default:
 			bts, err = msgp.Skip(bts)
 			if err != nil {
@@ -548,10 +487,6 @@ func (z *dataUsageCache) Msgsize() (s int) {
 			s += msgp.StringPrefixSize + len(za0001) + za0002.Msgsize()
 		}
 	}
-	s += 6 + msgp.ArrayHeaderSize
-	for za0003 := range z.Disks {
-		s += msgp.StringPrefixSize + len(z.Disks[za0003])
-	}
 	return
 }
 
@@ -811,54 +746,35 @@ func (z *dataUsageCacheV2) DecodeMsg(dc *msgp.Reader) (err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, err = dc.ReadArrayHeader()
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], err = dc.ReadString()
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, err = dc.ReadMapHeader()
+			var zb0002 uint32
+			zb0002, err = dc.ReadMapHeader()
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV2, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV2, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				zb0003--
-				var za0002 string
-				var za0003 dataUsageEntryV2
-				za0002, err = dc.ReadString()
+			for zb0002 > 0 {
+				zb0002--
+				var za0001 string
+				var za0002 dataUsageEntryV2
+				za0001, err = dc.ReadString()
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				err = za0003.DecodeMsg(dc)
+				err = za0002.DecodeMsg(dc)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			err = dc.Skip()
@@ -895,54 +811,35 @@ func (z *dataUsageCacheV2) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, bts, err = msgp.ReadArrayHeaderBytes(bts)
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], bts, err = msgp.ReadStringBytes(bts)
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, bts, err = msgp.ReadMapHeaderBytes(bts)
+			var zb0002 uint32
+			zb0002, bts, err = msgp.ReadMapHeaderBytes(bts)
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV2, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV2, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				var za0002 string
-				var za0003 dataUsageEntryV2
-				zb0003--
-				za0002, bts, err = msgp.ReadStringBytes(bts)
+			for zb0002 > 0 {
+				var za0001 string
+				var za0002 dataUsageEntryV2
+				zb0002--
+				za0001, bts, err = msgp.ReadStringBytes(bts)
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				bts, err = za0003.UnmarshalMsg(bts)
+				bts, err = za0002.UnmarshalMsg(bts)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			bts, err = msgp.Skip(bts)
@@ -958,15 +855,11 @@ func (z *dataUsageCacheV2) 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 *dataUsageCacheV2) Msgsize() (s int) {
-	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.ArrayHeaderSize
-	for za0001 := range z.Disks {
-		s += msgp.StringPrefixSize + len(z.Disks[za0001])
-	}
-	s += 6 + msgp.MapHeaderSize
+	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.MapHeaderSize
 	if z.Cache != nil {
-		for za0002, za0003 := range z.Cache {
-			_ = za0003
-			s += msgp.StringPrefixSize + len(za0002) + za0003.Msgsize()
+		for za0001, za0002 := range z.Cache {
+			_ = za0002
+			s += msgp.StringPrefixSize + len(za0001) + za0002.Msgsize()
 		}
 	}
 	return
@@ -996,54 +889,35 @@ func (z *dataUsageCacheV3) DecodeMsg(dc *msgp.Reader) (err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, err = dc.ReadArrayHeader()
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], err = dc.ReadString()
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, err = dc.ReadMapHeader()
+			var zb0002 uint32
+			zb0002, err = dc.ReadMapHeader()
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV3, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV3, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				zb0003--
-				var za0002 string
-				var za0003 dataUsageEntryV3
-				za0002, err = dc.ReadString()
+			for zb0002 > 0 {
+				zb0002--
+				var za0001 string
+				var za0002 dataUsageEntryV3
+				za0001, err = dc.ReadString()
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				err = za0003.DecodeMsg(dc)
+				err = za0002.DecodeMsg(dc)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			err = dc.Skip()
@@ -1080,54 +954,35 @@ func (z *dataUsageCacheV3) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, bts, err = msgp.ReadArrayHeaderBytes(bts)
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], bts, err = msgp.ReadStringBytes(bts)
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, bts, err = msgp.ReadMapHeaderBytes(bts)
+			var zb0002 uint32
+			zb0002, bts, err = msgp.ReadMapHeaderBytes(bts)
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV3, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV3, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				var za0002 string
-				var za0003 dataUsageEntryV3
-				zb0003--
-				za0002, bts, err = msgp.ReadStringBytes(bts)
+			for zb0002 > 0 {
+				var za0001 string
+				var za0002 dataUsageEntryV3
+				zb0002--
+				za0001, bts, err = msgp.ReadStringBytes(bts)
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				bts, err = za0003.UnmarshalMsg(bts)
+				bts, err = za0002.UnmarshalMsg(bts)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			bts, err = msgp.Skip(bts)
@@ -1143,15 +998,11 @@ func (z *dataUsageCacheV3) 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 *dataUsageCacheV3) Msgsize() (s int) {
-	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.ArrayHeaderSize
-	for za0001 := range z.Disks {
-		s += msgp.StringPrefixSize + len(z.Disks[za0001])
-	}
-	s += 6 + msgp.MapHeaderSize
+	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.MapHeaderSize
 	if z.Cache != nil {
-		for za0002, za0003 := range z.Cache {
-			_ = za0003
-			s += msgp.StringPrefixSize + len(za0002) + za0003.Msgsize()
+		for za0001, za0002 := range z.Cache {
+			_ = za0002
+			s += msgp.StringPrefixSize + len(za0001) + za0002.Msgsize()
 		}
 	}
 	return
@@ -1181,54 +1032,35 @@ func (z *dataUsageCacheV4) DecodeMsg(dc *msgp.Reader) (err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, err = dc.ReadArrayHeader()
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], err = dc.ReadString()
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, err = dc.ReadMapHeader()
+			var zb0002 uint32
+			zb0002, err = dc.ReadMapHeader()
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV4, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV4, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				zb0003--
-				var za0002 string
-				var za0003 dataUsageEntryV4
-				za0002, err = dc.ReadString()
+			for zb0002 > 0 {
+				zb0002--
+				var za0001 string
+				var za0002 dataUsageEntryV4
+				za0001, err = dc.ReadString()
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				err = za0003.DecodeMsg(dc)
+				err = za0002.DecodeMsg(dc)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			err = dc.Skip()
@@ -1265,54 +1097,35 @@ func (z *dataUsageCacheV4) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, bts, err = msgp.ReadArrayHeaderBytes(bts)
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], bts, err = msgp.ReadStringBytes(bts)
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, bts, err = msgp.ReadMapHeaderBytes(bts)
+			var zb0002 uint32
+			zb0002, bts, err = msgp.ReadMapHeaderBytes(bts)
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV4, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV4, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				var za0002 string
-				var za0003 dataUsageEntryV4
-				zb0003--
-				za0002, bts, err = msgp.ReadStringBytes(bts)
+			for zb0002 > 0 {
+				var za0001 string
+				var za0002 dataUsageEntryV4
+				zb0002--
+				za0001, bts, err = msgp.ReadStringBytes(bts)
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				bts, err = za0003.UnmarshalMsg(bts)
+				bts, err = za0002.UnmarshalMsg(bts)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			bts, err = msgp.Skip(bts)
@@ -1328,15 +1141,11 @@ func (z *dataUsageCacheV4) 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 *dataUsageCacheV4) Msgsize() (s int) {
-	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.ArrayHeaderSize
-	for za0001 := range z.Disks {
-		s += msgp.StringPrefixSize + len(z.Disks[za0001])
-	}
-	s += 6 + msgp.MapHeaderSize
+	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.MapHeaderSize
 	if z.Cache != nil {
-		for za0002, za0003 := range z.Cache {
-			_ = za0003
-			s += msgp.StringPrefixSize + len(za0002) + za0003.Msgsize()
+		for za0001, za0002 := range z.Cache {
+			_ = za0002
+			s += msgp.StringPrefixSize + len(za0001) + za0002.Msgsize()
 		}
 	}
 	return
@@ -1366,54 +1175,35 @@ func (z *dataUsageCacheV5) DecodeMsg(dc *msgp.Reader) (err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, err = dc.ReadArrayHeader()
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], err = dc.ReadString()
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, err = dc.ReadMapHeader()
+			var zb0002 uint32
+			zb0002, err = dc.ReadMapHeader()
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV5, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV5, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				zb0003--
-				var za0002 string
-				var za0003 dataUsageEntryV5
-				za0002, err = dc.ReadString()
+			for zb0002 > 0 {
+				zb0002--
+				var za0001 string
+				var za0002 dataUsageEntryV5
+				za0001, err = dc.ReadString()
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				err = za0003.DecodeMsg(dc)
+				err = za0002.DecodeMsg(dc)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			err = dc.Skip()
@@ -1450,54 +1240,35 @@ func (z *dataUsageCacheV5) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, bts, err = msgp.ReadArrayHeaderBytes(bts)
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], bts, err = msgp.ReadStringBytes(bts)
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, bts, err = msgp.ReadMapHeaderBytes(bts)
+			var zb0002 uint32
+			zb0002, bts, err = msgp.ReadMapHeaderBytes(bts)
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV5, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV5, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				var za0002 string
-				var za0003 dataUsageEntryV5
-				zb0003--
-				za0002, bts, err = msgp.ReadStringBytes(bts)
+			for zb0002 > 0 {
+				var za0001 string
+				var za0002 dataUsageEntryV5
+				zb0002--
+				za0001, bts, err = msgp.ReadStringBytes(bts)
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				bts, err = za0003.UnmarshalMsg(bts)
+				bts, err = za0002.UnmarshalMsg(bts)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			bts, err = msgp.Skip(bts)
@@ -1513,15 +1284,11 @@ func (z *dataUsageCacheV5) 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 *dataUsageCacheV5) Msgsize() (s int) {
-	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.ArrayHeaderSize
-	for za0001 := range z.Disks {
-		s += msgp.StringPrefixSize + len(z.Disks[za0001])
-	}
-	s += 6 + msgp.MapHeaderSize
+	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.MapHeaderSize
 	if z.Cache != nil {
-		for za0002, za0003 := range z.Cache {
-			_ = za0003
-			s += msgp.StringPrefixSize + len(za0002) + za0003.Msgsize()
+		for za0001, za0002 := range z.Cache {
+			_ = za0002
+			s += msgp.StringPrefixSize + len(za0001) + za0002.Msgsize()
 		}
 	}
 	return
@@ -1551,54 +1318,35 @@ func (z *dataUsageCacheV6) DecodeMsg(dc *msgp.Reader) (err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, err = dc.ReadArrayHeader()
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], err = dc.ReadString()
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, err = dc.ReadMapHeader()
+			var zb0002 uint32
+			zb0002, err = dc.ReadMapHeader()
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV6, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV6, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				zb0003--
-				var za0002 string
-				var za0003 dataUsageEntryV6
-				za0002, err = dc.ReadString()
+			for zb0002 > 0 {
+				zb0002--
+				var za0001 string
+				var za0002 dataUsageEntryV6
+				za0001, err = dc.ReadString()
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				err = za0003.DecodeMsg(dc)
+				err = za0002.DecodeMsg(dc)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			err = dc.Skip()
@@ -1635,54 +1383,35 @@ func (z *dataUsageCacheV6) UnmarshalMsg(bts []byte) (o []byte, err error) {
 				err = msgp.WrapError(err, "Info")
 				return
 			}
-		case "Disks":
-			var zb0002 uint32
-			zb0002, bts, err = msgp.ReadArrayHeaderBytes(bts)
-			if err != nil {
-				err = msgp.WrapError(err, "Disks")
-				return
-			}
-			if cap(z.Disks) >= int(zb0002) {
-				z.Disks = (z.Disks)[:zb0002]
-			} else {
-				z.Disks = make([]string, zb0002)
-			}
-			for za0001 := range z.Disks {
-				z.Disks[za0001], bts, err = msgp.ReadStringBytes(bts)
-				if err != nil {
-					err = msgp.WrapError(err, "Disks", za0001)
-					return
-				}
-			}
 		case "Cache":
-			var zb0003 uint32
-			zb0003, bts, err = msgp.ReadMapHeaderBytes(bts)
+			var zb0002 uint32
+			zb0002, bts, err = msgp.ReadMapHeaderBytes(bts)
 			if err != nil {
 				err = msgp.WrapError(err, "Cache")
 				return
 			}
 			if z.Cache == nil {
-				z.Cache = make(map[string]dataUsageEntryV6, zb0003)
+				z.Cache = make(map[string]dataUsageEntryV6, zb0002)
 			} else if len(z.Cache) > 0 {
 				for key := range z.Cache {
 					delete(z.Cache, key)
 				}
 			}
-			for zb0003 > 0 {
-				var za0002 string
-				var za0003 dataUsageEntryV6
-				zb0003--
-				za0002, bts, err = msgp.ReadStringBytes(bts)
+			for zb0002 > 0 {
+				var za0001 string
+				var za0002 dataUsageEntryV6
+				zb0002--
+				za0001, bts, err = msgp.ReadStringBytes(bts)
 				if err != nil {
 					err = msgp.WrapError(err, "Cache")
 					return
 				}
-				bts, err = za0003.UnmarshalMsg(bts)
+				bts, err = za0002.UnmarshalMsg(bts)
 				if err != nil {
-					err = msgp.WrapError(err, "Cache", za0002)
+					err = msgp.WrapError(err, "Cache", za0001)
 					return
 				}
-				z.Cache[za0002] = za0003
+				z.Cache[za0001] = za0002
 			}
 		default:
 			bts, err = msgp.Skip(bts)
@@ -1698,15 +1427,11 @@ func (z *dataUsageCacheV6) 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 *dataUsageCacheV6) Msgsize() (s int) {
-	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.ArrayHeaderSize
-	for za0001 := range z.Disks {
-		s += msgp.StringPrefixSize + len(z.Disks[za0001])
-	}
-	s += 6 + msgp.MapHeaderSize
+	s = 1 + 5 + z.Info.Msgsize() + 6 + msgp.MapHeaderSize
 	if z.Cache != nil {
-		for za0002, za0003 := range z.Cache {
-			_ = za0003
-			s += msgp.StringPrefixSize + len(za0002) + za0003.Msgsize()
+		for za0001, za0002 := range z.Cache {
+			_ = za0002
+			s += msgp.StringPrefixSize + len(za0001) + za0002.Msgsize()
 		}
 	}
 	return

cmd/data-usage.go

@@ -18,13 +18,11 @@
 package cmd
 
 import (
-	"bytes"
 	"context"
-	"net/http"
+	"errors"
 	"strings"
 
 	jsoniter "github.com/json-iterator/go"
-	"github.com/minio/minio/internal/hash"
 	"github.com/minio/minio/internal/logger"
 )
 
@@ -32,28 +30,24 @@ const (
 	dataUsageRoot   = SlashSeparator
 	dataUsageBucket = minioMetaBucket + SlashSeparator + bucketMetaPrefix
 
-	dataUsageObjName   = ".usage.json"
+	dataUsageObjName       = ".usage.json"
+	dataUsageObjNamePath   = bucketMetaPrefix + SlashSeparator + dataUsageObjName
+	dataUsageBloomName     = ".bloomcycle.bin"
+	dataUsageBloomNamePath = bucketMetaPrefix + SlashSeparator + dataUsageBloomName
+
 	dataUsageCacheName = ".usage-cache.bin"
-	dataUsageBloomName = ".bloomcycle.bin"
 )
 
 // storeDataUsageInBackend will store all objects sent on the gui channel until closed.
 func storeDataUsageInBackend(ctx context.Context, objAPI ObjectLayer, dui <-chan DataUsageInfo) {
 	for dataUsageInfo := range dui {
-		var json = jsoniter.ConfigCompatibleWithStandardLibrary
+		json := jsoniter.ConfigCompatibleWithStandardLibrary
 		dataUsageJSON, err := json.Marshal(dataUsageInfo)
 		if err != nil {
 			logger.LogIf(ctx, err)
 			continue
 		}
-		size := int64(len(dataUsageJSON))
-		r, err := hash.NewReader(bytes.NewReader(dataUsageJSON), size, "", "", size)
-		if err != nil {
-			logger.LogIf(ctx, err)
-			continue
-		}
-		_, err = objAPI.PutObject(ctx, dataUsageBucket, dataUsageObjName, NewPutObjReader(r), ObjectOptions{})
-		if !isErrBucketNotFound(err) {
+		if err = saveConfig(ctx, objAPI, dataUsageObjNamePath, dataUsageJSON); err != nil {
 			logger.LogIf(ctx, err)
 		}
 	}
@@ -95,18 +89,17 @@ func loadPrefixUsageFromBackend(ctx context.Context, objAPI ObjectLayer, bucket
 }
 
 func loadDataUsageFromBackend(ctx context.Context, objAPI ObjectLayer) (DataUsageInfo, error) {
-	r, err := objAPI.GetObjectNInfo(ctx, dataUsageBucket, dataUsageObjName, nil, http.Header{}, readLock, ObjectOptions{})
+	buf, err := readConfig(ctx, objAPI, dataUsageObjNamePath)
 	if err != nil {
-		if isErrObjectNotFound(err) || isErrBucketNotFound(err) {
+		if errors.Is(err, errConfigNotFound) {
 			return DataUsageInfo{}, nil
 		}
-		return DataUsageInfo{}, toObjectErr(err, dataUsageBucket, dataUsageObjName)
+		return DataUsageInfo{}, toObjectErr(err, minioMetaBucket, dataUsageObjNamePath)
 	}
-	defer r.Close()
 
 	var dataUsageInfo DataUsageInfo
-	var json = jsoniter.ConfigCompatibleWithStandardLibrary
-	if err = json.NewDecoder(r).Decode(&dataUsageInfo); err != nil {
+	json := jsoniter.ConfigCompatibleWithStandardLibrary
+	if err = json.Unmarshal(buf, &dataUsageInfo); err != nil {
 		return DataUsageInfo{}, err
 	}
 	// For forward compatibility reasons, we need to add this code.

cmd/data-usage_test.go

@@ -41,7 +41,7 @@ func TestDataUsageUpdate(t *testing.T) {
 	}
 	const bucket = "bucket"
 	defer os.RemoveAll(base)
-	var files = []usageTestFile{
+	files := []usageTestFile{
 		{name: "rootfile", size: 10000},
 		{name: "rootfile2", size: 10000},
 		{name: "dir1/d1file", size: 2000},
@@ -67,13 +67,13 @@ func TestDataUsageUpdate(t *testing.T) {
 		return
 	}
 
-	got, err := scanDataFolder(context.Background(), base, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize)
+	got, err := scanDataFolder(context.Background(), 0, 0, base, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize)
 	if err != nil {
 		t.Fatal(err)
 	}
 
 	// Test dirs
-	var want = []struct {
+	want := []struct {
 		path       string
 		isNil      bool
 		size, objs int
@@ -178,7 +178,7 @@ func TestDataUsageUpdate(t *testing.T) {
 	}
 	// Changed dir must be picked up in this many cycles.
 	for i := 0; i < dataUsageUpdateDirCycles; i++ {
-		got, err = scanDataFolder(context.Background(), base, got, getSize)
+		got, err = scanDataFolder(context.Background(), 0, 0, base, got, getSize)
 		got.Info.NextCycle++
 		if err != nil {
 			t.Fatal(err)
@@ -257,7 +257,7 @@ func TestDataUsageUpdatePrefix(t *testing.T) {
 	}
 	scannerSleeper.Update(0, 0)
 	defer os.RemoveAll(base)
-	var files = []usageTestFile{
+	files := []usageTestFile{
 		{name: "bucket/rootfile", size: 10000},
 		{name: "bucket/rootfile2", size: 10000},
 		{name: "bucket/dir1/d1file", size: 2000},
@@ -289,7 +289,7 @@ func TestDataUsageUpdatePrefix(t *testing.T) {
 		}
 		return
 	}
-	got, err := scanDataFolder(context.Background(), base, dataUsageCache{Info: dataUsageCacheInfo{Name: "bucket"}}, getSize)
+	got, err := scanDataFolder(context.Background(), 0, 0, base, dataUsageCache{Info: dataUsageCacheInfo{Name: "bucket"}}, getSize)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -302,7 +302,7 @@ func TestDataUsageUpdatePrefix(t *testing.T) {
 	}
 
 	// Test dirs
-	var want = []struct {
+	want := []struct {
 		path       string
 		isNil      bool
 		size, objs int
@@ -423,7 +423,7 @@ func TestDataUsageUpdatePrefix(t *testing.T) {
 	}
 	// Changed dir must be picked up in this many cycles.
 	for i := 0; i < dataUsageUpdateDirCycles; i++ {
-		got, err = scanDataFolder(context.Background(), base, got, getSize)
+		got, err = scanDataFolder(context.Background(), 0, 0, base, got, getSize)
 		got.Info.NextCycle++
 		if err != nil {
 			t.Fatal(err)
@@ -543,7 +543,7 @@ func TestDataUsageCacheSerialize(t *testing.T) {
 	}
 	const bucket = "abucket"
 	defer os.RemoveAll(base)
-	var files = []usageTestFile{
+	files := []usageTestFile{
 		{name: "rootfile", size: 10000},
 		{name: "rootfile2", size: 10000},
 		{name: "dir1/d1file", size: 2000},
@@ -575,7 +575,7 @@ func TestDataUsageCacheSerialize(t *testing.T) {
 		}
 		return
 	}
-	want, err := scanDataFolder(context.Background(), base, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize)
+	want, err := scanDataFolder(context.Background(), 0, 0, base, dataUsageCache{Info: dataUsageCacheInfo{Name: bucket}}, getSize)
 	if err != nil {
 		t.Fatal(err)
 	}

cmd/disk-cache-backend.go

@@ -190,7 +190,7 @@ func newDiskCache(ctx context.Context, dir string, config cache.Config) (*diskCa
 		quotaPct = config.Quota
 	}
 
-	if err := os.MkdirAll(dir, 0777); err != nil {
+	if err := os.MkdirAll(dir, 0o777); err != nil {
 		return nil, fmt.Errorf("Unable to initialize '%s' dir, %w", dir, err)
 	}
 	cache := diskCache{
@@ -619,10 +619,10 @@ func (c *diskCache) saveMetadata(ctx context.Context, bucket, object string, met
 	cachedPath := getCacheSHADir(c.dir, bucket, object)
 	metaPath := pathJoin(cachedPath, cacheMetaJSONFile)
 	// Create cache directory if needed
-	if err := os.MkdirAll(cachedPath, 0777); err != nil {
+	if err := os.MkdirAll(cachedPath, 0o777); err != nil {
 		return err
 	}
-	f, err := os.OpenFile(metaPath, os.O_RDWR|os.O_CREATE, 0666)
+	f, err := os.OpenFile(metaPath, os.O_RDWR|os.O_CREATE, 0o666)
 	if err != nil {
 		return err
 	}
@@ -682,10 +682,10 @@ func (c *diskCache) updateMetadata(ctx context.Context, bucket, object, etag str
 	cachedPath := getCacheSHADir(c.dir, bucket, object)
 	metaPath := pathJoin(cachedPath, cacheMetaJSONFile)
 	// Create cache directory if needed
-	if err := os.MkdirAll(cachedPath, 0777); err != nil {
+	if err := os.MkdirAll(cachedPath, 0o777); err != nil {
 		return err
 	}
-	f, err := os.OpenFile(metaPath, os.O_RDWR, 0666)
+	f, err := os.OpenFile(metaPath, os.O_RDWR, 0o666)
 	if err != nil {
 		return err
 	}
@@ -732,7 +732,7 @@ func getCacheWriteBackSHADir(dir, bucket, object string) string {
 
 // Cache data to disk with bitrot checksum added for each block of 1MB
 func (c *diskCache) bitrotWriteToCache(cachePath, fileName string, reader io.Reader, size uint64) (int64, string, error) {
-	if err := os.MkdirAll(cachePath, 0777); err != nil {
+	if err := os.MkdirAll(cachePath, 0o777); err != nil {
 		return 0, "", err
 	}
 	filePath := pathJoin(cachePath, fileName)
@@ -807,6 +807,7 @@ func newCacheEncryptReader(content io.Reader, bucket, object string, metadata ma
 	}
 	return reader, nil
 }
+
 func newCacheEncryptMetadata(bucket, object string, metadata map[string]string) ([]byte, error) {
 	var sealedKey crypto.SealedKey
 	if globalCacheKMS == nil {
@@ -827,6 +828,7 @@ func newCacheEncryptMetadata(bucket, object string, metadata map[string]string)
 	metadata[SSECacheEncrypted] = ""
 	return objectKey[:], nil
 }
+
 func (c *diskCache) GetLockContext(ctx context.Context, bucket, object string) (RWLocker, LockContext, error) {
 	cachePath := getCacheSHADir(c.dir, bucket, object)
 	cLock := c.NewNSLockFn(cachePath)
@@ -879,12 +881,12 @@ func (c *diskCache) put(ctx context.Context, bucket, object string, data io.Read
 		cachePath = getCacheWriteBackSHADir(c.dir, bucket, object)
 	}
 
-	if err := os.MkdirAll(cachePath, 0777); err != nil {
+	if err := os.MkdirAll(cachePath, 0o777); err != nil {
 		return oi, err
 	}
-	var metadata = cloneMSS(opts.UserDefined)
-	var reader = data
-	var actualSize = uint64(size)
+	metadata := cloneMSS(opts.UserDefined)
+	reader := data
+	actualSize := uint64(size)
 	if globalCacheKMS != nil {
 		reader, err = newCacheEncryptReader(data, bucket, object, metadata)
 		if err != nil {
@@ -933,14 +935,14 @@ func (c *diskCache) putRange(ctx context.Context, bucket, object string, data io
 		return errDiskFull
 	}
 	cachePath := getCacheSHADir(c.dir, bucket, object)
-	if err := os.MkdirAll(cachePath, 0777); err != nil {
+	if err := os.MkdirAll(cachePath, 0o777); err != nil {
 		return err
 	}
-	var metadata = cloneMSS(opts.UserDefined)
-	var reader = data
-	var actualSize = uint64(rlen)
+	metadata := cloneMSS(opts.UserDefined)
+	reader := data
+	actualSize := uint64(rlen)
 	// objSize is the actual size of object (with encryption overhead if any)
-	var objSize = uint64(size)
+	objSize := uint64(size)
 	if globalCacheKMS != nil {
 		reader, err = newCacheEncryptReader(data, bucket, object, metadata)
 		if err != nil {
@@ -1269,12 +1271,12 @@ func (c *diskCache) NewMultipartUpload(ctx context.Context, bucket, object, uID
 
 	cachePath := getMultipartCacheSHADir(c.dir, bucket, object)
 	uploadIDDir := path.Join(cachePath, uploadID)
-	if err := os.MkdirAll(uploadIDDir, 0777); err != nil {
+	if err := os.MkdirAll(uploadIDDir, 0o777); err != nil {
 		return uploadID, err
 	}
 	metaPath := pathJoin(uploadIDDir, cacheMetaJSONFile)
 
-	f, err := os.OpenFile(metaPath, os.O_RDWR|os.O_CREATE, 0666)
+	f, err := os.OpenFile(metaPath, os.O_RDWR|os.O_CREATE, 0o666)
 	if err != nil {
 		return uploadID, err
 	}
@@ -1331,7 +1333,7 @@ func (c *diskCache) PutObjectPart(ctx context.Context, bucket, object, uploadID
 		return oi, errDiskFull
 	}
 	reader := data
-	var actualSize = uint64(size)
+	actualSize := uint64(size)
 	if globalCacheKMS != nil {
 		reader, err = newCachePartEncryptReader(ctx, bucket, object, partID, data, size, meta.Meta)
 		if err != nil {
@@ -1380,7 +1382,7 @@ func (c *diskCache) SavePartMetadata(ctx context.Context, bucket, object, upload
 	defer uploadLock.Unlock(ulkctx.Cancel)
 
 	metaPath := pathJoin(uploadDir, cacheMetaJSONFile)
-	f, err := os.OpenFile(metaPath, os.O_RDWR, 0666)
+	f, err := os.OpenFile(metaPath, os.O_RDWR, 0o666)
 	if err != nil {
 		return err
 	}
@@ -1558,7 +1560,7 @@ func (c *diskCache) CompleteMultipartUpload(ctx context.Context, bucket, object,
 	uploadMeta.Hits++
 	metaPath := pathJoin(uploadIDDir, cacheMetaJSONFile)
 
-	f, err := os.OpenFile(metaPath, os.O_RDWR|os.O_CREATE, 0666)
+	f, err := os.OpenFile(metaPath, os.O_RDWR|os.O_CREATE, 0o666)
 	if err != nil {
 		return oi, err
 	}

cmd/disk-cache-utils.go

@@ -113,7 +113,6 @@ func cacheControlOpts(o ObjectInfo) *cacheControl {
 		if strings.EqualFold(k, "cache-control") {
 			headerVal = v
 		}
-
 	}
 	if headerVal == "" {
 		return nil
@@ -581,6 +580,7 @@ func (t *multiWriter) Write(p []byte) (n int, err error) {
 	}
 	return len(p), nil
 }
+
 func cacheMultiWriter(w1 io.Writer, w2 *io.PipeWriter) io.Writer {
 	return &multiWriter{backendWriter: w1, cacheWriter: w2}
 }

cmd/disk-cache-utils_test.go

@@ -38,7 +38,7 @@ func TestGetCacheControlOpts(t *testing.T) {
 		{"max-age=2592000, no-store", timeSentinel, &cacheControl{maxAge: 2592000, sMaxAge: 0, noStore: true, minFresh: 0, expiry: time.Time{}}, false},
 		{"must-revalidate, max-age=600", timeSentinel, &cacheControl{maxAge: 600, sMaxAge: 0, minFresh: 0, expiry: time.Time{}}, false},
 		{"s-maxAge=2500, max-age=600", timeSentinel, &cacheControl{maxAge: 600, sMaxAge: 2500, minFresh: 0, expiry: time.Time{}}, false},
-		{"s-maxAge=2500, max-age=600", expiry, &cacheControl{maxAge: 600, sMaxAge: 2500, minFresh: 0, expiry: time.Date(2015, time.October, 21, 07, 28, 00, 00, time.UTC)}, false},
+		{"s-maxAge=2500, max-age=600", expiry, &cacheControl{maxAge: 600, sMaxAge: 2500, minFresh: 0, expiry: time.Date(2015, time.October, 21, 0o7, 28, 0o0, 0o0, time.UTC)}, false},
 		{"s-maxAge=2500, max-age=600s", timeSentinel, &cacheControl{maxAge: 600, sMaxAge: 2500, minFresh: 0, expiry: time.Time{}}, true},
 	}
 
@@ -61,7 +61,6 @@ func TestGetCacheControlOpts(t *testing.T) {
 }
 
 func TestIsMetadataSame(t *testing.T) {
-
 	testCases := []struct {
 		m1       map[string]string
 		m2       map[string]string
@@ -148,6 +147,7 @@ func TestNewFileScorer(t *testing.T) {
 		t.Fatal("unexpected file list", fs.queueString())
 	}
 }
+
 func TestBytesToClear(t *testing.T) {
 	testCases := []struct {
 		total         int64

cmd/disk-cache.go

@@ -137,7 +137,6 @@ func (c *cacheObjects) incHitsToMeta(ctx context.Context, dcache *diskCache, buc
 
 // Backend metadata could have changed through server side copy - reset cache metadata if that is the case
 func (c *cacheObjects) updateMetadataIfChanged(ctx context.Context, dcache *diskCache, bucket, object string, bkObjectInfo, cacheObjInfo ObjectInfo, rs *HTTPRangeSpec) error {
-
 	bkMeta := make(map[string]string, len(bkObjectInfo.UserDefined))
 	cacheMeta := make(map[string]string, len(cacheObjInfo.UserDefined))
 	for k, v := range bkObjectInfo.UserDefined {
@@ -1154,7 +1153,7 @@ func (c *cacheObjects) CompleteMultipartUpload(ctx context.Context, bucket, obje
 	if err == nil {
 		// fill cache in the background
 		go func() {
-			_, err := dcache.CompleteMultipartUpload(ctx, bucket, object, uploadID, uploadedParts, oi, opts)
+			_, err := dcache.CompleteMultipartUpload(bgContext(ctx), bucket, object, uploadID, uploadedParts, oi, opts)
 			if err != nil {
 				// fill cache in the background
 				bReader, bErr := c.InnerGetObjectNInfoFn(GlobalContext, bucket, object, nil, http.Header{}, readLock, ObjectOptions{})

cmd/dynamic-timeouts_test.go

@@ -26,7 +26,6 @@ import (
 )
 
 func TestDynamicTimeoutSingleIncrease(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	initial := timeout.Timeout()
@@ -43,7 +42,6 @@ func TestDynamicTimeoutSingleIncrease(t *testing.T) {
 }
 
 func TestDynamicTimeoutDualIncrease(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	initial := timeout.Timeout()
@@ -66,7 +64,6 @@ func TestDynamicTimeoutDualIncrease(t *testing.T) {
 }
 
 func TestDynamicTimeoutSingleDecrease(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	initial := timeout.Timeout()
@@ -83,7 +80,6 @@ func TestDynamicTimeoutSingleDecrease(t *testing.T) {
 }
 
 func TestDynamicTimeoutDualDecrease(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	initial := timeout.Timeout()
@@ -106,7 +102,6 @@ func TestDynamicTimeoutDualDecrease(t *testing.T) {
 }
 
 func TestDynamicTimeoutManyDecreases(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	initial := timeout.Timeout()
@@ -116,7 +111,6 @@ func TestDynamicTimeoutManyDecreases(t *testing.T) {
 		for i := 0; i < dynamicTimeoutLogSize; i++ {
 			timeout.LogSuccess(successTimeout)
 		}
-
 	}
 
 	adjusted := timeout.Timeout()
@@ -151,7 +145,6 @@ func TestDynamicTimeoutConcurrent(t *testing.T) {
 }
 
 func TestDynamicTimeoutHitMinimum(t *testing.T) {
-
 	const minimum = 30 * time.Second
 	timeout := newDynamicTimeout(time.Minute, minimum)
 
@@ -172,7 +165,6 @@ func TestDynamicTimeoutHitMinimum(t *testing.T) {
 }
 
 func testDynamicTimeoutAdjust(t *testing.T, timeout *dynamicTimeout, f func() float64) {
-
 	const successTimeout = 20 * time.Second
 
 	for i := 0; i < dynamicTimeoutLogSize; i++ {
@@ -192,7 +184,6 @@ func testDynamicTimeoutAdjust(t *testing.T, timeout *dynamicTimeout, f func() fl
 }
 
 func TestDynamicTimeoutAdjustExponential(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	rand.Seed(0)
@@ -200,9 +191,7 @@ func TestDynamicTimeoutAdjustExponential(t *testing.T) {
 	initial := timeout.Timeout()
 
 	for try := 0; try < 10; try++ {
-
 		testDynamicTimeoutAdjust(t, timeout, rand.ExpFloat64)
-
 	}
 
 	adjusted := timeout.Timeout()
@@ -212,7 +201,6 @@ func TestDynamicTimeoutAdjustExponential(t *testing.T) {
 }
 
 func TestDynamicTimeoutAdjustNormalized(t *testing.T) {
-
 	timeout := newDynamicTimeout(time.Minute, time.Second)
 
 	rand.Seed(0)
@@ -220,11 +208,9 @@ func TestDynamicTimeoutAdjustNormalized(t *testing.T) {
 	initial := timeout.Timeout()
 
 	for try := 0; try < 10; try++ {
-
 		testDynamicTimeoutAdjust(t, timeout, func() float64 {
 			return 1.0 + rand.NormFloat64()
 		})
-
 	}
 
 	adjusted := timeout.Timeout()

cmd/encryption-v1.go

@@ -186,7 +186,7 @@ func rotateKey(oldKey []byte, newKeyID string, newKey []byte, bucket, object str
 		// client provided it. Therefore, we create a copy
 		// of the client provided context and add the bucket
 		// key, if not present.
-		var kmsCtx = kms.Context{}
+		kmsCtx := kms.Context{}
 		for k, v := range ctx {
 			kmsCtx[k] = v
 		}
@@ -253,7 +253,7 @@ func newEncryptMetadata(kind crypto.Type, keyID string, key []byte, bucket, obje
 		// client provided it. Therefore, we create a copy
 		// of the client provided context and add the bucket
 		// key, if not present.
-		var kmsCtx = kms.Context{}
+		kmsCtx := kms.Context{}
 		for k, v := range ctx {
 			kmsCtx[k] = v
 		}
@@ -443,7 +443,6 @@ func newDecryptReaderWithObjectKey(client io.Reader, objectEncryptionKey []byte,
 // DecryptBlocksRequestR - same as DecryptBlocksRequest but with a
 // reader
 func DecryptBlocksRequestR(inputReader io.Reader, h http.Header, seqNumber uint32, partStart int, oi ObjectInfo, copySource bool) (io.Reader, error) {
-
 	bucket, object := oi.Bucket, oi.Name
 	// Single part case
 	if !oi.isMultipart() {

cmd/encryption-v1_test.go

@@ -64,7 +64,6 @@ func TestEncryptRequest(t *testing.T) {
 			req.Header.Set(k, v)
 		}
 		_, _, err := EncryptRequest(content, req, "bucket", "object", test.metadata)
-
 		if err != nil {
 			t.Fatalf("Test %d: Failed to encrypt request: %v", i, err)
 		}
@@ -285,14 +284,13 @@ func TestGetDecryptedRange(t *testing.T) {
 	)
 
 	// Single part object tests
-	var (
-		mkSPObj = func(s int64) ObjectInfo {
-			return ObjectInfo{
-				Size:        getEncSize(s),
-				UserDefined: udMap(false),
-			}
+
+	mkSPObj := func(s int64) ObjectInfo {
+		return ObjectInfo{
+			Size:        getEncSize(s),
+			UserDefined: udMap(false),
 		}
-	)
+	}
 
 	testSP := []struct {
 		decSz int64
@@ -325,7 +323,7 @@ func TestGetDecryptedRange(t *testing.T) {
 			if err != nil {
 				t.Errorf("Case %d: unexpected err: %v", i, err)
 			}
-			var rLen = pkgSz + 32
+			rLen := pkgSz + 32
 			if test.decSz < pkgSz {
 				rLen = test.decSz + 32
 			}
@@ -341,7 +339,7 @@ func TestGetDecryptedRange(t *testing.T) {
 			if err != nil {
 				t.Errorf("Case %d: unexpected err: %v", i, err)
 			}
-			var rLen = (pkgSz + 32) * 2
+			rLen := (pkgSz + 32) * 2
 			if test.decSz < 2*pkgSz {
 				rLen = (pkgSz + 32) + (test.decSz - pkgSz + 32)
 			}
@@ -356,7 +354,7 @@ func TestGetDecryptedRange(t *testing.T) {
 			if err != nil {
 				t.Errorf("Case %d: unexpected err: %v", i, err)
 			}
-			var rLen = (pkgSz + 32) * 2
+			rLen := (pkgSz + 32) * 2
 			if test.decSz-pkgSz < 2*pkgSz {
 				rLen = (pkgSz + 32) + (test.decSz - pkgSz + 32*2)
 			}
@@ -551,60 +549,90 @@ var getDefaultOptsTests = []struct {
 	encryptionType encrypt.Type
 	err            error
 }{
-	{headers: http.Header{xhttp.AmzServerSideEncryptionCustomerAlgorithm: []string{"AES256"},
-		xhttp.AmzServerSideEncryptionCustomerKey:    []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
-		xhttp.AmzServerSideEncryptionCustomerKeyMD5: []string{"7PpPLAK26ONlVUGOWlusfg=="}},
+	{
+		headers: http.Header{
+			xhttp.AmzServerSideEncryptionCustomerAlgorithm: []string{"AES256"},
+			xhttp.AmzServerSideEncryptionCustomerKey:       []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
+			xhttp.AmzServerSideEncryptionCustomerKeyMD5:    []string{"7PpPLAK26ONlVUGOWlusfg=="},
+		},
 		copySource:     false,
 		metadata:       nil,
 		encryptionType: encrypt.SSEC,
-		err:            nil}, // 0
-	{headers: http.Header{xhttp.AmzServerSideEncryptionCustomerAlgorithm: []string{"AES256"},
-		xhttp.AmzServerSideEncryptionCustomerKey:    []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
-		xhttp.AmzServerSideEncryptionCustomerKeyMD5: []string{"7PpPLAK26ONlVUGOWlusfg=="}},
+		err:            nil,
+	}, // 0
+	{
+		headers: http.Header{
+			xhttp.AmzServerSideEncryptionCustomerAlgorithm: []string{"AES256"},
+			xhttp.AmzServerSideEncryptionCustomerKey:       []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
+			xhttp.AmzServerSideEncryptionCustomerKeyMD5:    []string{"7PpPLAK26ONlVUGOWlusfg=="},
+		},
 		copySource:     true,
 		metadata:       nil,
 		encryptionType: "",
-		err:            nil}, // 1
-	{headers: http.Header{xhttp.AmzServerSideEncryptionCustomerAlgorithm: []string{"AES256"},
-		xhttp.AmzServerSideEncryptionCustomerKey:    []string{"Mz"},
-		xhttp.AmzServerSideEncryptionCustomerKeyMD5: []string{"7PpPLAK26ONlVUGOWlusfg=="}},
+		err:            nil,
+	}, // 1
+	{
+		headers: http.Header{
+			xhttp.AmzServerSideEncryptionCustomerAlgorithm: []string{"AES256"},
+			xhttp.AmzServerSideEncryptionCustomerKey:       []string{"Mz"},
+			xhttp.AmzServerSideEncryptionCustomerKeyMD5:    []string{"7PpPLAK26ONlVUGOWlusfg=="},
+		},
 		copySource:     false,
 		metadata:       nil,
 		encryptionType: "",
-		err:            crypto.ErrInvalidCustomerKey}, // 2
-	{headers: http.Header{xhttp.AmzServerSideEncryption: []string{"AES256"}},
+		err:            crypto.ErrInvalidCustomerKey,
+	}, // 2
+	{
+		headers:        http.Header{xhttp.AmzServerSideEncryption: []string{"AES256"}},
 		copySource:     false,
 		metadata:       nil,
 		encryptionType: encrypt.S3,
-		err:            nil}, // 3
-	{headers: http.Header{},
+		err:            nil,
+	}, // 3
+	{
+		headers:    http.Header{},
 		copySource: false,
-		metadata: map[string]string{crypto.MetaSealedKeyS3: base64.StdEncoding.EncodeToString(make([]byte, 64)),
+		metadata: map[string]string{
+			crypto.MetaSealedKeyS3:       base64.StdEncoding.EncodeToString(make([]byte, 64)),
 			crypto.MetaKeyID:             "kms-key",
-			crypto.MetaDataEncryptionKey: "m-key"},
+			crypto.MetaDataEncryptionKey: "m-key",
+		},
 		encryptionType: encrypt.S3,
-		err:            nil}, // 4
-	{headers: http.Header{},
+		err:            nil,
+	}, // 4
+	{
+		headers:    http.Header{},
 		copySource: true,
-		metadata: map[string]string{crypto.MetaSealedKeyS3: base64.StdEncoding.EncodeToString(make([]byte, 64)),
+		metadata: map[string]string{
+			crypto.MetaSealedKeyS3:       base64.StdEncoding.EncodeToString(make([]byte, 64)),
 			crypto.MetaKeyID:             "kms-key",
-			crypto.MetaDataEncryptionKey: "m-key"},
+			crypto.MetaDataEncryptionKey: "m-key",
+		},
 		encryptionType: "",
-		err:            nil}, // 5
-	{headers: http.Header{xhttp.AmzServerSideEncryptionCopyCustomerAlgorithm: []string{"AES256"},
-		xhttp.AmzServerSideEncryptionCopyCustomerKey:    []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
-		xhttp.AmzServerSideEncryptionCopyCustomerKeyMD5: []string{"7PpPLAK26ONlVUGOWlusfg=="}},
+		err:            nil,
+	}, // 5
+	{
+		headers: http.Header{
+			xhttp.AmzServerSideEncryptionCopyCustomerAlgorithm: []string{"AES256"},
+			xhttp.AmzServerSideEncryptionCopyCustomerKey:       []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
+			xhttp.AmzServerSideEncryptionCopyCustomerKeyMD5:    []string{"7PpPLAK26ONlVUGOWlusfg=="},
+		},
 		copySource:     true,
 		metadata:       nil,
 		encryptionType: encrypt.SSEC,
-		err:            nil}, // 6
-	{headers: http.Header{xhttp.AmzServerSideEncryptionCopyCustomerAlgorithm: []string{"AES256"},
-		xhttp.AmzServerSideEncryptionCopyCustomerKey:    []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
-		xhttp.AmzServerSideEncryptionCopyCustomerKeyMD5: []string{"7PpPLAK26ONlVUGOWlusfg=="}},
+		err:            nil,
+	}, // 6
+	{
+		headers: http.Header{
+			xhttp.AmzServerSideEncryptionCopyCustomerAlgorithm: []string{"AES256"},
+			xhttp.AmzServerSideEncryptionCopyCustomerKey:       []string{"MzJieXRlc2xvbmdzZWNyZXRrZXltdXN0cHJvdmlkZWQ="},
+			xhttp.AmzServerSideEncryptionCopyCustomerKeyMD5:    []string{"7PpPLAK26ONlVUGOWlusfg=="},
+		},
 		copySource:     false,
 		metadata:       nil,
 		encryptionType: "",
-		err:            nil}, // 7
+		err:            nil,
+	}, // 7
 }
 
 func TestGetDefaultOpts(t *testing.T) {

cmd/endpoint-ellipses.go

@@ -90,7 +90,7 @@ func commonSetDriveCount(divisibleSize uint64, setCounts []uint64) (setSize uint
 // input argument patterns, the symmetry calculation is to ensure that
 // we also use uniform number of drives common across all ellipses patterns.
 func possibleSetCountsWithSymmetry(setCounts []uint64, argPatterns []ellipses.ArgPattern) []uint64 {
-	var newSetCounts = make(map[uint64]struct{})
+	newSetCounts := make(map[uint64]struct{})
 	for _, ss := range setCounts {
 		var symmetry bool
 		for _, argPattern := range argPatterns {
@@ -224,7 +224,7 @@ func (s endpointSet) getEndpoints() (endpoints []string) {
 // this function also intelligently decides on what will
 // be the right set size etc.
 func (s endpointSet) Get() (sets [][]string) {
-	var k = uint64(0)
+	k := uint64(0)
 	endpoints := s.getEndpoints()
 	for i := range s.setIndexes {
 		for j := range s.setIndexes[i] {
@@ -253,7 +253,7 @@ func getTotalSizes(argPatterns []ellipses.ArgPattern) []uint64 {
 // of endpoints following the ellipses pattern, this is what is used
 // by the object layer for initializing itself.
 func parseEndpointSet(customSetDriveCount uint64, args ...string) (ep endpointSet, err error) {
-	var argPatterns = make([]ellipses.ArgPattern, len(args))
+	argPatterns := make([]ellipses.ArgPattern, len(args))
 	for i, arg := range args {
 		patterns, perr := ellipses.FindEllipsesPatterns(arg)
 		if perr != nil {
@@ -332,15 +332,13 @@ const (
 	EnvErasureSetDriveCount = "MINIO_ERASURE_SET_DRIVE_COUNT"
 )
 
-var (
-	globalCustomErasureDriveCount = false
-)
+var globalCustomErasureDriveCount = false
 
 // CreateServerEndpoints - validates and creates new endpoints from input args, supports
 // both ellipses and without ellipses transparently.
 func createServerEndpoints(serverAddr string, args ...string) (
-	endpointServerPools EndpointServerPools, setupType SetupType, err error) {
-
+	endpointServerPools EndpointServerPools, setupType SetupType, err error,
+) {
 	if len(args) == 0 {
 		return nil, -1, errInvalidArgument
 	}
@@ -355,9 +353,11 @@ func createServerEndpoints(serverAddr string, args ...string) (
 			return nil, -1, err
 		}
 		endpointServerPools = append(endpointServerPools, PoolEndpoints{
+			Legacy:       true,
 			SetCount:     len(setArgs),
 			DrivesPerSet: len(setArgs[0]),
 			Endpoints:    endpointList,
+			CmdLine:      strings.Join(args, " "),
 		})
 		setupType = newSetupType
 		return endpointServerPools, setupType, nil
@@ -378,6 +378,7 @@ func createServerEndpoints(serverAddr string, args ...string) (
 			SetCount:     len(setArgs),
 			DrivesPerSet: len(setArgs[0]),
 			Endpoints:    endpointList,
+			CmdLine:      arg,
 		}); err != nil {
 			return nil, -1, err
 		}

cmd/endpoint-ellipses_test.go

@@ -72,7 +72,8 @@ func TestGetDivisibleSize(t *testing.T) {
 	testCases := []struct {
 		totalSizes []uint64
 		result     uint64
-	}{{[]uint64{24, 32, 16}, 8},
+	}{
+		{[]uint64{24, 32, 16}, 8},
 		{[]uint64{32, 8, 4}, 4},
 		{[]uint64{8, 8, 8}, 8},
 		{[]uint64{24}, 24},
@@ -168,7 +169,7 @@ func TestGetSetIndexesEnvOverride(t *testing.T) {
 	for _, testCase := range testCases {
 		testCase := testCase
 		t.Run("", func(t *testing.T) {
-			var argPatterns = make([]ellipses.ArgPattern, len(testCase.args))
+			argPatterns := make([]ellipses.ArgPattern, len(testCase.args))
 			for i, arg := range testCase.args {
 				patterns, err := ellipses.FindEllipsesPatterns(arg)
 				if err != nil {
@@ -290,7 +291,7 @@ func TestGetSetIndexes(t *testing.T) {
 	for _, testCase := range testCases {
 		testCase := testCase
 		t.Run("", func(t *testing.T) {
-			var argPatterns = make([]ellipses.ArgPattern, len(testCase.args))
+			argPatterns := make([]ellipses.ArgPattern, len(testCase.args))
 			for i, arg := range testCase.args {
 				patterns, err := ellipses.FindEllipsesPatterns(arg)
 				if err != nil {
@@ -552,8 +553,10 @@ func TestParseEndpointSet(t *testing.T) {
 					},
 				},
 				nil,
-				[][]uint64{{16, 16, 16, 16, 16, 16, 16, 16,
-					16, 16, 16, 16, 16, 16, 16, 16}},
+				[][]uint64{{
+					16, 16, 16, 16, 16, 16, 16, 16,
+					16, 16, 16, 16, 16, 16, 16, 16,
+				}},
 			},
 			true,
 		},

cmd/endpoint.go

@@ -197,14 +197,28 @@ func NewEndpoint(arg string) (ep Endpoint, e error) {
 // PoolEndpoints represent endpoints in a given pool
 // along with its setCount and setDriveCount.
 type PoolEndpoints struct {
+	// indicates if endpoints are provided in non-ellipses style
+	Legacy       bool
 	SetCount     int
 	DrivesPerSet int
 	Endpoints    Endpoints
+	CmdLine      string
 }
 
 // EndpointServerPools - list of list of endpoints
 type EndpointServerPools []PoolEndpoints
 
+// GetPoolIdx return pool index
+func (l EndpointServerPools) GetPoolIdx(pool string) int {
+	for id, ep := range globalEndpoints {
+		if ep.CmdLine != pool {
+			continue
+		}
+		return id
+	}
+	return -1
+}
+
 // GetLocalPoolIdx returns the pool which endpoint belongs to locally.
 // if ep is remote this code will return -1 poolIndex
 func (l EndpointServerPools) GetLocalPoolIdx(ep Endpoint) int {
@@ -220,6 +234,13 @@ func (l EndpointServerPools) GetLocalPoolIdx(ep Endpoint) int {
 	return -1
 }
 
+// Legacy returns 'true' if the MinIO server commandline was
+// provided with no ellipses pattern, those are considered
+// legacy deployments.
+func (l EndpointServerPools) Legacy() bool {
+	return len(l) == 1 && l[0].Legacy
+}
+
 // Add add pool endpoints
 func (l *EndpointServerPools) Add(zeps PoolEndpoints) error {
 	existSet := set.NewStringSet()
@@ -622,20 +643,33 @@ func CreateEndpoints(serverAddr string, foundLocal bool, args ...[]string) (Endp
 		}
 	}
 
-	// Check whether same path is not used in endpoints of a host on different port.
-	{
-		pathIPMap := make(map[string]set.StringSet)
-		for _, endpoint := range endpoints {
-			host := endpoint.Hostname()
-			hostIPSet, _ := getHostIP(host)
-			if IPSet, ok := pathIPMap[endpoint.Path]; ok {
-				if !IPSet.Intersection(hostIPSet).IsEmpty() {
-					return endpoints, setupType,
-						config.ErrInvalidErasureEndpoints(nil).Msg(fmt.Sprintf("path '%s' can not be served by different port on same address", endpoint.Path))
+	orchestrated := IsKubernetes() || IsDocker()
+	if !orchestrated {
+		// Check whether same path is not used in endpoints of a host on different port.
+		// Only verify this on baremetal setups, DNS is not available in orchestrated
+		// environments so we can't do much here.
+		{
+			pathIPMap := make(map[string]set.StringSet)
+			hostIPCache := make(map[string]set.StringSet)
+			for _, endpoint := range endpoints {
+				host := endpoint.Hostname()
+				hostIPSet, ok := hostIPCache[host]
+				if !ok {
+					hostIPSet, err = getHostIP(host)
+					if err != nil {
+						return endpoints, setupType, config.ErrInvalidErasureEndpoints(nil).Msg(fmt.Sprintf("host '%s' cannot resolve: %s", host, err))
+					}
+					hostIPCache[host] = hostIPSet
+				}
+				if IPSet, ok := pathIPMap[endpoint.Path]; ok {
+					if !IPSet.Intersection(hostIPSet).IsEmpty() {
+						return endpoints, setupType,
+							config.ErrInvalidErasureEndpoints(nil).Msg(fmt.Sprintf("same path '%s' can not be served by different port on same address", endpoint.Path))
+					}
+					pathIPMap[endpoint.Path] = IPSet.Union(hostIPSet)
+				} else {
+					pathIPMap[endpoint.Path] = hostIPSet
 				}
-				pathIPMap[endpoint.Path] = IPSet.Union(hostIPSet)
-			} else {
-				pathIPMap[endpoint.Path] = hostIPSet
 			}
 		}
 	}

cmd/endpoint_test.go

@@ -238,13 +238,18 @@ func TestCreateEndpoints(t *testing.T) {
 		{"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
-		{":1234", [][]string{{"/d1", "/d2", "/d3", "/d4"}}, ":1234",
+		{
+			":1234",
+			[][]string{{"/d1", "/d2", "/d3", "/d4"}},
+			":1234",
 			Endpoints{
 				Endpoint{URL: &url.URL{Path: mustAbs("/d1")}, IsLocal: true},
 				Endpoint{URL: &url.URL{Path: mustAbs("/d2")}, IsLocal: true},
 				Endpoint{URL: &url.URL{Path: mustAbs("/d3")}, IsLocal: true},
 				Endpoint{URL: &url.URL{Path: mustAbs("/d4")}, IsLocal: true},
-			}, ErasureSetupType, nil},
+			},
+			ErasureSetupType, nil,
+		},
 		// DistErasure Setup with URLEndpointType
 		{":9000", [][]string{{"http://localhost/d1", "http://localhost/d2", "http://localhost/d3", "http://localhost/d4"}}, ":9000", Endpoints{
 			Endpoint{URL: &url.URL{Scheme: "http", Host: "localhost", Path: "/d1"}, IsLocal: true},
@@ -350,12 +355,18 @@ func TestGetLocalPeer(t *testing.T) {
 		expectedResult string
 	}{
 		{[]string{"/d1", "/d2", "d3", "d4"}, "127.0.0.1:9000"},
-		{[]string{"http://localhost:9000/d1", "http://localhost:9000/d2", "http://example.org:9000/d3", "http://example.com:9000/d4"},
-			"localhost:9000"},
-		{[]string{"http://localhost:9000/d1", "http://example.org:9000/d2", "http://example.com:9000/d3", "http://example.net:9000/d4"},
-			"localhost:9000"},
-		{[]string{"http://localhost:9000/d1", "http://localhost:9001/d2", "http://localhost:9002/d3", "http://localhost:9003/d4"},
-			"localhost:9000"},
+		{
+			[]string{"http://localhost:9000/d1", "http://localhost:9000/d2", "http://example.org:9000/d3", "http://example.com:9000/d4"},
+			"localhost:9000",
+		},
+		{
+			[]string{"http://localhost:9000/d1", "http://example.org:9000/d2", "http://example.com:9000/d3", "http://example.net:9000/d4"},
+			"localhost:9000",
+		},
+		{
+			[]string{"http://localhost:9000/d1", "http://localhost:9001/d2", "http://localhost:9002/d3", "http://localhost:9003/d4"},
+			"localhost:9000",
+		},
 	}
 
 	for i, testCase := range testCases {

cmd/erasure-bucket.go

@@ -64,8 +64,7 @@ func (er erasureObjects) MakeBucketWithLocation(ctx context.Context, bucket stri
 		}, index)
 	}
 
-	writeQuorum := getWriteQuorum(len(storageDisks))
-	err := reduceWriteQuorumErrs(ctx, g.Wait(), bucketOpIgnoredErrs, writeQuorum)
+	err := reduceWriteQuorumErrs(ctx, g.Wait(), bucketOpIgnoredErrs, er.defaultWQuorum())
 	return toObjectErr(err, bucket)
 }
 
@@ -92,7 +91,7 @@ func (er erasureObjects) getBucketInfo(ctx context.Context, bucketName string) (
 	storageDisks := er.getDisks()
 
 	g := errgroup.WithNErrs(len(storageDisks))
-	var bucketsInfo = make([]BucketInfo, len(storageDisks))
+	bucketsInfo := make([]BucketInfo, len(storageDisks))
 	// Undo previous make bucket entry on all underlying storage disks.
 	for index := range storageDisks {
 		index := index
@@ -121,8 +120,7 @@ func (er erasureObjects) getBucketInfo(ctx context.Context, bucketName string) (
 	// reduce to one error based on read quorum.
 	// `nil` is deliberately passed for ignoredErrs
 	// because these errors were already ignored.
-	readQuorum := getReadQuorum(len(storageDisks))
-	return BucketInfo{}, reduceReadQuorumErrs(ctx, errs, nil, readQuorum)
+	return BucketInfo{}, reduceReadQuorumErrs(ctx, errs, nil, er.defaultRQuorum())
 }
 
 // GetBucketInfo - returns BucketInfo for a bucket.
@@ -167,8 +165,7 @@ func (er erasureObjects) DeleteBucket(ctx context.Context, bucket string, opts D
 		return nil
 	}
 
-	writeQuorum := getWriteQuorum(len(storageDisks))
-	err := reduceWriteQuorumErrs(ctx, dErrs, bucketOpIgnoredErrs, writeQuorum)
+	err := reduceWriteQuorumErrs(ctx, dErrs, bucketOpIgnoredErrs, er.defaultWQuorum())
 	if err == errErasureWriteQuorum && !opts.NoRecreate {
 		undoDeleteBucket(storageDisks, bucket)
 	}

cmd/erasure-coding.go

@@ -94,7 +94,7 @@ func (e *Erasure) EncodeData(ctx context.Context, data []byte) ([][]byte, error)
 // It only decodes the data blocks but does not verify them.
 // It returns an error if the decoding failed.
 func (e *Erasure) DecodeDataBlocks(data [][]byte) error {
-	var isZero = 0
+	isZero := 0
 	for _, b := range data {
 		if len(b) == 0 {
 			isZero++

cmd/erasure-common.go

@@ -22,6 +22,38 @@ import (
 	"sync"
 )
 
+func (er erasureObjects) getOnlineDisks() (newDisks []StorageAPI) {
+	disks := er.getDisks()
+	var wg sync.WaitGroup
+	var mu sync.Mutex
+	for _, i := range hashOrder(UTCNow().String(), len(disks)) {
+		i := i
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			if disks[i-1] == nil {
+				return
+			}
+			di, err := disks[i-1].DiskInfo(context.Background())
+			if err != nil || di.Healing {
+				// - Do not consume disks which are not reachable
+				//   unformatted or simply not accessible for some reason.
+				//
+				// - Do not consume disks which are being healed
+				//
+				// - Future: skip busy disks
+				return
+			}
+
+			mu.Lock()
+			newDisks = append(newDisks, disks[i-1])
+			mu.Unlock()
+		}()
+	}
+	wg.Wait()
+	return newDisks
+}
+
 func (er erasureObjects) getLoadBalancedLocalDisks() (newDisks []StorageAPI) {
 	disks := er.getDisks()
 	// Based on the random shuffling return back randomized disks.
@@ -50,7 +82,7 @@ func (er erasureObjects) getLoadBalancedDisks(optimized bool) []StorageAPI {
 
 	var wg sync.WaitGroup
 	var mu sync.Mutex
-	var newDisks = map[uint64][]StorageAPI{}
+	newDisks := map[uint64][]StorageAPI{}
 	// Based on the random shuffling return back randomized disks.
 	for _, i := range hashOrder(UTCNow().String(), len(disks)) {
 		i := i

cmd/erasure-decode.go

@@ -322,8 +322,8 @@ func (e Erasure) Heal(ctx context.Context, writers []io.Writer, readers []io.Rea
 			errs:        make([]error, len(writers)),
 		}
 
-		err = w.Write(ctx, bufs)
-		if err != nil {
+		if err = w.Write(ctx, bufs); err != nil {
+			logger.LogIf(ctx, err)
 			return err
 		}
 	}

cmd/erasure-encode.go

@@ -52,7 +52,10 @@ func (p *parallelWriter) Write(ctx context.Context, blocks [][]byte) error {
 			if p.errs[i] == nil {
 				if n != len(blocks[i]) {
 					p.errs[i] = io.ErrShortWrite
+					p.writers[i] = nil
 				}
+			} else {
+				p.writers[i] = nil
 			}
 		}(i)
 	}
@@ -93,6 +96,7 @@ func (e *Erasure) Encode(ctx context.Context, src io.Reader, writers []io.Writer
 		blocks, err = e.EncodeData(ctx, buf[:n])
 		if err != nil {
 			logger.LogIf(ctx, err)
+
 			return 0, err
 		}
 

cmd/erasure-healing-common.go

@@ -26,29 +26,59 @@ import (
 )
 
 // commonTime returns a maximally occurring time from a list of time.
-func commonTime(modTimes []time.Time) (modTime time.Time) {
-	timeOccurenceMap := make(map[int64]int, len(modTimes))
+func commonTimeAndOccurence(times []time.Time, group time.Duration) (maxTime time.Time, maxima int) {
+	timeOccurenceMap := make(map[int64]int, len(times))
+	groupNano := group.Nanoseconds()
 	// Ignore the uuid sentinel and count the rest.
-	for _, t := range modTimes {
+	for _, t := range times {
 		if t.Equal(timeSentinel) {
 			continue
 		}
-		timeOccurenceMap[t.UnixNano()]++
+		nano := t.UnixNano()
+		if group > 0 {
+			for k := range timeOccurenceMap {
+				if k == nano {
+					// We add to ourself later
+					continue
+				}
+				diff := k - nano
+				if diff < 0 {
+					diff = -diff
+				}
+				// We are within the limit
+				if diff < groupNano {
+					timeOccurenceMap[k]++
+				}
+			}
+		}
+		// Add ourself...
+		timeOccurenceMap[nano]++
 	}
 
-	var maxima int // Counter for remembering max occurrence of elements.
+	maxima = 0 // Counter for remembering max occurrence of elements.
+	latest := int64(0)
 
 	// Find the common cardinality from previously collected
 	// occurrences of elements.
 	for nano, count := range timeOccurenceMap {
-		t := time.Unix(0, nano).UTC()
-		if count > maxima || (count == maxima && t.After(modTime)) {
+		if count < maxima {
+			continue
+		}
+
+		// We are at or above maxima
+		if count > maxima || nano > latest {
 			maxima = count
-			modTime = t
+			latest = nano
 		}
 	}
 
-	// Return the collected common modTime.
+	// Return the collected common max time, with maxima
+	return time.Unix(0, latest).UTC(), maxima
+}
+
+// commonTime returns a maximally occurring time from a list of time.
+func commonTime(modTimes []time.Time) (modTime time.Time) {
+	modTime, _ = commonTimeAndOccurence(modTimes, 0)
 	return modTime
 }
 
@@ -88,6 +118,19 @@ func filterOnlineDisksInplace(fi FileInfo, partsMetadata []FileInfo, onlineDisks
 	}
 }
 
+// Extracts list of disk mtimes from FileInfo slice and returns, skips
+// slice elements that have errors.
+func listObjectDiskMtimes(partsMetadata []FileInfo) (diskMTimes []time.Time) {
+	diskMTimes = bootModtimes(len(partsMetadata))
+	for index, metadata := range partsMetadata {
+		if metadata.IsValid() {
+			// Once the file is found, save the disk mtime saved on disk.
+			diskMTimes[index] = metadata.DiskMTime
+		}
+	}
+	return diskMTimes
+}
+
 // Notes:
 // There are 5 possible states a disk could be in,
 // 1. __online__             - has the latest copy of xl.meta - returned by listOnlineDisks
@@ -182,8 +225,15 @@ func getLatestFileInfo(ctx context.Context, partsMetadata []FileInfo, errs []err
 // - slice of errors about the state of data files on disk - can have
 //   a not-found error or a hash-mismatch error.
 func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetadata []FileInfo,
-	errs []error, latestMeta FileInfo,
-	bucket, object string, scanMode madmin.HealScanMode) ([]StorageAPI, []error) {
+	errs []error, latestMeta FileInfo, bucket, object string,
+	scanMode madmin.HealScanMode) ([]StorageAPI, []error, time.Time,
+) {
+	var diskMTime time.Time
+	var shardFix bool
+	if !latestMeta.DataShardFixed() {
+		diskMTime = pickValidDiskTimeWithQuorum(partsMetadata,
+			latestMeta.Erasure.DataBlocks)
+	}
 
 	availableDisks := make([]StorageAPI, len(onlineDisks))
 	dataErrs := make([]error, len(onlineDisks))
@@ -234,6 +284,8 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 
 		if erasureDistributionReliable {
 			if !meta.IsValid() {
+				partsMetadata[i] = FileInfo{}
+				dataErrs[i] = errFileCorrupt
 				continue
 			}
 
@@ -256,6 +308,23 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 			}
 		}
 
+		if !diskMTime.Equal(timeSentinel) && !diskMTime.IsZero() {
+			if !partsMetadata[i].AcceptableDelta(diskMTime, shardDiskTimeDelta) {
+				// not with in acceptable delta, skip.
+				// If disk mTime mismatches it is considered outdated
+				// https://github.com/minio/minio/pull/13803
+				//
+				// This check only is active if we could find maximally
+				// occurring disk mtimes that are somewhat same across
+				// the quorum. Allowing to skip those shards which we
+				// might think are wrong.
+				shardFix = true
+				partsMetadata[i] = FileInfo{}
+				dataErrs[i] = errFileCorrupt
+				continue
+			}
+		}
+
 		// Always check data, if we got it.
 		if (len(meta.Data) > 0 || meta.Size == 0) && len(meta.Parts) > 0 {
 			checksumInfo := meta.Erasure.GetChecksumInfo(meta.Parts[0].Number)
@@ -298,5 +367,10 @@ func disksWithAllParts(ctx context.Context, onlineDisks []StorageAPI, partsMetad
 		}
 	}
 
-	return availableDisks, dataErrs
+	if shardFix {
+		// Only when shard is fixed return an appropriate disk mtime value.
+		return availableDisks, dataErrs, diskMTime
+	} // else return timeSentinel for disk mtime
+
+	return availableDisks, dataErrs, timeSentinel
 }

cmd/erasure-healing-common_test.go

@@ -251,7 +251,8 @@ func TestListOnlineDisks(t *testing.T) {
 				t.Fatalf("Expected modTime to be equal to %v but was found to be %v",
 					test.expectedTime, modTime)
 			}
-			availableDisks, newErrs := disksWithAllParts(ctx, onlineDisks, partsMetadata, test.errs, fi, bucket, object, madmin.HealDeepScan)
+			availableDisks, newErrs, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
+				test.errs, fi, bucket, object, madmin.HealDeepScan)
 			test.errs = newErrs
 
 			if test._tamperBackend != noTamper {
@@ -433,7 +434,8 @@ func TestListOnlineDisksSmallObjects(t *testing.T) {
 					test.expectedTime, modTime)
 			}
 
-			availableDisks, newErrs := disksWithAllParts(ctx, onlineDisks, partsMetadata, test.errs, fi, bucket, object, madmin.HealDeepScan)
+			availableDisks, newErrs, _ := disksWithAllParts(ctx, onlineDisks, partsMetadata,
+				test.errs, fi, bucket, object, madmin.HealDeepScan)
 			test.errs = newErrs
 
 			if test._tamperBackend != noTamper {
@@ -494,7 +496,8 @@ func TestDisksWithAllParts(t *testing.T) {
 
 	erasureDisks, _ = listOnlineDisks(erasureDisks, partsMetadata, errs)
 
-	filteredDisks, errs := disksWithAllParts(ctx, erasureDisks, partsMetadata, errs, fi, bucket, object, madmin.HealDeepScan)
+	filteredDisks, errs, _ := disksWithAllParts(ctx, erasureDisks, partsMetadata,
+		errs, fi, bucket, object, madmin.HealDeepScan)
 
 	if len(filteredDisks) != len(erasureDisks) {
 		t.Errorf("Unexpected number of disks: %d", len(filteredDisks))
@@ -515,7 +518,8 @@ func TestDisksWithAllParts(t *testing.T) {
 	partsMetadata[0].ModTime = partsMetadata[0].ModTime.Add(-1 * time.Hour)
 
 	errs = make([]error, len(erasureDisks))
-	filteredDisks, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata, errs, fi, bucket, object, madmin.HealDeepScan)
+	filteredDisks, _, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata,
+		errs, fi, bucket, object, madmin.HealDeepScan)
 
 	if len(filteredDisks) != len(erasureDisks) {
 		t.Errorf("Unexpected number of disks: %d", len(filteredDisks))
@@ -535,7 +539,8 @@ func TestDisksWithAllParts(t *testing.T) {
 	partsMetadata[1].DataDir = "foo-random"
 
 	errs = make([]error, len(erasureDisks))
-	filteredDisks, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata, errs, fi, bucket, object, madmin.HealDeepScan)
+	filteredDisks, _, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata,
+		errs, fi, bucket, object, madmin.HealDeepScan)
 
 	if len(filteredDisks) != len(erasureDisks) {
 		t.Errorf("Unexpected number of disks: %d", len(filteredDisks))
@@ -571,7 +576,8 @@ func TestDisksWithAllParts(t *testing.T) {
 	}
 
 	errs = make([]error, len(erasureDisks))
-	filteredDisks, errs = disksWithAllParts(ctx, erasureDisks, partsMetadata, errs, fi, bucket, object, madmin.HealDeepScan)
+	filteredDisks, errs, _ = disksWithAllParts(ctx, erasureDisks, partsMetadata,
+		errs, fi, bucket, object, madmin.HealDeepScan)
 
 	if len(filteredDisks) != len(erasureDisks) {
 		t.Errorf("Unexpected number of disks: %d", len(filteredDisks))

cmd/erasure-healing.go

@@ -24,12 +24,38 @@ import (
 	"fmt"
 	"io"
 	"sync"
+	"time"
 
 	"github.com/minio/madmin-go"
 	"github.com/minio/minio/internal/logger"
 	"github.com/minio/minio/internal/sync/errgroup"
 )
 
+const reservedMetadataPrefixLowerDataShardFix = ReservedMetadataPrefixLower + "data-shard-fix"
+
+// AcceptableDelta returns 'true' if the fi.DiskMTime is under
+// acceptable delta of "delta" duration with maxTime.
+//
+// This code is primarily used for heuristic detection of
+// incorrect shards, as per https://github.com/minio/minio/pull/13803
+//
+// This check only is active if we could find maximally
+// occurring disk mtimes that are somewhat same across
+// the quorum. Allowing to skip those shards which we
+// might think are wrong.
+func (fi FileInfo) AcceptableDelta(maxTime time.Time, delta time.Duration) bool {
+	diff := maxTime.Sub(fi.DiskMTime)
+	if diff < 0 {
+		diff = -diff
+	}
+	return diff < delta
+}
+
+// DataShardFixed - data shard fixed?
+func (fi FileInfo) DataShardFixed() bool {
+	return fi.Metadata[reservedMetadataPrefixLowerDataShardFix] == "true"
+}
+
 // Heals a bucket if it doesn't exist on one of the disks, additionally
 // also heals the missing entries for bucket metadata files
 // `policy.json, notification.xml, listeners.json`.
@@ -220,10 +246,15 @@ func shouldHealObjectOnDisk(erErr, dataErr error, meta FileInfo, latestMeta File
 	switch {
 	case errors.Is(erErr, errFileNotFound) || errors.Is(erErr, errFileVersionNotFound):
 		return true
-	case errors.Is(erErr, errCorruptedFormat):
+	case errors.Is(erErr, errFileCorrupt):
 		return true
 	}
 	if erErr == nil {
+		if meta.XLV1 {
+			// Legacy means heal always
+			// always check first.
+			return true
+		}
 		if !meta.Deleted && !meta.IsRemote() {
 			// If xl.meta was read fine but there may be problem with the part.N files.
 			if IsErr(dataErr, []error{
@@ -234,19 +265,7 @@ func shouldHealObjectOnDisk(erErr, dataErr error, meta FileInfo, latestMeta File
 				return true
 			}
 		}
-		if !latestMeta.MetadataEquals(meta) {
-			return true
-		}
-		if !latestMeta.TransitionInfoEquals(meta) {
-			return true
-		}
-		if !latestMeta.ReplicationInfoEquals(meta) {
-			return true
-		}
-		if !latestMeta.ModTime.Equal(meta.ModTime) {
-			return true
-		}
-		if meta.XLV1 {
+		if !latestMeta.Equals(meta) {
 			return true
 		}
 	}
@@ -267,12 +286,10 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 
 	// Initialize heal result object
 	result = madmin.HealResultItem{
-		Type:         madmin.HealItemObject,
-		Bucket:       bucket,
-		Object:       object,
-		DiskCount:    len(storageDisks),
-		ParityBlocks: er.defaultParityCount,
-		DataBlocks:   len(storageDisks) - er.defaultParityCount,
+		Type:      madmin.HealItemObject,
+		Bucket:    bucket,
+		Object:    object,
+		DiskCount: len(storageDisks),
 	}
 
 	if !opts.NoLock {
@@ -287,18 +304,27 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 
 	// Re-read when we have lock...
 	partsMetadata, errs := readAllFileInfo(ctx, storageDisks, bucket, object, versionID, true)
-
-	if _, err = getLatestFileInfo(ctx, partsMetadata, errs); err != nil {
-		return er.purgeObjectDangling(ctx, bucket, object, versionID, partsMetadata, errs, []error{}, opts)
+	if isAllNotFound(errs) {
+		// Nothing to do, file is already gone.
+		return er.defaultHealResult(FileInfo{}, storageDisks, storageEndpoints,
+			errs, bucket, object, versionID), nil
 	}
 
-	// List of disks having latest version of the object er.meta
+	readQuorum, _, err := objectQuorumFromMeta(ctx, partsMetadata, errs, er.defaultParityCount)
+	if err != nil {
+		return er.purgeObjectDangling(ctx, bucket, object, versionID, partsMetadata, errs, nil, opts)
+	}
+
+	result.ParityBlocks = result.DiskCount - readQuorum
+	result.DataBlocks = readQuorum
+
+	// List of disks having latest version of the object xl.meta
 	// (by modtime).
 	onlineDisks, modTime := listOnlineDisks(storageDisks, partsMetadata, errs)
 
 	// Latest FileInfo for reference. If a valid metadata is not
 	// present, it is as good as object not found.
-	latestMeta, err := pickValidFileInfo(ctx, partsMetadata, modTime, result.DataBlocks)
+	latestMeta, err := pickValidFileInfo(ctx, partsMetadata, modTime, readQuorum)
 	if err != nil {
 		return result, toObjectErr(err, bucket, object, versionID)
 	}
@@ -312,7 +338,7 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 	// used here for reconstruction. This is done to ensure that
 	// we do not skip drives that have inconsistent metadata to be
 	// skipped from purging when they are stale.
-	availableDisks, dataErrs := disksWithAllParts(ctx, onlineDisks, partsMetadata,
+	availableDisks, dataErrs, diskMTime := disksWithAllParts(ctx, onlineDisks, partsMetadata,
 		errs, latestMeta, bucket, object, scanMode)
 
 	// Loop to find number of disks with valid data, per-drive
@@ -330,10 +356,6 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 			// If data is sane on any one disk, we can
 			// extract the correct object size.
 			result.ObjectSize = partsMetadata[i].Size
-			if partsMetadata[i].Erasure.ParityBlocks > 0 && partsMetadata[i].Erasure.DataBlocks > 0 {
-				result.ParityBlocks = partsMetadata[i].Erasure.ParityBlocks
-				result.DataBlocks = partsMetadata[i].Erasure.DataBlocks
-			}
 		case errs[i] == errDiskNotFound, dataErrs[i] == errDiskNotFound:
 			driveState = madmin.DriveStateOffline
 		case errs[i] == errFileNotFound, errs[i] == errFileVersionNotFound, errs[i] == errVolumeNotFound:
@@ -373,18 +395,14 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 	}
 
 	if isAllNotFound(errs) {
-		err = toObjectErr(errFileNotFound, bucket, object)
-		if versionID != "" {
-			err = toObjectErr(errFileVersionNotFound, bucket, object, versionID)
-		}
 		// File is fully gone, fileInfo is empty.
 		return er.defaultHealResult(FileInfo{}, storageDisks, storageEndpoints, errs,
-			bucket, object, versionID), err
+			bucket, object, versionID), nil
 	}
 
 	// If less than read quorum number of disks have all the parts
 	// of the data, we can't reconstruct the erasure-coded data.
-	if numAvailableDisks < result.DataBlocks {
+	if numAvailableDisks < readQuorum {
 		return er.purgeObjectDangling(ctx, bucket, object, versionID, partsMetadata, errs, dataErrs, opts)
 	}
 
@@ -444,9 +462,6 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 	copyPartsMetadata = shufflePartsMetadata(copyPartsMetadata, latestMeta.Erasure.Distribution)
 
 	if !latestMeta.Deleted && !latestMeta.IsRemote() {
-		result.DataBlocks = latestMeta.Erasure.DataBlocks
-		result.ParityBlocks = latestMeta.Erasure.ParityBlocks
-
 		// Heal each part. erasureHealFile() will write the healed
 		// part to .minio/tmp/uuid/ which needs to be renamed later to
 		// the final location.
@@ -525,14 +540,14 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 
 			// If all disks are having errors, we give up.
 			if disksToHealCount == 0 {
-				return result, fmt.Errorf("all disks had write errors, unable to heal")
+				return result, fmt.Errorf("all disks had write errors, unable to heal %s/%s", bucket, object)
 			}
 
 		}
 
 	}
 
-	defer er.deleteObject(context.Background(), minioMetaTmpBucket, tmpID, len(storageDisks)/2+1)
+	defer er.renameAll(context.Background(), minioMetaTmpBucket, tmpID)
 
 	// Rename from tmp location to the actual location.
 	for i, disk := range outDatedDisks {
@@ -562,6 +577,20 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
 		}
 	}
 
+	if !diskMTime.Equal(timeSentinel) && !diskMTime.IsZero() {
+		// Update metadata to indicate special fix.
+		_, err = er.PutObjectMetadata(ctx, bucket, object, ObjectOptions{
+			NoLock: true,
+			UserDefined: map[string]string{
+				reservedMetadataPrefixLowerDataShardFix: "true",
+				// another reserved metadata to capture original disk-mtime
+				// captured for this version of the object, to be used
+				// possibly in future to heal other versions if possible.
+				ReservedMetadataPrefixLower + "disk-mtime": diskMTime.String(),
+			},
+		})
+	}
+
 	// Set the size of the object in the heal result
 	result.ObjectSize = latestMeta.Size
 
@@ -630,7 +659,7 @@ func (er erasureObjects) healObjectDir(ctx context.Context, bucket, object strin
 	}
 	if dryRun || danglingObject || isAllNotFound(errs) {
 		// Nothing to do, file is already gone.
-		return hr, toObjectErr(errFileNotFound, bucket, object)
+		return hr, nil
 	}
 	for i, err := range errs {
 		if err == errVolumeNotFound || err == errFileNotFound {
@@ -735,8 +764,15 @@ func statAllDirs(ctx context.Context, storageDisks []StorageAPI, bucket, prefix
 // A 0 length slice will always return false.
 func isAllNotFound(errs []error) bool {
 	for _, err := range errs {
-		if errors.Is(err, errFileNotFound) || errors.Is(err, errVolumeNotFound) || errors.Is(err, errFileVersionNotFound) {
-			continue
+		if err != nil {
+			switch err.Error() {
+			case errFileNotFound.Error():
+				fallthrough
+			case errVolumeNotFound.Error():
+				fallthrough
+			case errFileVersionNotFound.Error():
+				continue
+			}
 		}
 		return false
 	}
@@ -768,29 +804,28 @@ func isObjectDirDangling(errs []error) (ok bool) {
 }
 
 func (er erasureObjects) purgeObjectDangling(ctx context.Context, bucket, object, versionID string,
-	metaArr []FileInfo, errs []error, dataErrs []error, opts madmin.HealOpts) (madmin.HealResultItem, error) {
-
+	metaArr []FileInfo, errs []error, dataErrs []error, opts madmin.HealOpts) (madmin.HealResultItem, error,
+) {
 	storageDisks := er.getDisks()
 	storageEndpoints := er.getEndpoints()
 	// Check if the object is dangling, if yes and user requested
 	// remove we simply delete it from namespace.
 	m, ok := isObjectDangling(metaArr, errs, dataErrs)
 	if ok {
-		parityBlocks := m.Erasure.ParityBlocks
-		if m.Erasure.ParityBlocks == 0 {
-			parityBlocks = er.defaultParityCount
-		}
-		dataBlocks := m.Erasure.DataBlocks
-		if m.Erasure.DataBlocks == 0 {
-			dataBlocks = len(storageDisks) - parityBlocks
+		parityBlocks := er.defaultParityCount
+		dataBlocks := len(storageDisks) - parityBlocks
+		if m.IsValid() {
+			parityBlocks = m.Erasure.ParityBlocks
+			dataBlocks = m.Erasure.DataBlocks
 		}
+
 		writeQuorum := dataBlocks
 		if dataBlocks == parityBlocks {
 			writeQuorum++
 		}
+
 		var err error
-		var returnNotFound bool
-		if !opts.DryRun && opts.Remove {
+		if opts.Remove {
 			err = er.deleteObjectVersion(ctx, bucket, object, writeQuorum, FileInfo{
 				VersionID: versionID,
 			}, false)
@@ -806,19 +841,10 @@ func (er erasureObjects) purgeObjectDangling(ctx context.Context, bucket, object
 				m.Size = 0
 			}
 
-			// Delete successfully purged dangling content, return ObjectNotFound/VersionNotFound instead.
-			if countErrs(errs, nil) == len(errs) {
-				returnNotFound = true
-			}
-		}
-		if returnNotFound {
-			err = toObjectErr(errFileNotFound, bucket, object)
-			if versionID != "" {
-				err = toObjectErr(errFileVersionNotFound, bucket, object, versionID)
-			}
-			return er.defaultHealResult(m, storageDisks, storageEndpoints,
-				errs, bucket, object, versionID), err
+			return er.defaultHealResult(FileInfo{}, storageDisks, storageEndpoints,
+				errs, bucket, object, versionID), nil
 		}
+
 		return er.defaultHealResult(m, storageDisks, storageEndpoints,
 			errs, bucket, object, versionID), toObjectErr(err, bucket, object, versionID)
 	}
@@ -835,54 +861,64 @@ func (er erasureObjects) purgeObjectDangling(ctx context.Context, bucket, object
 // files is lesser than number of data blocks.
 func isObjectDangling(metaArr []FileInfo, errs []error, dataErrs []error) (validMeta FileInfo, ok bool) {
 	// We can consider an object data not reliable
-	// when er.meta is not found in read quorum disks.
-	// or when er.meta is not readable in read quorum disks.
-	var notFoundErasureMeta, corruptedErasureMeta int
-	for _, readErr := range errs {
-		if errors.Is(readErr, errFileNotFound) || errors.Is(readErr, errFileVersionNotFound) {
-			notFoundErasureMeta++
-		} else if errors.Is(readErr, errCorruptedFormat) {
-			corruptedErasureMeta++
-		}
-	}
-	var notFoundParts int
-	for i := range dataErrs {
-		// Only count part errors, if the error is not
-		// same as er.meta error. This is to avoid
-		// double counting when both parts and er.meta
-		// are not available.
-		if errs[i] != dataErrs[i] {
-			if IsErr(dataErrs[i], []error{
-				errFileNotFound,
-				errFileVersionNotFound,
-			}...) {
-				notFoundParts++
+	// when xl.meta is not found in read quorum disks.
+	// or when xl.meta is not readable in read quorum disks.
+	danglingErrsCount := func(cerrs []error) (int, int) {
+		var (
+			notFoundCount  int
+			corruptedCount int
+		)
+		for _, readErr := range cerrs {
+			if errors.Is(readErr, errFileNotFound) || errors.Is(readErr, errFileVersionNotFound) {
+				notFoundCount++
+			} else if errors.Is(readErr, errFileCorrupt) {
+				corruptedCount++
 			}
 		}
+		return notFoundCount, corruptedCount
 	}
 
+	ndataErrs := make([]error, len(dataErrs))
+	for i := range dataErrs {
+		if errs[i] != dataErrs[i] {
+			// Only count part errors, if the error is not
+			// same as xl.meta error. This is to avoid
+			// double counting when both parts and xl.meta
+			// are not available.
+			ndataErrs[i] = dataErrs[i]
+		}
+	}
+
+	notFoundMetaErrs, corruptedMetaErrs := danglingErrsCount(errs)
+	notFoundPartsErrs, corruptedPartsErrs := danglingErrsCount(ndataErrs)
+
 	for _, m := range metaArr {
-		if !m.IsValid() {
-			continue
+		if m.IsValid() {
+			validMeta = m
+			break
 		}
-		validMeta = m
-		break
 	}
 
-	if validMeta.Deleted || validMeta.IsRemote() {
-		// notFoundParts is ignored since a
+	if !validMeta.IsValid() {
+		// We have no idea what this file is, leave it as is.
+		return validMeta, false
+	}
+
+	if validMeta.Deleted {
+		// notFoundPartsErrs is ignored since
 		// - delete marker does not have any parts
-		// - transition status of complete has no parts
-		return validMeta, corruptedErasureMeta+notFoundErasureMeta > len(errs)/2
+		return validMeta, corruptedMetaErrs+notFoundMetaErrs > len(errs)/2
 	}
 
-	// We couldn't find any valid meta we are indeed corrupted, return true right away.
-	if validMeta.Erasure.DataBlocks == 0 {
-		return validMeta, true
+	totalErrs := notFoundMetaErrs + corruptedMetaErrs + notFoundPartsErrs + corruptedPartsErrs
+	if validMeta.IsRemote() {
+		// notFoundPartsErrs is ignored since
+		// - transition status of complete has no parts
+		totalErrs = notFoundMetaErrs + corruptedMetaErrs
 	}
 
 	// We have valid meta, now verify if we have enough files with parity blocks.
-	return validMeta, corruptedErasureMeta+notFoundErasureMeta+notFoundParts > validMeta.Erasure.ParityBlocks
+	return validMeta, totalErrs > validMeta.Erasure.ParityBlocks
 }
 
 // HealObject - heal the given object, automatically deletes the object if stale/corrupted if `remove` is true.
@@ -921,13 +957,9 @@ func (er erasureObjects) HealObject(ctx context.Context, bucket, object, version
 	// This allows to quickly check if all is ok or all are missing.
 	_, errs := readAllFileInfo(healCtx, storageDisks, bucket, object, versionID, false)
 	if isAllNotFound(errs) {
-		err = toObjectErr(errFileNotFound, bucket, object)
-		if versionID != "" {
-			err = toObjectErr(errFileVersionNotFound, bucket, object, versionID)
-		}
 		// Nothing to do, file is already gone.
 		return er.defaultHealResult(FileInfo{}, storageDisks, storageEndpoints,
-			errs, bucket, object, versionID), err
+			errs, bucket, object, versionID), nil
 	}
 
 	// Heal the object.

cmd/erasure-healing_test.go

@@ -22,6 +22,7 @@ import (
 	"context"
 	"crypto/rand"
 	"crypto/sha256"
+	"errors"
 	"io"
 	"os"
 	"path"
@@ -33,6 +34,192 @@ import (
 	"github.com/minio/madmin-go"
 )
 
+// Tests isObjectDangling function
+func TestIsObjectDangling(t *testing.T) {
+	fi := newFileInfo("test-object", 2, 2)
+	fi.Erasure.Index = 1
+
+	testCases := []struct {
+		name             string
+		metaArr          []FileInfo
+		errs             []error
+		dataErrs         []error
+		expectedMeta     FileInfo
+		expectedDangling bool
+	}{
+		{
+			name: "FileInfoExists-case1",
+			metaArr: []FileInfo{
+				{},
+				{},
+				fi,
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errDiskNotFound,
+				nil,
+				nil,
+			},
+			dataErrs:         nil,
+			expectedMeta:     fi,
+			expectedDangling: false,
+		},
+		{
+			name: "FileInfoExists-case2",
+			metaArr: []FileInfo{
+				{},
+				{},
+				fi,
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileNotFound,
+				nil,
+				nil,
+			},
+			dataErrs:         nil,
+			expectedMeta:     fi,
+			expectedDangling: false,
+		},
+		{
+			name: "FileInfoUndecided-case1",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errDiskNotFound,
+				errDiskNotFound,
+				nil,
+			},
+			dataErrs:         nil,
+			expectedMeta:     fi,
+			expectedDangling: false,
+		},
+		{
+			name:    "FileInfoUndecided-case2",
+			metaArr: []FileInfo{},
+			errs: []error{
+				errFileNotFound,
+				errDiskNotFound,
+				errDiskNotFound,
+				errFileNotFound,
+			},
+			dataErrs:         nil,
+			expectedMeta:     FileInfo{},
+			expectedDangling: false,
+		},
+		{
+			name:    "FileInfoUndecided-case3(file deleted)",
+			metaArr: []FileInfo{},
+			errs: []error{
+				errFileNotFound,
+				errFileNotFound,
+				errFileNotFound,
+				errFileNotFound,
+			},
+			dataErrs:         nil,
+			expectedMeta:     FileInfo{},
+			expectedDangling: false,
+		},
+		{
+			name: "FileInfoDecided-case1",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileNotFound,
+				errFileNotFound,
+				nil,
+			},
+			dataErrs:         nil,
+			expectedMeta:     fi,
+			expectedDangling: true,
+		},
+		{
+			name: "FileInfoDecided-case2",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				fi,
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileCorrupt,
+				errFileCorrupt,
+				nil,
+			},
+			dataErrs:         nil,
+			expectedMeta:     fi,
+			expectedDangling: true,
+		},
+		{
+			name: "FileInfoDecided-case2-delete-marker",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				{Deleted: true},
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileCorrupt,
+				errFileCorrupt,
+				nil,
+			},
+			dataErrs:         nil,
+			expectedMeta:     FileInfo{Deleted: true},
+			expectedDangling: true,
+		},
+		{
+			name: "FileInfoDecided-case2-(duplicate data errors)",
+			metaArr: []FileInfo{
+				{},
+				{},
+				{},
+				{Deleted: true},
+			},
+			errs: []error{
+				errFileNotFound,
+				errFileCorrupt,
+				errFileCorrupt,
+				nil,
+			},
+			dataErrs: []error{
+				errFileNotFound,
+				errFileCorrupt,
+				nil,
+				nil,
+			},
+			expectedMeta:     FileInfo{Deleted: true},
+			expectedDangling: true,
+		},
+		// Add new cases as seen
+	}
+	for _, testCase := range testCases {
+		testCase := testCase
+		t.Run(testCase.name, func(t *testing.T) {
+			gotMeta, dangling := isObjectDangling(testCase.metaArr, testCase.errs, testCase.dataErrs)
+			if !gotMeta.Equals(testCase.expectedMeta) {
+				t.Errorf("Expected %#v, got %#v", testCase.expectedMeta, gotMeta)
+			}
+			if dangling != testCase.expectedDangling {
+				t.Errorf("Expected dangling %t, got %t", testCase.expectedDangling, dangling)
+			}
+		})
+	}
+}
+
 // Tests both object and bucket healing.
 func TestHealing(t *testing.T) {
 	ctx, cancel := context.WithCancel(context.Background())
@@ -93,16 +280,16 @@ func TestHealing(t *testing.T) {
 	}
 
 	// After heal the meta file should be as expected.
-	if !reflect.DeepEqual(fileInfoPreHeal, fileInfoPostHeal) {
+	if !fileInfoPreHeal.Equals(fileInfoPostHeal) {
 		t.Fatal("HealObject failed")
 	}
 
-	err = os.RemoveAll(path.Join(fsDirs[0], bucket, object, "er.meta"))
+	err = os.RemoveAll(path.Join(fsDirs[0], bucket, object, "xl.meta"))
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	// Write er.meta with different modtime to simulate the case where a disk had
+	// Write xl.meta with different modtime to simulate the case where a disk had
 	// gone down when an object was replaced by a new object.
 	fileInfoOutDated := fileInfoPreHeal
 	fileInfoOutDated.ModTime = time.Now()
@@ -122,7 +309,7 @@ func TestHealing(t *testing.T) {
 	}
 
 	// After heal the meta file should be as expected.
-	if !reflect.DeepEqual(fileInfoPreHeal, fileInfoPostHeal) {
+	if !fileInfoPreHeal.Equals(fileInfoPostHeal) {
 		t.Fatal("HealObject failed")
 	}
 
@@ -328,6 +515,298 @@ func TestHealingDanglingObject(t *testing.T) {
 	}
 }
 
+func TestHealCorrectQuorum(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	resetGlobalHealState()
+	defer resetGlobalHealState()
+
+	nDisks := 32
+	fsDirs, err := getRandomDisks(nDisks)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	defer removeRoots(fsDirs)
+
+	pools := mustGetPoolEndpoints(fsDirs[:16]...)
+	pools = append(pools, mustGetPoolEndpoints(fsDirs[16:]...)...)
+
+	// Everything is fine, should return nil
+	objLayer, _, err := initObjectLayer(ctx, pools)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	bucket := getRandomBucketName()
+	object := getRandomObjectName()
+	data := bytes.Repeat([]byte("a"), 5*1024*1024)
+	var opts ObjectOptions
+
+	err = objLayer.MakeBucketWithLocation(ctx, bucket, BucketOptions{})
+	if err != nil {
+		t.Fatalf("Failed to make a bucket - %v", err)
+	}
+
+	// Create an object with multiple parts uploaded in decreasing
+	// part number.
+	uploadID, err := objLayer.NewMultipartUpload(ctx, bucket, object, opts)
+	if err != nil {
+		t.Fatalf("Failed to create a multipart upload - %v", err)
+	}
+
+	var uploadedParts []CompletePart
+	for _, partID := range []int{2, 1} {
+		pInfo, err1 := objLayer.PutObjectPart(ctx, bucket, object, uploadID, partID, mustGetPutObjReader(t, bytes.NewReader(data), int64(len(data)), "", ""), opts)
+		if err1 != nil {
+			t.Fatalf("Failed to upload a part - %v", err1)
+		}
+		uploadedParts = append(uploadedParts, CompletePart{
+			PartNumber: pInfo.PartNumber,
+			ETag:       pInfo.ETag,
+		})
+	}
+
+	_, err = objLayer.CompleteMultipartUpload(ctx, bucket, object, uploadID, uploadedParts, ObjectOptions{})
+	if err != nil {
+		t.Fatalf("Failed to complete multipart upload - %v", err)
+	}
+
+	cfgFile := pathJoin(bucketMetaPrefix, bucket, ".test.bin")
+	if err = saveConfig(ctx, objLayer, cfgFile, data); err != nil {
+		t.Fatal(err)
+	}
+
+	hopts := madmin.HealOpts{
+		DryRun:   false,
+		Remove:   true,
+		ScanMode: madmin.HealNormalScan,
+	}
+
+	// Test 1: Remove the object backend files from the first disk.
+	z := objLayer.(*erasureServerPools)
+	for _, set := range z.serverPools {
+		er := set.sets[0]
+		erasureDisks := er.getDisks()
+
+		fileInfos, errs := readAllFileInfo(ctx, erasureDisks, bucket, object, "", false)
+		nfi, err := getLatestFileInfo(ctx, fileInfos, errs)
+		if errors.Is(err, errFileNotFound) {
+			continue
+		}
+		if err != nil {
+			t.Fatalf("Failed to getLatestFileInfo - %v", err)
+		}
+
+		for i := 0; i < nfi.Erasure.ParityBlocks; i++ {
+			erasureDisks[i].Delete(context.Background(), bucket, pathJoin(object, xlStorageFormatFile), false)
+		}
+
+		// Try healing now, it should heal the content properly.
+		_, err = objLayer.HealObject(ctx, bucket, object, "", hopts)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		fileInfos, errs = readAllFileInfo(ctx, erasureDisks, bucket, object, "", false)
+		if countErrs(errs, nil) != len(fileInfos) {
+			t.Fatal("Expected all xl.meta healed, but partial heal detected")
+		}
+
+		fileInfos, errs = readAllFileInfo(ctx, erasureDisks, minioMetaBucket, cfgFile, "", false)
+		nfi, err = getLatestFileInfo(ctx, fileInfos, errs)
+		if errors.Is(err, errFileNotFound) {
+			continue
+		}
+		if err != nil {
+			t.Fatalf("Failed to getLatestFileInfo - %v", err)
+		}
+
+		for i := 0; i < nfi.Erasure.ParityBlocks; i++ {
+			erasureDisks[i].Delete(context.Background(), minioMetaBucket, pathJoin(cfgFile, xlStorageFormatFile), false)
+		}
+
+		// Try healing now, it should heal the content properly.
+		_, err = objLayer.HealObject(ctx, minioMetaBucket, cfgFile, "", hopts)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		fileInfos, errs = readAllFileInfo(ctx, erasureDisks, minioMetaBucket, cfgFile, "", false)
+		if countErrs(errs, nil) != len(fileInfos) {
+			t.Fatal("Expected all xl.meta healed, but partial heal detected")
+		}
+	}
+}
+
+func TestHealObjectCorruptedPools(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	resetGlobalHealState()
+	defer resetGlobalHealState()
+
+	nDisks := 32
+	fsDirs, err := getRandomDisks(nDisks)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	defer removeRoots(fsDirs)
+
+	pools := mustGetPoolEndpoints(fsDirs[:16]...)
+	pools = append(pools, mustGetPoolEndpoints(fsDirs[16:]...)...)
+
+	// Everything is fine, should return nil
+	objLayer, _, err := initObjectLayer(ctx, pools)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	bucket := getRandomBucketName()
+	object := getRandomObjectName()
+	data := bytes.Repeat([]byte("a"), 5*1024*1024)
+	var opts ObjectOptions
+
+	err = objLayer.MakeBucketWithLocation(ctx, bucket, BucketOptions{})
+	if err != nil {
+		t.Fatalf("Failed to make a bucket - %v", err)
+	}
+
+	// Create an object with multiple parts uploaded in decreasing
+	// part number.
+	uploadID, err := objLayer.NewMultipartUpload(ctx, bucket, object, opts)
+	if err != nil {
+		t.Fatalf("Failed to create a multipart upload - %v", err)
+	}
+
+	var uploadedParts []CompletePart
+	for _, partID := range []int{2, 1} {
+		pInfo, err1 := objLayer.PutObjectPart(ctx, bucket, object, uploadID, partID, mustGetPutObjReader(t, bytes.NewReader(data), int64(len(data)), "", ""), opts)
+		if err1 != nil {
+			t.Fatalf("Failed to upload a part - %v", err1)
+		}
+		uploadedParts = append(uploadedParts, CompletePart{
+			PartNumber: pInfo.PartNumber,
+			ETag:       pInfo.ETag,
+		})
+	}
+
+	_, err = objLayer.CompleteMultipartUpload(ctx, bucket, object, uploadID, uploadedParts, ObjectOptions{})
+	if err != nil {
+		t.Fatalf("Failed to complete multipart upload - %v", err)
+	}
+
+	// Test 1: Remove the object backend files from the first disk.
+	z := objLayer.(*erasureServerPools)
+	for _, set := range z.serverPools {
+		er := set.sets[0]
+		erasureDisks := er.getDisks()
+		firstDisk := erasureDisks[0]
+		err = firstDisk.Delete(context.Background(), bucket, pathJoin(object, xlStorageFormatFile), false)
+		if err != nil {
+			t.Fatalf("Failed to delete a file - %v", err)
+		}
+
+		_, err = objLayer.HealObject(ctx, bucket, object, "", madmin.HealOpts{ScanMode: madmin.HealNormalScan})
+		if err != nil {
+			t.Fatalf("Failed to heal object - %v", err)
+		}
+
+		fileInfos, errs := readAllFileInfo(ctx, erasureDisks, bucket, object, "", false)
+		fi, err := getLatestFileInfo(ctx, fileInfos, errs)
+		if errors.Is(err, errFileNotFound) {
+			continue
+		}
+		if err != nil {
+			t.Fatalf("Failed to getLatestFileInfo - %v", err)
+		}
+
+		if _, err = firstDisk.StatInfoFile(context.Background(), bucket, object+"/"+xlStorageFormatFile, false); err != nil {
+			t.Errorf("Expected xl.meta file to be present but stat failed - %v", err)
+		}
+
+		err = firstDisk.Delete(context.Background(), bucket, pathJoin(object, fi.DataDir, "part.1"), false)
+		if err != nil {
+			t.Errorf("Failure during deleting part.1 - %v", err)
+		}
+
+		err = firstDisk.WriteAll(context.Background(), bucket, pathJoin(object, fi.DataDir, "part.1"), []byte{})
+		if err != nil {
+			t.Errorf("Failure during creating part.1 - %v", err)
+		}
+
+		_, err = objLayer.HealObject(ctx, bucket, object, "", madmin.HealOpts{DryRun: false, Remove: true, ScanMode: madmin.HealDeepScan})
+		if err != nil {
+			t.Errorf("Expected nil but received %v", err)
+		}
+
+		fileInfos, errs = readAllFileInfo(ctx, erasureDisks, bucket, object, "", false)
+		nfi, err := getLatestFileInfo(ctx, fileInfos, errs)
+		if err != nil {
+			t.Fatalf("Failed to getLatestFileInfo - %v", err)
+		}
+
+		if !reflect.DeepEqual(fi, nfi) {
+			t.Fatalf("FileInfo not equal after healing")
+		}
+
+		err = firstDisk.Delete(context.Background(), bucket, pathJoin(object, fi.DataDir, "part.1"), false)
+		if err != nil {
+			t.Errorf("Failure during deleting part.1 - %v", err)
+		}
+
+		bdata := bytes.Repeat([]byte("b"), int(nfi.Size))
+		err = firstDisk.WriteAll(context.Background(), bucket, pathJoin(object, fi.DataDir, "part.1"), bdata)
+		if err != nil {
+			t.Errorf("Failure during creating part.1 - %v", err)
+		}
+
+		_, err = objLayer.HealObject(ctx, bucket, object, "", madmin.HealOpts{DryRun: false, Remove: true, ScanMode: madmin.HealDeepScan})
+		if err != nil {
+			t.Errorf("Expected nil but received %v", err)
+		}
+
+		fileInfos, errs = readAllFileInfo(ctx, erasureDisks, bucket, object, "", false)
+		nfi, err = getLatestFileInfo(ctx, fileInfos, errs)
+		if err != nil {
+			t.Fatalf("Failed to getLatestFileInfo - %v", err)
+		}
+
+		if !reflect.DeepEqual(fi, nfi) {
+			t.Fatalf("FileInfo not equal after healing")
+		}
+
+		// Test 4: checks if HealObject returns an error when xl.meta is not found
+		// in more than read quorum number of disks, to create a corrupted situation.
+		for i := 0; i <= nfi.Erasure.DataBlocks; i++ {
+			erasureDisks[i].Delete(context.Background(), bucket, pathJoin(object, xlStorageFormatFile), false)
+		}
+
+		// Try healing now, expect to receive errFileNotFound.
+		_, err = objLayer.HealObject(ctx, bucket, object, "", madmin.HealOpts{DryRun: false, Remove: true, ScanMode: madmin.HealDeepScan})
+		if err != nil {
+			if _, ok := err.(ObjectNotFound); !ok {
+				t.Errorf("Expect %v but received %v", ObjectNotFound{Bucket: bucket, Object: object}, err)
+			}
+		}
+
+		// since majority of xl.meta's are not available, object should be successfully deleted.
+		_, err = objLayer.GetObjectInfo(ctx, bucket, object, ObjectOptions{})
+		if _, ok := err.(ObjectNotFound); !ok {
+			t.Errorf("Expect %v but received %v", ObjectNotFound{Bucket: bucket, Object: object}, err)
+		}
+
+		for i := 0; i < (nfi.Erasure.DataBlocks + nfi.Erasure.ParityBlocks); i++ {
+			_, err = erasureDisks[i].StatInfoFile(context.Background(), bucket, pathJoin(object, xlStorageFormatFile), false)
+			if err == nil {
+				t.Errorf("Expected xl.meta file to be not present, but succeeeded")
+			}
+		}
+	}
+}
+
 func TestHealObjectCorrupted(t *testing.T) {
 	ctx, cancel := context.WithCancel(context.Background())
 	defer cancel()
@@ -405,7 +884,7 @@ func TestHealObjectCorrupted(t *testing.T) {
 	}
 
 	if _, err = firstDisk.StatInfoFile(context.Background(), bucket, object+"/"+xlStorageFormatFile, false); err != nil {
-		t.Errorf("Expected er.meta file to be present but stat failed - %v", err)
+		t.Errorf("Expected xl.meta file to be present but stat failed - %v", err)
 	}
 
 	err = firstDisk.Delete(context.Background(), bucket, pathJoin(object, fi.DataDir, "part.1"), false)
@@ -549,7 +1028,7 @@ func TestHealObjectErasure(t *testing.T) {
 	}
 
 	if _, err = firstDisk.StatInfoFile(context.Background(), bucket, object+"/"+xlStorageFormatFile, false); err != nil {
-		t.Errorf("Expected er.meta file to be present but stat failed - %v", err)
+		t.Errorf("Expected xl.meta file to be present but stat failed - %v", err)
 	}
 
 	erasureDisks := er.getDisks()
@@ -565,7 +1044,7 @@ func TestHealObjectErasure(t *testing.T) {
 
 	// Try healing now, expect to receive errDiskNotFound.
 	_, err = obj.HealObject(ctx, bucket, object, "", madmin.HealOpts{ScanMode: madmin.HealDeepScan})
-	// since majority of er.meta's are not available, object quorum can't be read properly and error will be errErasureReadQuorum
+	// since majority of xl.meta's are not available, object quorum can't be read properly and error will be errErasureReadQuorum
 	if _, ok := err.(InsufficientReadQuorum); !ok {
 		t.Errorf("Expected %v but received %v", InsufficientReadQuorum{}, err)
 	}
@@ -699,7 +1178,8 @@ func TestHealLastDataShard(t *testing.T) {
 				t.Fatalf("Failed to make a bucket - %v", err)
 			}
 
-			_, err = obj.PutObject(ctx, bucket, object, mustGetPutObjReader(t, bytes.NewReader(data), int64(len(data)), "", ""), opts)
+			_, err = obj.PutObject(ctx, bucket, object,
+				mustGetPutObjReader(t, bytes.NewReader(data), int64(len(data)), "", ""), opts)
 
 			if err != nil {
 				t.Fatal(err)
@@ -724,7 +1204,9 @@ func TestHealLastDataShard(t *testing.T) {
 			if err != nil {
 				t.Fatalf("Failed to delete a file - %v", err)
 			}
-			_, err = obj.HealObject(ctx, bucket, object, "", madmin.HealOpts{ScanMode: madmin.HealNormalScan})
+			_, err = obj.HealObject(ctx, bucket, object, "", madmin.HealOpts{
+				ScanMode: madmin.HealNormalScan,
+			})
 			if err != nil {
 				t.Fatal(err)
 			}
@@ -738,21 +1220,23 @@ func TestHealLastDataShard(t *testing.T) {
 			firstHealedH := sha256.New()
 			_, err = io.Copy(firstHealedH, firstGr)
 			if err != nil {
-				t.Fatal()
+				t.Fatal(err)
 			}
 			firstHealedDataSha256 := firstHealedH.Sum(nil)
 
 			if !bytes.Equal(actualSha256, firstHealedDataSha256) {
-				t.Fatal("object healed wrong")
+				t.Fatalf("object healed wrong, expected %x, got %x",
+					actualSha256, firstHealedDataSha256)
 			}
 
 			// remove another data shard
-			err = removeAll(pathJoin(shuffledDisks[1].String(), bucket, object))
-			if err != nil {
+			if err = removeAll(pathJoin(shuffledDisks[1].String(), bucket, object)); err != nil {
 				t.Fatalf("Failed to delete a file - %v", err)
 			}
 
-			_, err = obj.HealObject(ctx, bucket, object, "", madmin.HealOpts{ScanMode: madmin.HealNormalScan})
+			_, err = obj.HealObject(ctx, bucket, object, "", madmin.HealOpts{
+				ScanMode: madmin.HealNormalScan,
+			})
 			if err != nil {
 				t.Fatal(err)
 			}
@@ -766,12 +1250,13 @@ func TestHealLastDataShard(t *testing.T) {
 			secondHealedH := sha256.New()
 			_, err = io.Copy(secondHealedH, secondGr)
 			if err != nil {
-				t.Fatal()
+				t.Fatal(err)
 			}
 			secondHealedDataSha256 := secondHealedH.Sum(nil)
 
 			if !bytes.Equal(actualSha256, secondHealedDataSha256) {
-				t.Fatal("object healed wrong")
+				t.Fatalf("object healed wrong, expected %x, got %x",
+					actualSha256, secondHealedDataSha256)
 			}
 		})
 	}

cmd/erasure-metadata-utils_test.go

@@ -87,9 +87,13 @@ func TestReduceErrs(t *testing.T) {
 			errDiskNotFound,
 		}, []error{errDiskNotFound}, errVolumeNotFound},
 		{[]error{}, []error{}, errErasureReadQuorum},
-		{[]error{errFileNotFound, errFileNotFound, errFileNotFound,
-			errFileNotFound, errFileNotFound, nil, nil, nil, nil, nil},
-			nil, nil},
+		{
+			[]error{
+				errFileNotFound, errFileNotFound, errFileNotFound,
+				errFileNotFound, errFileNotFound, nil, nil, nil, nil, nil,
+			},
+			nil, nil,
+		},
 		// Checks if wrapped context cancelation errors are grouped as one.
 		{canceledErrs, nil, context.Canceled},
 	}

cmd/erasure-metadata.go

@@ -351,6 +351,17 @@ func findFileInfoInQuorum(ctx context.Context, metaArr []FileInfo, modTime time.
 	return FileInfo{}, errErasureReadQuorum
 }
 
+func pickValidDiskTimeWithQuorum(metaArr []FileInfo, quorum int) time.Time {
+	diskMTimes := listObjectDiskMtimes(metaArr)
+
+	diskMTime, diskMaxima := commonTimeAndOccurence(diskMTimes, shardDiskTimeDelta)
+	if diskMaxima >= quorum {
+		return diskMTime
+	}
+
+	return timeSentinel
+}
+
 // pickValidFileInfo - picks one valid FileInfo content and returns from a
 // slice of FileInfo.
 func pickValidFileInfo(ctx context.Context, metaArr []FileInfo, modTime time.Time, quorum int) (FileInfo, error) {

cmd/erasure-multipart.go

@@ -29,6 +29,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/klauspost/readahead"
 	"github.com/minio/minio-go/v7/pkg/set"
 	xhttp "github.com/minio/minio/internal/http"
 	"github.com/minio/minio/internal/logger"
@@ -392,6 +393,53 @@ func (er erasureObjects) CopyObjectPart(ctx context.Context, srcBucket, srcObjec
 	return partInfo, nil
 }
 
+func undoRenamePart(disks []StorageAPI, srcBucket, srcEntry, dstBucket, dstEntry string, errs []error) {
+	// Undo rename object on disks where RenameFile succeeded.
+	g := errgroup.WithNErrs(len(disks))
+	for index, disk := range disks {
+		if disk == nil {
+			continue
+		}
+		index := index
+		g.Go(func() error {
+			if errs[index] == nil {
+				_ = disks[index].RenameFile(context.TODO(), dstBucket, dstEntry, srcBucket, srcEntry)
+			}
+			return nil
+		}, index)
+	}
+	g.Wait()
+}
+
+// renamePart - renames multipart part to its relevant location under uploadID.
+func renamePart(ctx context.Context, disks []StorageAPI, srcBucket, srcEntry, dstBucket, dstEntry string, writeQuorum int) ([]StorageAPI, error) {
+	g := errgroup.WithNErrs(len(disks))
+
+	// Rename file on all underlying storage disks.
+	for index := range disks {
+		index := index
+		g.Go(func() error {
+			if disks[index] == nil {
+				return errDiskNotFound
+			}
+			return disks[index].RenameFile(ctx, srcBucket, srcEntry, dstBucket, dstEntry)
+		}, index)
+	}
+
+	// Wait for all renames to finish.
+	errs := g.Wait()
+
+	// We can safely allow RenameFile errors up to len(er.getDisks()) - writeQuorum
+	// otherwise return failure. Cleanup successful renames.
+	err := reduceWriteQuorumErrs(ctx, errs, objectOpIgnoredErrs, writeQuorum)
+	if err == errErasureWriteQuorum {
+		// Undo all the partial rename operations.
+		undoRenamePart(disks, srcBucket, srcEntry, dstBucket, dstEntry, errs)
+	}
+
+	return evalDisks(disks, errs), err
+}
+
 // PutObjectPart - reads incoming stream and internally erasure codes
 // them. This call is similar to single put operation but it is part
 // of the multipart transaction.
@@ -481,7 +529,7 @@ func (er erasureObjects) PutObjectPart(ctx context.Context, bucket, object, uplo
 	var online int
 	defer func() {
 		if online != len(onlineDisks) {
-			er.deleteObject(context.Background(), minioMetaTmpBucket, tmpPart, writeQuorum)
+			er.renameAll(context.Background(), minioMetaTmpBucket, tmpPart)
 		}
 	}()
 
@@ -521,7 +569,22 @@ func (er erasureObjects) PutObjectPart(ctx context.Context, bucket, object, uplo
 		writers[i] = newBitrotWriter(disk, minioMetaTmpBucket, tmpPartPath, erasure.ShardFileSize(data.Size()), DefaultBitrotAlgorithm, erasure.ShardSize())
 	}
 
-	n, err := erasure.Encode(pctx, data, writers, buffer, writeQuorum)
+	toEncode := io.Reader(data)
+	if data.Size() > bigFileThreshold {
+		// Add input readahead.
+		// We use 2 buffers, so we always have a full buffer of input.
+		bufA := er.bp.Get()
+		bufB := er.bp.Get()
+		defer er.bp.Put(bufA)
+		defer er.bp.Put(bufB)
+		ra, err := readahead.NewReaderBuffer(data, [][]byte{bufA[:fi.Erasure.BlockSize], bufB[:fi.Erasure.BlockSize]})
+		if err == nil {
+			toEncode = ra
+			defer ra.Close()
+		}
+	}
+
+	n, err := erasure.Encode(pctx, toEncode, writers, buffer, writeQuorum)
 	closeBitrotWriters(writers)
 	if err != nil {
 		return pi, toObjectErr(err, bucket, object)
@@ -555,7 +618,7 @@ func (er erasureObjects) PutObjectPart(ctx context.Context, bucket, object, uplo
 
 	// Rename temporary part file to its final location.
 	partPath := pathJoin(uploadIDPath, fi.DataDir, partSuffix)
-	onlineDisks, err = rename(wctx, onlineDisks, minioMetaTmpBucket, tmpPartPath, minioMetaMultipartBucket, partPath, false, writeQuorum, nil)
+	onlineDisks, err = renamePart(wctx, onlineDisks, minioMetaTmpBucket, tmpPartPath, minioMetaMultipartBucket, partPath, writeQuorum)
 	if err != nil {
 		return pi, toObjectErr(err, minioMetaMultipartBucket, partPath)
 	}
@@ -820,7 +883,7 @@ func (er erasureObjects) CompleteMultipartUpload(ctx context.Context, bucket str
 	onlineDisks, partsMetadata = shuffleDisksAndPartsMetadataByIndex(onlineDisks, partsMetadata, fi)
 
 	// Save current erasure metadata for validation.
-	var currentFI = fi
+	currentFI := fi
 
 	// Allocate parts similar to incoming slice.
 	fi.Parts = make([]ObjectPartInfo, len(parts))
@@ -978,21 +1041,8 @@ func (er erasureObjects) AbortMultipartUpload(ctx context.Context, bucket, objec
 		return toObjectErr(err, bucket, object, uploadID)
 	}
 
-	uploadIDPath := er.getUploadIDDir(bucket, object, uploadID)
-
-	// Read metadata associated with the object from all disks.
-	partsMetadata, errs := readAllFileInfo(ctx, er.getDisks(), minioMetaMultipartBucket, uploadIDPath, "", false)
-
-	// get Quorum for this object
-	_, writeQuorum, err := objectQuorumFromMeta(ctx, partsMetadata, errs, er.defaultParityCount)
-	if err != nil {
-		return toObjectErr(err, bucket, object, uploadID)
-	}
-
 	// Cleanup all uploaded parts.
-	if err = er.deleteObject(ctx, minioMetaMultipartBucket, uploadIDPath, writeQuorum); err != nil {
-		return toObjectErr(err, bucket, object, uploadID)
-	}
+	er.renameAll(ctx, minioMetaMultipartBucket, er.getUploadIDDir(bucket, object, uploadID))
 
 	// Successfully purged.
 	return nil

cmd/erasure-object.go

@@ -29,6 +29,7 @@ import (
 	"strings"
 	"sync"
 
+	"github.com/klauspost/readahead"
 	"github.com/minio/madmin-go"
 	"github.com/minio/minio-go/v7/pkg/tags"
 	"github.com/minio/minio/internal/bucket/lifecycle"
@@ -145,7 +146,7 @@ func (er erasureObjects) CopyObject(ctx context.Context, srcBucket, srcObject, d
 // 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) {
 	var unlockOnDefer bool
-	var nsUnlocker = func() {}
+	nsUnlocker := func() {}
 	defer func() {
 		if unlockOnDefer {
 			nsUnlocker()
@@ -179,6 +180,30 @@ func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object stri
 		return nil, toObjectErr(err, bucket, object)
 	}
 
+	if !fi.DataShardFixed() {
+		diskMTime := pickValidDiskTimeWithQuorum(metaArr, fi.Erasure.DataBlocks)
+		if !diskMTime.Equal(timeSentinel) && !diskMTime.IsZero() {
+			for index := range onlineDisks {
+				if onlineDisks[index] == OfflineDisk {
+					continue
+				}
+				if !metaArr[index].IsValid() {
+					continue
+				}
+				if !metaArr[index].AcceptableDelta(diskMTime, shardDiskTimeDelta) {
+					// If disk mTime mismatches it is considered outdated
+					// https://github.com/minio/minio/pull/13803
+					//
+					// This check only is active if we could find maximally
+					// occurring disk mtimes that are somewhat same across
+					// the quorum. Allowing to skip those shards which we
+					// might think are wrong.
+					onlineDisks[index] = OfflineDisk
+				}
+			}
+		}
+	}
+
 	objInfo := fi.ToObjectInfo(bucket, object)
 	if objInfo.DeleteMarker {
 		if opts.VersionID == "" {
@@ -384,7 +409,8 @@ func (er erasureObjects) getObjectFileInfo(ctx context.Context, bucket, object s
 	}
 
 	if reducedErr := reduceReadQuorumErrs(ctx, errs, objectOpIgnoredErrs, readQuorum); reducedErr != nil {
-		if reducedErr == errErasureReadQuorum && bucket != minioMetaBucket {
+		if errors.Is(reducedErr, errErasureReadQuorum) && !strings.HasPrefix(bucket, minioMetaBucket) {
+			// Skip buckets that live at `.minio.sys` bucket.
 			if _, ok := isObjectDangling(metaArr, errs, nil); ok {
 				reducedErr = errFileNotFound
 				if opts.VersionID != "" {
@@ -451,7 +477,6 @@ func (er erasureObjects) getObjectInfo(ctx context.Context, bucket, object strin
 	fi, _, _, err := er.getObjectFileInfo(ctx, bucket, object, opts, false)
 	if err != nil {
 		return objInfo, toObjectErr(err, bucket, object)
-
 	}
 	objInfo = fi.ToObjectInfo(bucket, object)
 	if opts.VersionID != "" && !fi.VersionPurgeStatus().Empty() {
@@ -474,7 +499,7 @@ func (er erasureObjects) getObjectInfo(ctx context.Context, bucket, object strin
 func (er erasureObjects) getObjectInfoAndQuorum(ctx context.Context, bucket, object string, opts ObjectOptions) (objInfo ObjectInfo, wquorum int, err error) {
 	fi, _, _, err := er.getObjectFileInfo(ctx, bucket, object, opts, false)
 	if err != nil {
-		return objInfo, getWriteQuorum(len(er.getDisks())), toObjectErr(err, bucket, object)
+		return objInfo, er.defaultWQuorum(), toObjectErr(err, bucket, object)
 	}
 
 	wquorum = fi.Erasure.DataBlocks
@@ -499,31 +524,6 @@ func (er erasureObjects) getObjectInfoAndQuorum(ctx context.Context, bucket, obj
 	return objInfo, wquorum, nil
 }
 
-func undoRename(disks []StorageAPI, srcBucket, srcEntry, dstBucket, dstEntry string, isDir bool, errs []error) {
-	// Undo rename object on disks where RenameFile succeeded.
-
-	// If srcEntry/dstEntry are objects then add a trailing slash to copy
-	// over all the parts inside the object directory
-	if isDir {
-		srcEntry = retainSlash(srcEntry)
-		dstEntry = retainSlash(dstEntry)
-	}
-	g := errgroup.WithNErrs(len(disks))
-	for index, disk := range disks {
-		if disk == nil {
-			continue
-		}
-		index := index
-		g.Go(func() error {
-			if errs[index] == nil {
-				_ = disks[index].RenameFile(context.TODO(), dstBucket, dstEntry, srcBucket, srcEntry)
-			}
-			return nil
-		}, index)
-	}
-	g.Wait()
-}
-
 // Similar to rename but renames data from srcEntry to dstEntry at dataDir
 func renameData(ctx context.Context, disks []StorageAPI, srcBucket, srcEntry string, metadata []FileInfo, dstBucket, dstEntry string, writeQuorum int) ([]StorageAPI, error) {
 	defer NSUpdated(dstBucket, dstEntry)
@@ -564,46 +564,6 @@ func renameData(ctx context.Context, disks []StorageAPI, srcBucket, srcEntry str
 	return evalDisks(disks, errs), err
 }
 
-// rename - common function that renamePart and renameObject use to rename
-// the respective underlying storage layer representations.
-func rename(ctx context.Context, disks []StorageAPI, srcBucket, srcEntry, dstBucket, dstEntry string, isDir bool, writeQuorum int, ignoredErr []error) ([]StorageAPI, error) {
-	if isDir {
-		dstEntry = retainSlash(dstEntry)
-		srcEntry = retainSlash(srcEntry)
-	}
-	defer NSUpdated(dstBucket, dstEntry)
-
-	g := errgroup.WithNErrs(len(disks))
-
-	// Rename file on all underlying storage disks.
-	for index := range disks {
-		index := index
-		g.Go(func() error {
-			if disks[index] == nil {
-				return errDiskNotFound
-			}
-			if err := disks[index].RenameFile(ctx, srcBucket, srcEntry, dstBucket, dstEntry); err != nil {
-				if !IsErrIgnored(err, ignoredErr...) {
-					return err
-				}
-			}
-			return nil
-		}, index)
-	}
-
-	// Wait for all renames to finish.
-	errs := g.Wait()
-
-	// We can safely allow RenameFile errors up to len(er.getDisks()) - writeQuorum
-	// otherwise return failure. Cleanup successful renames.
-	err := reduceWriteQuorumErrs(ctx, errs, objectOpIgnoredErrs, writeQuorum)
-	if err == errErasureWriteQuorum {
-		// Undo all the partial rename operations.
-		undoRename(disks, srcBucket, srcEntry, dstBucket, dstEntry, isDir, errs)
-	}
-	return evalDisks(disks, errs), err
-}
-
 func (er erasureObjects) putMetacacheObject(ctx context.Context, key string, r *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error) {
 	data := r.Reader
 
@@ -677,9 +637,10 @@ func (er erasureObjects) putMetacacheObject(ctx context.Context, key string, r *
 		if disk == nil {
 			continue
 		}
-
-		inlineBuffers[i] = bytes.NewBuffer(make([]byte, 0, shardFileSize))
-		writers[i] = newStreamingBitrotWriterBuffer(inlineBuffers[i], DefaultBitrotAlgorithm, erasure.ShardSize())
+		if disk.IsOnline() {
+			inlineBuffers[i] = bytes.NewBuffer(make([]byte, 0, shardFileSize))
+			writers[i] = newStreamingBitrotWriterBuffer(inlineBuffers[i], DefaultBitrotAlgorithm, erasure.ShardSize())
+		}
 	}
 
 	n, erasureErr := erasure.Encode(ctx, data, writers, buffer, writeQuorum)
@@ -696,6 +657,7 @@ func (er erasureObjects) putMetacacheObject(ctx context.Context, key string, r *
 
 	for i, w := range writers {
 		if w == nil {
+			// Make sure to avoid writing to disks which we couldn't complete in erasure.Encode()
 			onlineDisks[i] = nil
 			continue
 		}
@@ -878,7 +840,7 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 	var online int
 	defer func() {
 		if online != len(onlineDisks) {
-			er.deleteObject(context.Background(), minioMetaTmpBucket, tempObj, writeQuorum)
+			er.renameAll(context.Background(), minioMetaTmpBucket, tempObj)
 		}
 	}()
 
@@ -906,6 +868,10 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 			continue
 		}
 
+		if !disk.IsOnline() {
+			continue
+		}
+
 		if len(inlineBuffers) > 0 {
 			sz := shardFileSize
 			if sz < 0 {
@@ -915,10 +881,25 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 			writers[i] = newStreamingBitrotWriterBuffer(inlineBuffers[i], DefaultBitrotAlgorithm, erasure.ShardSize())
 			continue
 		}
+
 		writers[i] = newBitrotWriter(disk, minioMetaTmpBucket, tempErasureObj, shardFileSize, DefaultBitrotAlgorithm, erasure.ShardSize())
 	}
 
-	n, erasureErr := erasure.Encode(ctx, data, writers, buffer, writeQuorum)
+	toEncode := io.Reader(data)
+	if data.Size() > bigFileThreshold {
+		// We use 2 buffers, so we always have a full buffer of input.
+		bufA := er.bp.Get()
+		bufB := er.bp.Get()
+		defer er.bp.Put(bufA)
+		defer er.bp.Put(bufB)
+		ra, err := readahead.NewReaderBuffer(data, [][]byte{bufA[:fi.Erasure.BlockSize], bufB[:fi.Erasure.BlockSize]})
+		if err == nil {
+			toEncode = ra
+			defer ra.Close()
+		}
+		logger.LogIf(ctx, err)
+	}
+	n, erasureErr := erasure.Encode(ctx, toEncode, writers, buffer, writeQuorum)
 	closeBitrotWriters(writers)
 	if erasureErr != nil {
 		return ObjectInfo{}, toObjectErr(erasureErr, minioMetaTmpBucket, tempErasureObj)
@@ -988,6 +969,9 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 
 	// Rename the successfully written temporary object to final location.
 	if onlineDisks, err = renameData(ctx, onlineDisks, minioMetaTmpBucket, tempObj, partsMetadata, bucket, object, writeQuorum); err != nil {
+		if errors.Is(err, errFileNotFound) {
+			return ObjectInfo{}, toObjectErr(errErasureWriteQuorum, bucket, object)
+		}
 		logger.LogIf(ctx, err)
 		return ObjectInfo{}, toObjectErr(err, bucket, object)
 	}
@@ -1036,42 +1020,6 @@ func (er erasureObjects) deleteObjectVersion(ctx context.Context, bucket, object
 	return reduceWriteQuorumErrs(ctx, g.Wait(), objectOpIgnoredErrs, writeQuorum)
 }
 
-// deleteObject - wrapper for delete object, deletes an object from
-// all the disks in parallel, including `xl.meta` associated with the
-// object.
-func (er erasureObjects) deleteObject(ctx context.Context, bucket, object string, writeQuorum int) error {
-	var err error
-
-	disks := er.getDisks()
-	tmpObj := mustGetUUID()
-	if bucket == minioMetaTmpBucket {
-		tmpObj = object
-	} else {
-		// Rename the current object while requiring write quorum, but also consider
-		// that a non found object in a given disk as a success since it already
-		// confirms that the object doesn't have a part in that disk (already removed)
-		disks, err = rename(ctx, disks, bucket, object, minioMetaTmpBucket, tmpObj, true, writeQuorum,
-			[]error{errFileNotFound})
-		if err != nil {
-			return toObjectErr(err, bucket, object)
-		}
-	}
-
-	g := errgroup.WithNErrs(len(disks))
-	for index := range disks {
-		index := index
-		g.Go(func() error {
-			if disks[index] == nil {
-				return errDiskNotFound
-			}
-			return disks[index].Delete(ctx, minioMetaTmpBucket, tmpObj, true)
-		}, index)
-	}
-
-	// return errors if any during deletion
-	return reduceWriteQuorumErrs(ctx, g.Wait(), objectOpIgnoredErrs, writeQuorum)
-}
-
 // DeleteObjects deletes objects/versions in bulk, this function will still automatically split objects list
 // 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.
@@ -1089,7 +1037,7 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 		// class for objects which have reduced quorum
 		// storage class only needs to be honored for
 		// Read() requests alone which we already do.
-		writeQuorums[i] = getWriteQuorum(len(storageDisks))
+		writeQuorums[i] = len(storageDisks)/2 + 1
 	}
 
 	versionsMap := make(map[string]FileInfoVersions, len(objects))
@@ -1153,7 +1101,7 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 	}
 
 	// Initialize list of errors.
-	var delObjErrs = make([][]error, len(storageDisks))
+	delObjErrs := make([][]error, len(storageDisks))
 
 	var wg sync.WaitGroup
 	// Remove versions in bulk for each disk
@@ -1174,6 +1122,12 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 					continue
 				}
 				for _, v := range dedupVersions[i].Versions {
+					if err == errFileNotFound || err == errFileVersionNotFound {
+						if !dobjects[v.Idx].DeleteMarker {
+							// Not delete marker, if not found, ok.
+							continue
+						}
+					}
 					delObjErrs[index][v.Idx] = err
 				}
 			}
@@ -1199,9 +1153,7 @@ func (er erasureObjects) DeleteObjects(ctx context.Context, bucket string, objec
 			errs[objIndex] = toObjectErr(err, bucket, objects[objIndex].ObjectName)
 		}
 
-		if errs[objIndex] == nil {
-			defer NSUpdated(bucket, objects[objIndex].ObjectName)
-		}
+		defer NSUpdated(bucket, objects[objIndex].ObjectName)
 	}
 
 	// Check failed deletes across multiple objects
@@ -1444,14 +1396,16 @@ func (er erasureObjects) addPartial(bucket, object, versionID string, size int64
 }
 
 func (er erasureObjects) PutObjectMetadata(ctx context.Context, bucket, object string, opts ObjectOptions) (ObjectInfo, error) {
-	// Lock the object before updating tags.
-	lk := er.NewNSLock(bucket, object)
-	lkctx, err := lk.GetLock(ctx, globalOperationTimeout)
-	if err != nil {
-		return ObjectInfo{}, err
+	if !opts.NoLock {
+		// Lock the object before updating metadata.
+		lk := er.NewNSLock(bucket, object)
+		lkctx, err := lk.GetLock(ctx, globalOperationTimeout)
+		if err != nil {
+			return ObjectInfo{}, err
+		}
+		ctx = lkctx.Context()
+		defer lk.Unlock(lkctx.Cancel)
 	}
-	ctx = lkctx.Context()
-	defer lk.Unlock(lkctx.Cancel)
 
 	disks := er.getDisks()
 
@@ -1541,6 +1495,7 @@ func (er erasureObjects) PutObjectTags(ctx context.Context, bucket, object strin
 	filterOnlineDisksInplace(fi, metaArr, onlineDisks)
 
 	fi.Metadata[xhttp.AmzObjectTagging] = tags
+	fi.ReplicationState = opts.PutReplicationState()
 	for k, v := range opts.UserDefined {
 		fi.Metadata[k] = v
 	}
@@ -1574,7 +1529,7 @@ func (er erasureObjects) updateObjectMeta(ctx context.Context, bucket, object st
 	// Wait for all the routines.
 	mErrs := g.Wait()
 
-	return reduceWriteQuorumErrs(ctx, mErrs, objectOpIgnoredErrs, getWriteQuorum(len(onlineDisks)))
+	return reduceWriteQuorumErrs(ctx, mErrs, objectOpIgnoredErrs, er.defaultWQuorum())
 }
 
 // DeleteObjectTags - delete object tags from an existing object
@@ -1794,6 +1749,7 @@ func (er erasureObjects) restoreTransitionedObject(ctx context.Context, bucket s
 		})
 	}
 	_, err = er.CompleteMultipartUpload(ctx, bucket, object, uploadID, uploadedParts, ObjectOptions{
-		MTime: oi.ModTime})
+		MTime: oi.ModTime,
+	})
 	return setRestoreHeaderFn(oi, err)
 }

cmd/erasure-object_test.go

@@ -177,7 +177,11 @@ func TestErasureDeleteObjectsErasureSet(t *testing.T) {
 	toObjectNames := func(testCases []testCaseType) []ObjectToDelete {
 		names := make([]ObjectToDelete, len(testCases))
 		for i := range testCases {
-			names[i] = ObjectToDelete{ObjectName: testCases[i].object}
+			names[i] = ObjectToDelete{
+				ObjectV: ObjectV{
+					ObjectName: testCases[i].object,
+				},
+			}
 		}
 		return names
 	}
@@ -505,7 +509,6 @@ func TestGetObjectNoQuorum(t *testing.T) {
 			gr.Close()
 		}
 	}
-
 }
 
 func TestHeadObjectNoQuorum(t *testing.T) {

cmd/erasure-server-pool-decom.go

@@ -0,0 +1,996 @@
+// Copyright (c) 2015-2021 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"
+	"encoding/binary"
+	"errors"
+	"fmt"
+	"net/http"
+	"sort"
+	"time"
+
+	"github.com/dustin/go-humanize"
+	"github.com/minio/minio/internal/hash"
+	"github.com/minio/minio/internal/logger"
+	"github.com/minio/pkg/console"
+)
+
+// PoolDecommissionInfo currently decommissioning information
+type PoolDecommissionInfo struct {
+	StartTime   time.Time `json:"startTime" msg:"st"`
+	StartSize   int64     `json:"startSize" msg:"ss"`
+	TotalSize   int64     `json:"totalSize" msg:"ts"`
+	CurrentSize int64     `json:"currentSize" msg:"cs"`
+
+	Complete bool `json:"complete" msg:"cmp"`
+	Failed   bool `json:"failed" msg:"fl"`
+	Canceled bool `json:"canceled" msg:"cnl"`
+
+	// Internal information.
+	QueuedBuckets         []string `json:"-" msg:"bkts"`
+	DecommissionedBuckets []string `json:"-" msg:"dbkts"`
+
+	// Last bucket/object decommissioned.
+	Bucket string `json:"-" msg:"bkt"`
+	Object string `json:"-" msg:"obj"`
+
+	// Verbose information
+	ItemsDecommissioned     int64 `json:"-" msg:"id"`
+	ItemsDecommissionFailed int64 `json:"-" msg:"idf"`
+	BytesDone               int64 `json:"-" msg:"bd"`
+	BytesFailed             int64 `json:"-" msg:"bf"`
+}
+
+// bucketPop should be called when a bucket is done decommissioning.
+// Adds the bucket to the list of decommissioned buckets and updates resume numbers.
+func (pd *PoolDecommissionInfo) bucketPop(bucket string) {
+	pd.DecommissionedBuckets = append(pd.DecommissionedBuckets, bucket)
+	for i, b := range pd.QueuedBuckets {
+		if b == bucket {
+			// Bucket is done.
+			pd.QueuedBuckets = append(pd.QueuedBuckets[:i], pd.QueuedBuckets[i+1:]...)
+			// Clear tracker info.
+			if pd.Bucket == bucket {
+				pd.Bucket = "" // empty this out for next bucket
+				pd.Object = "" // empty this out for next object
+			}
+			return
+		}
+	}
+}
+
+func (pd *PoolDecommissionInfo) bucketsToDecommission() []string {
+	queuedBuckets := make([]string, len(pd.QueuedBuckets))
+	copy(queuedBuckets, pd.QueuedBuckets)
+	return queuedBuckets
+}
+
+func (pd *PoolDecommissionInfo) isBucketDecommissioned(bucket string) bool {
+	for _, b := range pd.DecommissionedBuckets {
+		if b == bucket {
+			return true
+		}
+	}
+	return false
+}
+
+func (pd *PoolDecommissionInfo) bucketPush(bucket string) {
+	for _, b := range pd.QueuedBuckets {
+		if pd.isBucketDecommissioned(b) {
+			return
+		}
+		if b == bucket {
+			return
+		}
+	}
+	pd.QueuedBuckets = append(pd.QueuedBuckets, bucket)
+	pd.Bucket = bucket
+}
+
+// PoolStatus captures current pool status
+type PoolStatus struct {
+	ID           int                   `json:"id" msg:"id"`
+	CmdLine      string                `json:"cmdline" msg:"cl"`
+	LastUpdate   time.Time             `json:"lastUpdate" msg:"lu"`
+	Decommission *PoolDecommissionInfo `json:"decommissionInfo,omitempty" msg:"dec"`
+}
+
+//go:generate msgp -file $GOFILE -unexported
+type poolMeta struct {
+	Version int          `msg:"v"`
+	Pools   []PoolStatus `msg:"pls"`
+}
+
+// A decommission resumable tells us if decommission is worth
+// resuming upon restart of a cluster.
+func (p *poolMeta) returnResumablePools(n int) []PoolStatus {
+	var newPools []PoolStatus
+	for _, pool := range p.Pools {
+		if pool.Decommission == nil {
+			continue
+		}
+		if pool.Decommission.Complete || pool.Decommission.Canceled {
+			// Do not resume decommission upon startup for
+			// - decommission complete
+			// - decommission canceled
+			continue
+		} // In all other situations we need to resume
+		newPools = append(newPools, pool)
+		if n > 0 && len(newPools) == n {
+			return newPools
+		}
+	}
+	return nil
+}
+
+func (p *poolMeta) DecommissionComplete(idx int) bool {
+	if p.Pools[idx].Decommission != nil {
+		p.Pools[idx].LastUpdate = UTCNow()
+		p.Pools[idx].Decommission.Complete = true
+		p.Pools[idx].Decommission.Failed = false
+		p.Pools[idx].Decommission.Canceled = false
+		return true
+	}
+	return false
+}
+
+func (p *poolMeta) DecommissionFailed(idx int) bool {
+	if p.Pools[idx].Decommission != nil {
+		p.Pools[idx].LastUpdate = UTCNow()
+		p.Pools[idx].Decommission.StartTime = time.Time{}
+		p.Pools[idx].Decommission.Complete = false
+		p.Pools[idx].Decommission.Failed = true
+		p.Pools[idx].Decommission.Canceled = false
+		return true
+	}
+	return false
+}
+
+func (p *poolMeta) DecommissionCancel(idx int) bool {
+	if p.Pools[idx].Decommission != nil {
+		p.Pools[idx].LastUpdate = UTCNow()
+		p.Pools[idx].Decommission.StartTime = time.Time{}
+		p.Pools[idx].Decommission.Complete = false
+		p.Pools[idx].Decommission.Failed = false
+		p.Pools[idx].Decommission.Canceled = true
+		return true
+	}
+	return false
+}
+
+func (p poolMeta) isBucketDecommissioned(idx int, bucket string) bool {
+	return p.Pools[idx].Decommission.isBucketDecommissioned(bucket)
+}
+
+func (p *poolMeta) BucketDone(idx int, bucket string) {
+	if p.Pools[idx].Decommission == nil {
+		// Decommission not in progress.
+		return
+	}
+	p.Pools[idx].Decommission.bucketPop(bucket)
+}
+
+func (p poolMeta) ResumeBucketObject(idx int) (bucket, object string) {
+	if p.Pools[idx].Decommission != nil {
+		bucket = p.Pools[idx].Decommission.Bucket
+		object = p.Pools[idx].Decommission.Object
+	}
+	return
+}
+
+func (p *poolMeta) TrackCurrentBucketObject(idx int, bucket string, object string) {
+	if p.Pools[idx].Decommission == nil {
+		// Decommission not in progress.
+		return
+	}
+	p.Pools[idx].Decommission.Bucket = bucket
+	p.Pools[idx].Decommission.Object = object
+}
+
+func (p *poolMeta) PendingBuckets(idx int) []string {
+	if p.Pools[idx].Decommission == nil {
+		// Decommission not in progress.
+		return nil
+	}
+
+	return p.Pools[idx].Decommission.bucketsToDecommission()
+}
+
+func (p *poolMeta) QueueBuckets(idx int, buckets []BucketInfo) {
+	// add new queued buckets
+	for _, bucket := range buckets {
+		p.Pools[idx].Decommission.bucketPush(bucket.Name)
+	}
+}
+
+var (
+	errDecommissionAlreadyRunning = errors.New("decommission is already in progress")
+	errDecommissionComplete       = errors.New("decommission is complete, please remove the servers from command-line")
+)
+
+func (p *poolMeta) Decommission(idx int, pi poolSpaceInfo) error {
+	for i, pool := range p.Pools {
+		if idx == i {
+			continue
+		}
+		if pool.Decommission != nil {
+			// Do not allow multiple decommissions at the same time.
+			// We shall for now only allow one pool decommission at
+			// a time.
+			return fmt.Errorf("%w at index: %d", errDecommissionAlreadyRunning, i)
+		}
+	}
+
+	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
+	}
+
+	// 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
+}
+
+func (p poolMeta) IsSuspended(idx int) bool {
+	return p.Pools[idx].Decommission != nil
+}
+
+func (p *poolMeta) validate(pools []*erasureSets) (bool, error) {
+	type poolInfo struct {
+		position     int
+		completed    bool
+		decomStarted bool // started but not finished yet
+	}
+
+	rememberedPools := make(map[string]poolInfo)
+	for idx, pool := range p.Pools {
+		complete := false
+		decomStarted := false
+		if pool.Decommission != nil {
+			if pool.Decommission.Complete {
+				complete = true
+			}
+			decomStarted = true
+		}
+		rememberedPools[pool.CmdLine] = poolInfo{
+			position:     idx,
+			completed:    complete,
+			decomStarted: decomStarted,
+		}
+	}
+
+	specifiedPools := make(map[string]int)
+	for idx, pool := range pools {
+		specifiedPools[pool.endpoints.CmdLine] = idx
+	}
+
+	// Check if specified pools need to remove decommissioned pool.
+	for k := range specifiedPools {
+		pi, ok := rememberedPools[k]
+		if ok && pi.completed {
+			return false, fmt.Errorf("pool(%s) = %s is decommissioned, please remove from server command line", humanize.Ordinal(pi.position+1), k)
+		}
+	}
+
+	// check if remembered pools are in right position or missing from command line.
+	for k, pi := range rememberedPools {
+		if pi.completed {
+			continue
+		}
+		_, ok := specifiedPools[k]
+		if !ok {
+			return false, fmt.Errorf("pool(%s) = %s is not specified, please specify on server command line", humanize.Ordinal(pi.position+1), k)
+		}
+	}
+
+	// check when remembered pools and specified pools are same they are at the expected position
+	if len(rememberedPools) == len(specifiedPools) {
+		for k, pi := range rememberedPools {
+			pos, ok := specifiedPools[k]
+			if !ok {
+				return false, fmt.Errorf("pool(%s) = %s is not specified, please specify on server command line", humanize.Ordinal(pi.position+1), k)
+			}
+			if pos != pi.position {
+				return false, fmt.Errorf("pool order change detected for %s, expected position is (%s) but found (%s)", k, humanize.Ordinal(pi.position+1), humanize.Ordinal(pos+1))
+			}
+		}
+	}
+
+	update := len(rememberedPools) != len(specifiedPools)
+	if update {
+		for k, pi := range rememberedPools {
+			if pi.decomStarted && !pi.completed {
+				return false, fmt.Errorf("pool(%s) = %s is being decommissioned, No changes should be made to the command line arguments. Please complete the decommission in progress", humanize.Ordinal(pi.position+1), k)
+			}
+		}
+	}
+	return update, nil
+}
+
+func (p *poolMeta) load(ctx context.Context, pool *erasureSets, pools []*erasureSets) error {
+	data, err := readConfig(ctx, pool, poolMetaName)
+	if err != nil {
+		if errors.Is(err, errConfigNotFound) || isErrObjectNotFound(err) {
+			return nil
+		}
+		return err
+	}
+	if len(data) == 0 {
+		// Seems to be empty create a new poolMeta object.
+		return nil
+	}
+	if len(data) <= 4 {
+		return fmt.Errorf("poolMeta: no data")
+	}
+	// Read header
+	switch binary.LittleEndian.Uint16(data[0:2]) {
+	case poolMetaFormat:
+	default:
+		return fmt.Errorf("poolMeta: unknown format: %d", binary.LittleEndian.Uint16(data[0:2]))
+	}
+	switch binary.LittleEndian.Uint16(data[2:4]) {
+	case poolMetaVersion:
+	default:
+		return fmt.Errorf("poolMeta: unknown version: %d", binary.LittleEndian.Uint16(data[2:4]))
+	}
+
+	// OK, parse data.
+	if _, err = p.UnmarshalMsg(data[4:]); err != nil {
+		return err
+	}
+
+	switch p.Version {
+	case poolMetaVersionV1:
+	default:
+		return fmt.Errorf("unexpected pool meta version: %d", p.Version)
+	}
+
+	return nil
+}
+
+func (p *poolMeta) CountItem(idx int, size int64, failed bool) {
+	pd := p.Pools[idx].Decommission
+	if pd != nil {
+		if failed {
+			pd.ItemsDecommissionFailed++
+			pd.BytesFailed += size
+		} else {
+			pd.ItemsDecommissioned++
+			pd.BytesDone += size
+		}
+		p.Pools[idx].Decommission = pd
+	}
+}
+
+func (p *poolMeta) updateAfter(ctx context.Context, idx int, pools []*erasureSets, duration time.Duration) (bool, error) {
+	if p.Pools[idx].Decommission == nil {
+		return false, errInvalidArgument
+	}
+	now := UTCNow()
+	if now.Sub(p.Pools[idx].LastUpdate) >= duration {
+		if serverDebugLog {
+			console.Debugf("decommission: persisting poolMeta on disk: threshold:%s, poolMeta:%#v\n", now.Sub(p.Pools[idx].LastUpdate), p.Pools[idx])
+		}
+		p.Pools[idx].LastUpdate = now
+		if err := p.save(ctx, pools); err != nil {
+			return false, err
+		}
+		return true, nil
+	}
+	return false, nil
+}
+
+func (p poolMeta) save(ctx context.Context, pools []*erasureSets) error {
+	data := make([]byte, 4, p.Msgsize()+4)
+
+	// Initialize the header.
+	binary.LittleEndian.PutUint16(data[0:2], poolMetaFormat)
+	binary.LittleEndian.PutUint16(data[2:4], poolMetaVersion)
+
+	buf, err := p.MarshalMsg(data)
+	if err != nil {
+		return err
+	}
+
+	// Saves on all pools to make sure decommissioning of first pool is allowed.
+	for _, eset := range pools {
+		if err = saveConfig(ctx, eset, poolMetaName, buf); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+const (
+	poolMetaName      = "pool.bin"
+	poolMetaFormat    = 1
+	poolMetaVersionV1 = 1
+	poolMetaVersion   = poolMetaVersionV1
+)
+
+// Init() initializes pools and saves additional information about them
+// in 'pool.bin', this is eventually used for decommissioning the pool.
+func (z *erasureServerPools) Init(ctx context.Context) error {
+	meta := poolMeta{}
+
+	if err := meta.load(ctx, z.serverPools[0], z.serverPools); err != nil {
+		return err
+	}
+
+	update, err := meta.validate(z.serverPools)
+	if err != nil {
+		return err
+	}
+
+	// if no update is needed return right away.
+	if !update {
+		// We are only supporting single pool decommission at this time
+		// so it makes sense to only resume single pools at any given
+		// time, in future meta.returnResumablePools() might take
+		// '-1' as argument to decommission multiple pools at a time
+		// but this is not a priority at the moment.
+		for _, pool := range meta.returnResumablePools(1) {
+			err := z.Decommission(ctx, pool.ID)
+			switch err {
+			case errDecommissionAlreadyRunning:
+				fallthrough
+			case nil:
+				go z.doDecommissionInRoutine(ctx, pool.ID)
+			}
+		}
+		z.poolMeta = meta
+
+		return nil
+	}
+
+	meta = poolMeta{} // to update write poolMeta fresh.
+	// looks like new pool was added we need to update,
+	// or this is a fresh installation (or an existing
+	// installation with pool removed)
+	meta.Version = poolMetaVersion
+	for idx, pool := range z.serverPools {
+		meta.Pools = append(meta.Pools, PoolStatus{
+			CmdLine:    pool.endpoints.CmdLine,
+			ID:         idx,
+			LastUpdate: UTCNow(),
+		})
+	}
+	if err = meta.save(ctx, z.serverPools); err != nil {
+		return err
+	}
+	z.poolMeta = meta
+	return nil
+}
+
+func (z *erasureServerPools) decommissionObject(ctx context.Context, bucket string, gr *GetObjectReader) (err error) {
+	defer gr.Close()
+	objInfo := gr.ObjInfo
+	if objInfo.isMultipart() {
+		uploadID, err := z.NewMultipartUpload(ctx, bucket, objInfo.Name, ObjectOptions{
+			VersionID:   objInfo.VersionID,
+			MTime:       objInfo.ModTime,
+			UserDefined: objInfo.UserDefined,
+		})
+		if err != nil {
+			return err
+		}
+		defer z.AbortMultipartUpload(ctx, bucket, objInfo.Name, uploadID, ObjectOptions{})
+		parts := make([]CompletePart, len(objInfo.Parts))
+		for i, part := range objInfo.Parts {
+			hr, err := hash.NewReader(gr, part.Size, "", "", part.Size)
+			if err != nil {
+				return err
+			}
+			pi, err := z.PutObjectPart(ctx, bucket, objInfo.Name, uploadID,
+				part.Number,
+				NewPutObjReader(hr),
+				ObjectOptions{})
+			if err != nil {
+				return err
+			}
+			parts[i] = CompletePart{
+				ETag:       pi.ETag,
+				PartNumber: pi.PartNumber,
+			}
+		}
+		_, err = z.CompleteMultipartUpload(ctx, bucket, objInfo.Name, uploadID, parts, ObjectOptions{
+			MTime: objInfo.ModTime,
+		})
+		return err
+	}
+	hr, err := hash.NewReader(gr, objInfo.Size, "", "", objInfo.Size)
+	if err != nil {
+		return err
+	}
+	_, err = z.PutObject(ctx,
+		bucket,
+		objInfo.Name,
+		NewPutObjReader(hr),
+		ObjectOptions{
+			VersionID:   objInfo.VersionID,
+			MTime:       objInfo.ModTime,
+			UserDefined: objInfo.UserDefined,
+		})
+	return err
+}
+
+// versionsSorter sorts FileInfo slices by version.
+//msgp:ignore versionsSorter
+type versionsSorter []FileInfo
+
+func (v versionsSorter) reverse() {
+	sort.Slice(v, func(i, j int) bool {
+		return v[i].ModTime.Before(v[j].ModTime)
+	})
+}
+
+func (z *erasureServerPools) decommissionPool(ctx context.Context, idx int, pool *erasureSets, bName string) error {
+	var forwardTo string
+	// If we resume to the same bucket, forward to last known item.
+	rbucket, robject := z.poolMeta.ResumeBucketObject(idx)
+	if rbucket != "" && rbucket == bName {
+		forwardTo = robject
+	}
+
+	versioned := globalBucketVersioningSys.Enabled(bName)
+	for _, set := range pool.sets {
+		disks := set.getOnlineDisks()
+		if len(disks) == 0 {
+			logger.LogIf(GlobalContext, fmt.Errorf("no online disks found for set with endpoints %s",
+				set.getEndpoints()))
+			continue
+		}
+
+		decommissionEntry := func(entry metaCacheEntry) {
+			if entry.isDir() {
+				return
+			}
+
+			fivs, err := entry.fileInfoVersions(bName)
+			if err != nil {
+				return
+			}
+
+			// We need a reversed order for Decommissioning,
+			// to create the appropriate stack.
+			versionsSorter(fivs.Versions).reverse()
+
+			for _, version := range fivs.Versions {
+				// TODO: Skip transitioned objects for now.
+				if version.IsRemote() {
+					continue
+				}
+				// We will skip decommissioning delete markers
+				// with single version, its as good as there
+				// is no data associated with the object.
+				if version.Deleted && len(fivs.Versions) == 1 {
+					continue
+				}
+				if version.Deleted {
+					_, err := z.DeleteObject(ctx,
+						bName,
+						version.Name,
+						ObjectOptions{
+							Versioned:         versioned,
+							VersionID:         version.VersionID,
+							MTime:             version.ModTime,
+							DeleteReplication: version.ReplicationState,
+						})
+					if err != nil {
+						logger.LogIf(ctx, err)
+						z.poolMetaMutex.Lock()
+						z.poolMeta.CountItem(idx, 0, true)
+						z.poolMetaMutex.Unlock()
+					} else {
+						set.DeleteObject(ctx,
+							bName,
+							version.Name,
+							ObjectOptions{
+								VersionID: version.VersionID,
+							})
+						z.poolMetaMutex.Lock()
+						z.poolMeta.CountItem(idx, 0, false)
+						z.poolMetaMutex.Unlock()
+					}
+					continue
+				}
+				gr, err := set.GetObjectNInfo(ctx,
+					bName,
+					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()
+					continue
+				}
+				// gr.Close() is ensured by decommissionObject().
+				if err = z.decommissionObject(ctx, bName, gr); err != nil {
+					logger.LogIf(ctx, err)
+					z.poolMetaMutex.Lock()
+					z.poolMeta.CountItem(idx, version.Size, true)
+					z.poolMetaMutex.Unlock()
+					continue
+				}
+				set.DeleteObject(ctx,
+					bName,
+					version.Name,
+					ObjectOptions{
+						VersionID: version.VersionID,
+					})
+				z.poolMetaMutex.Lock()
+				z.poolMeta.CountItem(idx, version.Size, false)
+				z.poolMetaMutex.Unlock()
+			}
+			z.poolMetaMutex.Lock()
+			z.poolMeta.TrackCurrentBucketObject(idx, bName, entry.name)
+			ok, err := z.poolMeta.updateAfter(ctx, idx, z.serverPools, 30*time.Second)
+			logger.LogIf(ctx, err)
+			if ok {
+				globalNotificationSys.ReloadPoolMeta(ctx)
+			}
+			z.poolMetaMutex.Unlock()
+		}
+
+		// How to resolve partial results.
+		resolver := metadataResolutionParams{
+			dirQuorum: len(disks) / 2, // make sure to capture all quorum ratios
+			objQuorum: len(disks) / 2, // make sure to capture all quorum ratios
+			bucket:    bName,
+		}
+
+		if err := listPathRaw(ctx, listPathRawOptions{
+			disks:          disks,
+			bucket:         bName,
+			recursive:      true,
+			forwardTo:      forwardTo,
+			minDisks:       len(disks) / 2, // to capture all quorum ratios
+			reportNotFound: false,
+			agreed:         decommissionEntry,
+			partial: func(entries metaCacheEntries, nAgreed int, errs []error) {
+				entry, ok := entries.resolve(&resolver)
+				if ok {
+					decommissionEntry(*entry)
+				}
+			},
+			finished: nil,
+		}); err != nil {
+			// Decommissioning failed and won't continue
+			return err
+		}
+	}
+	return nil
+}
+
+func (z *erasureServerPools) decommissionInBackground(ctx context.Context, idx int) error {
+	pool := z.serverPools[idx]
+	for _, bucket := range z.poolMeta.PendingBuckets(idx) {
+		if z.poolMeta.isBucketDecommissioned(idx, bucket) {
+			if serverDebugLog {
+				console.Debugln("decommission: already done, moving on", bucket)
+			}
+
+			z.poolMetaMutex.Lock()
+			z.poolMeta.BucketDone(idx, bucket) // remove from pendingBuckets and persist.
+			z.poolMeta.save(ctx, z.serverPools)
+			z.poolMetaMutex.Unlock()
+			continue
+		}
+		if serverDebugLog {
+			console.Debugln("decommission: currently on bucket", bucket)
+		}
+		if err := z.decommissionPool(ctx, idx, pool, bucket); err != nil {
+			return err
+		}
+		z.poolMetaMutex.Lock()
+		z.poolMeta.BucketDone(idx, bucket)
+		z.poolMeta.save(ctx, z.serverPools)
+		z.poolMetaMutex.Unlock()
+	}
+	return nil
+}
+
+func (z *erasureServerPools) doDecommissionInRoutine(ctx context.Context, idx int) {
+	z.poolMetaMutex.Lock()
+	var dctx context.Context
+	dctx, z.decommissionCancelers[idx] = context.WithCancel(GlobalContext)
+	z.poolMetaMutex.Unlock()
+
+	if err := z.decommissionInBackground(dctx, idx); err != nil {
+		logger.LogIf(GlobalContext, err)
+		logger.LogIf(GlobalContext, z.DecommissionFailed(dctx, idx))
+		return
+	}
+	// Complete the decommission..
+	logger.LogIf(GlobalContext, z.CompleteDecommission(dctx, idx))
+}
+
+func (z *erasureServerPools) IsSuspended(idx int) bool {
+	z.poolMetaMutex.Lock()
+	defer z.poolMetaMutex.Unlock()
+	return z.poolMeta.IsSuspended(idx)
+}
+
+// Decommission - start decommission session.
+func (z *erasureServerPools) Decommission(ctx context.Context, idx int) error {
+	if idx < 0 {
+		return errInvalidArgument
+	}
+
+	if z.SinglePool() {
+		return errInvalidArgument
+	}
+
+	// Make pool unwritable before decommissioning.
+	if err := z.StartDecommission(ctx, idx); err != nil {
+		return err
+	}
+
+	go z.doDecommissionInRoutine(ctx, idx)
+
+	// Successfully started decommissioning.
+	return nil
+}
+
+type decomError struct {
+	Err string
+}
+
+func (d decomError) Error() string {
+	return d.Err
+}
+
+type poolSpaceInfo struct {
+	Free  int64
+	Total int64
+	Used  int64
+}
+
+func (z *erasureServerPools) getDecommissionPoolSpaceInfo(idx int) (pi poolSpaceInfo, err error) {
+	if idx < 0 {
+		return pi, errInvalidArgument
+	}
+	if idx+1 > len(z.serverPools) {
+		return pi, errInvalidArgument
+	}
+	info, errs := z.serverPools[idx].StorageInfo(context.Background())
+	for _, err := range errs {
+		if err != nil {
+			return pi, errInvalidArgument
+		}
+	}
+	info.Backend = z.BackendInfo()
+	for _, disk := range info.Disks {
+		if disk.Healing {
+			return pi, decomError{
+				Err: fmt.Sprintf("%s drive is healing, decommission will not be started", disk.Endpoint),
+			}
+		}
+	}
+
+	usableTotal := int64(GetTotalUsableCapacity(info.Disks, info))
+	usableFree := int64(GetTotalUsableCapacityFree(info.Disks, info))
+	return poolSpaceInfo{
+		Total: usableTotal,
+		Free:  usableFree,
+		Used:  usableTotal - usableFree,
+	}, nil
+}
+
+func (z *erasureServerPools) Status(ctx context.Context, idx int) (PoolStatus, error) {
+	if idx < 0 {
+		return PoolStatus{}, errInvalidArgument
+	}
+
+	z.poolMetaMutex.RLock()
+	defer z.poolMetaMutex.RUnlock()
+
+	pi, err := z.getDecommissionPoolSpaceInfo(idx)
+	if err != nil {
+		return PoolStatus{}, errInvalidArgument
+	}
+
+	poolInfo := z.poolMeta.Pools[idx]
+	if poolInfo.Decommission != nil {
+		poolInfo.Decommission.TotalSize = pi.Total
+		poolInfo.Decommission.CurrentSize = poolInfo.Decommission.StartSize + poolInfo.Decommission.BytesDone
+	} else {
+		poolInfo.Decommission = &PoolDecommissionInfo{
+			TotalSize:   pi.Total,
+			CurrentSize: pi.Free,
+		}
+	}
+	return poolInfo, nil
+}
+
+func (z *erasureServerPools) ReloadPoolMeta(ctx context.Context) (err error) {
+	meta := poolMeta{}
+
+	if err = meta.load(ctx, z.serverPools[0], z.serverPools); err != nil {
+		return err
+	}
+
+	z.poolMetaMutex.Lock()
+	defer z.poolMetaMutex.Unlock()
+
+	z.poolMeta = meta
+	return nil
+}
+
+func (z *erasureServerPools) DecommissionCancel(ctx context.Context, idx int) (err error) {
+	if idx < 0 {
+		return errInvalidArgument
+	}
+
+	if z.SinglePool() {
+		return errInvalidArgument
+	}
+
+	z.poolMetaMutex.Lock()
+	defer z.poolMetaMutex.Unlock()
+
+	if z.poolMeta.DecommissionCancel(idx) {
+		z.decommissionCancelers[idx]() // cancel any active thread.
+		if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
+			return err
+		}
+		globalNotificationSys.ReloadPoolMeta(ctx)
+	}
+	return nil
+}
+
+func (z *erasureServerPools) DecommissionFailed(ctx context.Context, idx int) (err error) {
+	if idx < 0 {
+		return errInvalidArgument
+	}
+
+	if z.SinglePool() {
+		return errInvalidArgument
+	}
+
+	z.poolMetaMutex.Lock()
+	defer z.poolMetaMutex.Unlock()
+
+	if z.poolMeta.DecommissionFailed(idx) {
+		z.decommissionCancelers[idx]() // cancel any active thread.
+		if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
+			return err
+		}
+		globalNotificationSys.ReloadPoolMeta(ctx)
+	}
+	return nil
+}
+
+func (z *erasureServerPools) CompleteDecommission(ctx context.Context, idx int) (err error) {
+	if idx < 0 {
+		return errInvalidArgument
+	}
+
+	if z.SinglePool() {
+		return errInvalidArgument
+	}
+
+	z.poolMetaMutex.Lock()
+	defer z.poolMetaMutex.Unlock()
+
+	if z.poolMeta.DecommissionComplete(idx) {
+		if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
+			return err
+		}
+		globalNotificationSys.ReloadPoolMeta(ctx)
+	}
+	return nil
+}
+
+func (z *erasureServerPools) StartDecommission(ctx context.Context, idx int) (err error) {
+	if idx < 0 {
+		return errInvalidArgument
+	}
+
+	if z.SinglePool() {
+		return errInvalidArgument
+	}
+
+	buckets, err := z.ListBuckets(ctx)
+	if err != nil {
+		return err
+	}
+
+	// TODO: Support decommissioning transition tiers.
+	for _, bucket := range buckets {
+		if lc, err := globalLifecycleSys.Get(bucket.Name); err == nil {
+			if lc.HasTransition() {
+				return decomError{
+					Err: fmt.Sprintf("Bucket is part of transitioned tier %s: decommission is not allowed in Tier'd setups", bucket.Name),
+				}
+			}
+		}
+	}
+
+	// Buckets data are dispersed in multiple zones/sets, make
+	// sure to decommission the necessary metadata.
+	buckets = append(buckets, BucketInfo{
+		Name: pathJoin(minioMetaBucket, minioConfigPrefix),
+	})
+	buckets = append(buckets, BucketInfo{
+		Name: pathJoin(minioMetaBucket, bucketMetaPrefix),
+	})
+
+	var pool *erasureSets
+	for pidx := range z.serverPools {
+		if pidx == idx {
+			pool = z.serverPools[idx]
+			break
+		}
+	}
+
+	if pool == nil {
+		return errInvalidArgument
+	}
+
+	pi, err := z.getDecommissionPoolSpaceInfo(idx)
+	if err != nil {
+		return err
+	}
+
+	z.poolMetaMutex.Lock()
+	defer z.poolMetaMutex.Unlock()
+
+	if err = z.poolMeta.Decommission(idx, pi); err != nil {
+		return err
+	}
+	z.poolMeta.QueueBuckets(idx, buckets)
+	if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
+		return err
+	}
+	globalNotificationSys.ReloadPoolMeta(ctx)
+	return nil
+}

cmd/erasure-server-pool-decom_gen_test.go

@@ -0,0 +1,575 @@
+package cmd
+
+// Code generated by github.com/tinylib/msgp DO NOT EDIT.
+
+import (
+	"bytes"
+	"testing"
+
+	"github.com/tinylib/msgp/msgp"
+)
+
+func TestMarshalUnmarshalPoolDecommissionInfo(t *testing.T) {
+	v := PoolDecommissionInfo{}
+	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 BenchmarkMarshalMsgPoolDecommissionInfo(b *testing.B) {
+	v := PoolDecommissionInfo{}
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.MarshalMsg(nil)
+	}
+}
+
+func BenchmarkAppendMsgPoolDecommissionInfo(b *testing.B) {
+	v := PoolDecommissionInfo{}
+	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 BenchmarkUnmarshalPoolDecommissionInfo(b *testing.B) {
+	v := PoolDecommissionInfo{}
+	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 TestEncodeDecodePoolDecommissionInfo(t *testing.T) {
+	v := PoolDecommissionInfo{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+
+	m := v.Msgsize()
+	if buf.Len() > m {
+		t.Log("WARNING: TestEncodeDecodePoolDecommissionInfo Msgsize() is inaccurate")
+	}
+
+	vn := PoolDecommissionInfo{}
+	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 BenchmarkEncodePoolDecommissionInfo(b *testing.B) {
+	v := PoolDecommissionInfo{}
+	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 BenchmarkDecodePoolDecommissionInfo(b *testing.B) {
+	v := PoolDecommissionInfo{}
+	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 TestMarshalUnmarshalPoolStatus(t *testing.T) {
+	v := PoolStatus{}
+	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 BenchmarkMarshalMsgPoolStatus(b *testing.B) {
+	v := PoolStatus{}
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.MarshalMsg(nil)
+	}
+}
+
+func BenchmarkAppendMsgPoolStatus(b *testing.B) {
+	v := PoolStatus{}
+	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 BenchmarkUnmarshalPoolStatus(b *testing.B) {
+	v := PoolStatus{}
+	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 TestEncodeDecodePoolStatus(t *testing.T) {
+	v := PoolStatus{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+
+	m := v.Msgsize()
+	if buf.Len() > m {
+		t.Log("WARNING: TestEncodeDecodePoolStatus Msgsize() is inaccurate")
+	}
+
+	vn := PoolStatus{}
+	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 BenchmarkEncodePoolStatus(b *testing.B) {
+	v := PoolStatus{}
+	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 BenchmarkDecodePoolStatus(b *testing.B) {
+	v := PoolStatus{}
+	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 TestMarshalUnmarshaldecomError(t *testing.T) {
+	v := decomError{}
+	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 BenchmarkMarshalMsgdecomError(b *testing.B) {
+	v := decomError{}
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.MarshalMsg(nil)
+	}
+}
+
+func BenchmarkAppendMsgdecomError(b *testing.B) {
+	v := decomError{}
+	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 BenchmarkUnmarshaldecomError(b *testing.B) {
+	v := decomError{}
+	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 TestEncodeDecodedecomError(t *testing.T) {
+	v := decomError{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+
+	m := v.Msgsize()
+	if buf.Len() > m {
+		t.Log("WARNING: TestEncodeDecodedecomError Msgsize() is inaccurate")
+	}
+
+	vn := decomError{}
+	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 BenchmarkEncodedecomError(b *testing.B) {
+	v := decomError{}
+	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 BenchmarkDecodedecomError(b *testing.B) {
+	v := decomError{}
+	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 TestMarshalUnmarshalpoolMeta(t *testing.T) {
+	v := poolMeta{}
+	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 BenchmarkMarshalMsgpoolMeta(b *testing.B) {
+	v := poolMeta{}
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.MarshalMsg(nil)
+	}
+}
+
+func BenchmarkAppendMsgpoolMeta(b *testing.B) {
+	v := poolMeta{}
+	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 BenchmarkUnmarshalpoolMeta(b *testing.B) {
+	v := poolMeta{}
+	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 TestEncodeDecodepoolMeta(t *testing.T) {
+	v := poolMeta{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+
+	m := v.Msgsize()
+	if buf.Len() > m {
+		t.Log("WARNING: TestEncodeDecodepoolMeta Msgsize() is inaccurate")
+	}
+
+	vn := poolMeta{}
+	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 BenchmarkEncodepoolMeta(b *testing.B) {
+	v := poolMeta{}
+	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 BenchmarkDecodepoolMeta(b *testing.B) {
+	v := poolMeta{}
+	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 TestMarshalUnmarshalpoolSpaceInfo(t *testing.T) {
+	v := poolSpaceInfo{}
+	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 BenchmarkMarshalMsgpoolSpaceInfo(b *testing.B) {
+	v := poolSpaceInfo{}
+	b.ReportAllocs()
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		v.MarshalMsg(nil)
+	}
+}
+
+func BenchmarkAppendMsgpoolSpaceInfo(b *testing.B) {
+	v := poolSpaceInfo{}
+	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 BenchmarkUnmarshalpoolSpaceInfo(b *testing.B) {
+	v := poolSpaceInfo{}
+	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 TestEncodeDecodepoolSpaceInfo(t *testing.T) {
+	v := poolSpaceInfo{}
+	var buf bytes.Buffer
+	msgp.Encode(&buf, &v)
+
+	m := v.Msgsize()
+	if buf.Len() > m {
+		t.Log("WARNING: TestEncodeDecodepoolSpaceInfo Msgsize() is inaccurate")
+	}
+
+	vn := poolSpaceInfo{}
+	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 BenchmarkEncodepoolSpaceInfo(b *testing.B) {
+	v := poolSpaceInfo{}
+	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 BenchmarkDecodepoolSpaceInfo(b *testing.B) {
+	v := poolSpaceInfo{}
+	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)
+		}
+	}
+}