minio/cmd/admin-bucket-handlers.go
Anis Eleuch 95bf4a57b6
logging: Add subsystem to log API (#19002)
Create new code paths for multiple subsystems in the code. This will
make maintaing this easier later.

Also introduce bugLogIf() for errors that should not happen in the first
place.
2024-04-04 05:04:40 -07:00

1094 lines
32 KiB
Go

// 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 (
"bytes"
"encoding/base64"
"encoding/json"
"encoding/xml"
"errors"
"fmt"
"io"
"net/http"
"strings"
"time"
jsoniter "github.com/json-iterator/go"
"github.com/klauspost/compress/zip"
"github.com/minio/kms-go/kes"
"github.com/minio/madmin-go/v3"
"github.com/minio/minio-go/v7/pkg/tags"
"github.com/minio/minio/internal/bucket/lifecycle"
objectlock "github.com/minio/minio/internal/bucket/object/lock"
"github.com/minio/minio/internal/bucket/versioning"
"github.com/minio/minio/internal/event"
"github.com/minio/minio/internal/kms"
"github.com/minio/mux"
"github.com/minio/pkg/v2/policy"
)
const (
bucketQuotaConfigFile = "quota.json"
bucketTargetsFile = "bucket-targets.json"
)
// PutBucketQuotaConfigHandler - PUT Bucket quota configuration.
// ----------
// Places a quota configuration on the specified bucket. The quota
// specified in the quota configuration will be applied by default
// to enforce total quota for the specified bucket.
func (a adminAPIHandlers) PutBucketQuotaConfigHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
objectAPI, _ := validateAdminReq(ctx, w, r, policy.SetBucketQuotaAdminAction)
if objectAPI == nil {
return
}
vars := mux.Vars(r)
bucket := pathClean(vars["bucket"])
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
data, err := io.ReadAll(r.Body)
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
return
}
quotaConfig, err := parseBucketQuota(bucket, data)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
updatedAt, err := globalBucketMetadataSys.Update(ctx, bucket, bucketQuotaConfigFile, data)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
bucketMeta := madmin.SRBucketMeta{
Type: madmin.SRBucketMetaTypeQuotaConfig,
Bucket: bucket,
Quota: data,
UpdatedAt: updatedAt,
}
if quotaConfig.Size == 0 && quotaConfig.Quota == 0 {
bucketMeta.Quota = nil
}
// Call site replication hook.
replLogIf(ctx, globalSiteReplicationSys.BucketMetaHook(ctx, bucketMeta))
// Write success response.
writeSuccessResponseHeadersOnly(w)
}
// GetBucketQuotaConfigHandler - gets bucket quota configuration
func (a adminAPIHandlers) GetBucketQuotaConfigHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
objectAPI, _ := validateAdminReq(ctx, w, r, policy.GetBucketQuotaAdminAction)
if objectAPI == nil {
return
}
vars := mux.Vars(r)
bucket := pathClean(vars["bucket"])
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
config, _, err := globalBucketMetadataSys.GetQuotaConfig(ctx, bucket)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
configData, err := json.Marshal(config)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Write success response.
writeSuccessResponseJSON(w, configData)
}
// SetRemoteTargetHandler - sets a remote target for bucket
func (a adminAPIHandlers) SetRemoteTargetHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
vars := mux.Vars(r)
bucket := pathClean(vars["bucket"])
update := r.Form.Get("update") == "true"
// Get current object layer instance.
objectAPI, _ := validateAdminReq(ctx, w, r, policy.SetBucketTargetAction)
if objectAPI == nil {
return
}
// Check if bucket exists.
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
cred, _, s3Err := validateAdminSignature(ctx, r, "")
if s3Err != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(s3Err), r.URL)
return
}
password := cred.SecretKey
reqBytes, err := madmin.DecryptData(password, io.LimitReader(r.Body, r.ContentLength))
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminConfigBadJSON, err), r.URL)
return
}
var target madmin.BucketTarget
json := jsoniter.ConfigCompatibleWithStandardLibrary
if err = json.Unmarshal(reqBytes, &target); err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminConfigBadJSON, err), r.URL)
return
}
sameTarget, _ := isLocalHost(target.URL().Hostname(), target.URL().Port(), globalMinioPort)
if sameTarget && bucket == target.TargetBucket {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrBucketRemoteIdenticalToSource), r.URL)
return
}
target.SourceBucket = bucket
var ops []madmin.TargetUpdateType
if update {
ops = madmin.GetTargetUpdateOps(r.Form)
} else {
var exists bool // true if arn exists
target.Arn, exists = globalBucketTargetSys.getRemoteARN(bucket, &target, "")
if exists && target.Arn != "" { // return pre-existing ARN
data, err := json.Marshal(target.Arn)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Write success response.
writeSuccessResponseJSON(w, data)
return
}
}
if target.Arn == "" {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminConfigBadJSON, err), r.URL)
return
}
if globalSiteReplicationSys.isEnabled() && !update {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrRemoteTargetDenyAddError, err), r.URL)
return
}
if update {
// overlay the updates on existing target
tgt := globalBucketTargetSys.GetRemoteBucketTargetByArn(ctx, bucket, target.Arn)
if tgt.Empty() {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrRemoteTargetNotFoundError, err), r.URL)
return
}
for _, op := range ops {
switch op {
case madmin.CredentialsUpdateType:
if !globalSiteReplicationSys.isEnabled() {
// credentials update is possible only in bucket replication. User will never
// know the site replicator creds.
tgt.Credentials = target.Credentials
tgt.TargetBucket = target.TargetBucket
tgt.Secure = target.Secure
tgt.Endpoint = target.Endpoint
}
case madmin.SyncUpdateType:
tgt.ReplicationSync = target.ReplicationSync
case madmin.ProxyUpdateType:
tgt.DisableProxy = target.DisableProxy
case madmin.PathUpdateType:
tgt.Path = target.Path
case madmin.BandwidthLimitUpdateType:
tgt.BandwidthLimit = target.BandwidthLimit
case madmin.HealthCheckDurationUpdateType:
tgt.HealthCheckDuration = target.HealthCheckDuration
}
}
target = tgt
}
// enforce minimum bandwidth limit as 100MBps
if target.BandwidthLimit > 0 && target.BandwidthLimit < 100*1000*1000 {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrReplicationBandwidthLimitError, err), r.URL)
return
}
if err = globalBucketTargetSys.SetTarget(ctx, bucket, &target, update); err != nil {
switch err.(type) {
case RemoteTargetConnectionErr:
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrReplicationRemoteConnectionError, err), r.URL)
default:
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
}
return
}
targets, err := globalBucketTargetSys.ListBucketTargets(ctx, bucket)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
tgtBytes, err := json.Marshal(&targets)
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminConfigBadJSON, err), r.URL)
return
}
if _, err = globalBucketMetadataSys.Update(ctx, bucket, bucketTargetsFile, tgtBytes); err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
data, err := json.Marshal(target.Arn)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Write success response.
writeSuccessResponseJSON(w, data)
}
// ListRemoteTargetsHandler - lists remote target(s) for a bucket or gets a target
// for a particular ARN type
func (a adminAPIHandlers) ListRemoteTargetsHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
vars := mux.Vars(r)
bucket := pathClean(vars["bucket"])
arnType := vars["type"]
// Get current object layer instance.
objectAPI, _ := validateAdminReq(ctx, w, r, policy.GetBucketTargetAction)
if objectAPI == nil {
return
}
if bucket != "" {
// Check if bucket exists.
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
if _, err := globalBucketMetadataSys.GetBucketTargetsConfig(bucket); err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
}
targets := globalBucketTargetSys.ListTargets(ctx, bucket, arnType)
data, err := json.Marshal(targets)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Write success response.
writeSuccessResponseJSON(w, data)
}
// RemoveRemoteTargetHandler - removes a remote target for bucket with specified ARN
func (a adminAPIHandlers) RemoveRemoteTargetHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
vars := mux.Vars(r)
bucket := pathClean(vars["bucket"])
arn := vars["arn"]
// Get current object layer instance.
objectAPI, _ := validateAdminReq(ctx, w, r, policy.SetBucketTargetAction)
if objectAPI == nil {
return
}
// Check if bucket exists.
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
if err := globalBucketTargetSys.RemoveTarget(ctx, bucket, arn); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
targets, err := globalBucketTargetSys.ListBucketTargets(ctx, bucket)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
tgtBytes, err := json.Marshal(&targets)
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrAdminConfigBadJSON, err), r.URL)
return
}
if _, err = globalBucketMetadataSys.Update(ctx, bucket, bucketTargetsFile, tgtBytes); err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
// Write success response.
writeSuccessNoContent(w)
}
// ExportBucketMetadataHandler - exports all bucket metadata as a zipped file
func (a adminAPIHandlers) ExportBucketMetadataHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
bucket := pathClean(r.Form.Get("bucket"))
// Get current object layer instance.
objectAPI, _ := validateAdminReq(ctx, w, r, policy.ExportBucketMetadataAction)
if objectAPI == nil {
return
}
var (
buckets []BucketInfo
err error
)
if bucket != "" {
// Check if bucket exists.
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
buckets = append(buckets, BucketInfo{Name: bucket})
} else {
buckets, err = objectAPI.ListBuckets(ctx, BucketOptions{})
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
}
// Initialize a zip writer which will provide a zipped content
// of bucket metadata
zipWriter := zip.NewWriter(w)
defer zipWriter.Close()
rawDataFn := func(r io.Reader, filename string, sz int) {
header, zerr := zip.FileInfoHeader(dummyFileInfo{
name: filename,
size: int64(sz),
mode: 0o600,
modTime: time.Now(),
isDir: false,
sys: nil,
})
if zerr == nil {
header.Method = zip.Deflate
zwriter, zerr := zipWriter.CreateHeader(header)
if zerr == nil {
io.Copy(zwriter, r)
}
}
}
cfgFiles := []string{
bucketPolicyConfig,
bucketNotificationConfig,
bucketLifecycleConfig,
bucketSSEConfig,
bucketTaggingConfig,
bucketQuotaConfigFile,
objectLockConfig,
bucketVersioningConfig,
bucketReplicationConfig,
bucketTargetsFile,
}
for _, bi := range buckets {
for _, cfgFile := range cfgFiles {
cfgPath := pathJoin(bi.Name, cfgFile)
bucket := bi.Name
switch cfgFile {
case bucketNotificationConfig:
config, err := globalBucketMetadataSys.GetNotificationConfig(bucket)
if err != nil {
adminLogIf(ctx, err)
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketLifecycleConfig:
config, _, err := globalBucketMetadataSys.GetLifecycleConfig(bucket)
if err != nil {
if errors.Is(err, BucketLifecycleNotFound{Bucket: bucket}) {
continue
}
adminLogIf(ctx, err)
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketQuotaConfigFile:
config, _, err := globalBucketMetadataSys.GetQuotaConfig(ctx, bucket)
if err != nil {
if errors.Is(err, BucketQuotaConfigNotFound{Bucket: bucket}) {
continue
}
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
configData, err := json.Marshal(config)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketSSEConfig:
config, _, err := globalBucketMetadataSys.GetSSEConfig(bucket)
if err != nil {
if errors.Is(err, BucketSSEConfigNotFound{Bucket: bucket}) {
continue
}
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketTaggingConfig:
config, _, err := globalBucketMetadataSys.GetTaggingConfig(bucket)
if err != nil {
if errors.Is(err, BucketTaggingNotFound{Bucket: bucket}) {
continue
}
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case objectLockConfig:
config, _, err := globalBucketMetadataSys.GetObjectLockConfig(bucket)
if err != nil {
if errors.Is(err, BucketObjectLockConfigNotFound{Bucket: bucket}) {
continue
}
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketVersioningConfig:
config, _, err := globalBucketMetadataSys.GetVersioningConfig(bucket)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
// ignore empty versioning configs
if config.Status != versioning.Enabled && config.Status != versioning.Suspended {
continue
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketReplicationConfig:
config, _, err := globalBucketMetadataSys.GetReplicationConfig(ctx, bucket)
if err != nil {
if errors.Is(err, BucketReplicationConfigNotFound{Bucket: bucket}) {
continue
}
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
case bucketTargetsFile:
config, err := globalBucketMetadataSys.GetBucketTargetsConfig(bucket)
if err != nil {
if errors.Is(err, BucketRemoteTargetNotFound{Bucket: bucket}) {
continue
}
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
configData, err := xml.Marshal(config)
if err != nil {
writeErrorResponse(ctx, w, exportError(ctx, err, cfgFile, bucket), r.URL)
return
}
rawDataFn(bytes.NewReader(configData), cfgPath, len(configData))
}
}
}
}
type importMetaReport struct {
madmin.BucketMetaImportErrs
}
func (i *importMetaReport) SetStatus(bucket, fname string, err error) {
st := i.Buckets[bucket]
var errMsg string
if err != nil {
errMsg = err.Error()
}
switch fname {
case bucketPolicyConfig:
st.Policy = madmin.MetaStatus{IsSet: true, Err: errMsg}
case bucketNotificationConfig:
st.Notification = madmin.MetaStatus{IsSet: true, Err: errMsg}
case bucketLifecycleConfig:
st.Lifecycle = madmin.MetaStatus{IsSet: true, Err: errMsg}
case bucketSSEConfig:
st.SSEConfig = madmin.MetaStatus{IsSet: true, Err: errMsg}
case bucketTaggingConfig:
st.Tagging = madmin.MetaStatus{IsSet: true, Err: errMsg}
case bucketQuotaConfigFile:
st.Quota = madmin.MetaStatus{IsSet: true, Err: errMsg}
case objectLockConfig:
st.ObjectLock = madmin.MetaStatus{IsSet: true, Err: errMsg}
case bucketVersioningConfig:
st.Versioning = madmin.MetaStatus{IsSet: true, Err: errMsg}
default:
st.Err = errMsg
}
i.Buckets[bucket] = st
}
// ImportBucketMetadataHandler - imports all bucket metadata from a zipped file and overwrite bucket metadata config
// There are some caveats regarding the following:
// 1. object lock config - object lock should have been specified at time of bucket creation. Only default retention settings are imported here.
// 2. Replication config - is omitted from import as remote target credentials are not available from exported data for security reasons.
// 3. lifecycle config - if transition rules are present, tier name needs to have been defined.
func (a adminAPIHandlers) ImportBucketMetadataHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
// Get current object layer instance.
objectAPI, _ := validateAdminReq(ctx, w, r, policy.ImportBucketMetadataAction)
if objectAPI == nil {
return
}
data, err := io.ReadAll(r.Body)
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
return
}
reader := bytes.NewReader(data)
zr, err := zip.NewReader(reader, int64(len(data)))
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
return
}
rpt := importMetaReport{
madmin.BucketMetaImportErrs{
Buckets: make(map[string]madmin.BucketStatus, len(zr.File)),
},
}
bucketMap := make(map[string]*BucketMetadata, len(zr.File))
updatedAt := UTCNow()
for _, file := range zr.File {
slc := strings.Split(file.Name, slashSeparator)
if len(slc) != 2 { // expecting bucket/configfile in the zipfile
rpt.SetStatus(file.Name, "", fmt.Errorf("malformed zip - expecting format bucket/<config.json>"))
continue
}
bucket := slc[0]
meta, err := readBucketMetadata(ctx, objectAPI, bucket)
if err == nil {
bucketMap[bucket] = &meta
} else if err != errConfigNotFound {
rpt.SetStatus(bucket, "", err)
}
}
// import object lock config if any - order of import matters here.
for _, file := range zr.File {
slc := strings.Split(file.Name, slashSeparator)
if len(slc) != 2 { // expecting bucket/configfile in the zipfile
rpt.SetStatus(file.Name, "", fmt.Errorf("malformed zip - expecting format bucket/<config.json>"))
continue
}
bucket, fileName := slc[0], slc[1]
if fileName == objectLockConfig {
reader, err := file.Open()
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
config, err := objectlock.ParseObjectLockConfig(reader)
if err != nil {
rpt.SetStatus(bucket, fileName, fmt.Errorf("%s (%s)", errorCodes[ErrMalformedXML].Description, err))
continue
}
configData, err := xml.Marshal(config)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
if _, ok := bucketMap[bucket]; !ok {
opts := MakeBucketOptions{
LockEnabled: config.Enabled(),
ForceCreate: true, // ignore if it already exists
}
err = objectAPI.MakeBucket(ctx, bucket, opts)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
v, _ := globalBucketMetadataSys.Get(bucket)
bucketMap[bucket] = &v
}
bucketMap[bucket].ObjectLockConfigXML = configData
bucketMap[bucket].ObjectLockConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
}
}
// import versioning metadata
for _, file := range zr.File {
slc := strings.Split(file.Name, slashSeparator)
if len(slc) != 2 { // expecting bucket/configfile in the zipfile
rpt.SetStatus(file.Name, "", fmt.Errorf("malformed zip - expecting format bucket/<config.json>"))
continue
}
bucket, fileName := slc[0], slc[1]
if fileName == bucketVersioningConfig {
reader, err := file.Open()
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
v, err := versioning.ParseConfig(io.LimitReader(reader, maxBucketVersioningConfigSize))
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
if _, ok := bucketMap[bucket]; !ok {
if err = objectAPI.MakeBucket(ctx, bucket, MakeBucketOptions{
ForceCreate: true, // ignore if it already exists
}); err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
v, _ := globalBucketMetadataSys.Get(bucket)
bucketMap[bucket] = &v
}
if globalSiteReplicationSys.isEnabled() && v.Suspended() {
rpt.SetStatus(bucket, fileName, fmt.Errorf("Cluster replication is enabled for this site, so the versioning state cannot be suspended."))
continue
}
if rcfg, _ := globalBucketObjectLockSys.Get(bucket); rcfg.LockEnabled && v.Suspended() {
rpt.SetStatus(bucket, fileName, fmt.Errorf("An Object Lock configuration is present on this bucket, so the versioning state cannot be suspended."))
continue
}
if _, err := getReplicationConfig(ctx, bucket); err == nil && v.Suspended() {
rpt.SetStatus(bucket, fileName, fmt.Errorf("A replication configuration is present on this bucket, so the versioning state cannot be suspended."))
continue
}
configData, err := xml.Marshal(v)
if err != nil {
rpt.SetStatus(bucket, fileName, fmt.Errorf("%s (%s)", errorCodes[ErrMalformedXML].Description, err))
continue
}
bucketMap[bucket].VersioningConfigXML = configData
bucketMap[bucket].VersioningConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
}
}
for _, file := range zr.File {
reader, err := file.Open()
if err != nil {
rpt.SetStatus(file.Name, "", err)
continue
}
sz := file.FileInfo().Size()
slc := strings.Split(file.Name, slashSeparator)
if len(slc) != 2 { // expecting bucket/configfile in the zipfile
rpt.SetStatus(file.Name, "", fmt.Errorf("malformed zip - expecting format bucket/<config.json>"))
continue
}
bucket, fileName := slc[0], slc[1]
// create bucket if it does not exist yet.
if _, ok := bucketMap[bucket]; !ok {
err = objectAPI.MakeBucket(ctx, bucket, MakeBucketOptions{
ForceCreate: true, // ignore if it already exists
})
if err != nil {
rpt.SetStatus(bucket, "", err)
continue
}
v, _ := globalBucketMetadataSys.Get(bucket)
bucketMap[bucket] = &v
}
if _, ok := bucketMap[bucket]; !ok {
continue
}
switch fileName {
case bucketNotificationConfig:
config, err := event.ParseConfig(io.LimitReader(reader, sz), globalSite.Region, globalEventNotifier.targetList)
if err != nil {
rpt.SetStatus(bucket, fileName, fmt.Errorf("%s (%s)", errorCodes[ErrMalformedXML].Description, err))
continue
}
configData, err := xml.Marshal(config)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketMap[bucket].NotificationConfigXML = configData
rpt.SetStatus(bucket, fileName, nil)
case bucketPolicyConfig:
// Error out if Content-Length is beyond allowed size.
if sz > maxBucketPolicySize {
rpt.SetStatus(bucket, fileName, fmt.Errorf(ErrPolicyTooLarge.String()))
continue
}
bucketPolicyBytes, err := io.ReadAll(io.LimitReader(reader, sz))
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketPolicy, err := policy.ParseBucketPolicyConfig(bytes.NewReader(bucketPolicyBytes), bucket)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
// Version in policy must not be empty
if bucketPolicy.Version == "" {
rpt.SetStatus(bucket, fileName, fmt.Errorf(ErrPolicyInvalidVersion.String()))
continue
}
configData, err := json.Marshal(bucketPolicy)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketMap[bucket].PolicyConfigJSON = configData
bucketMap[bucket].PolicyConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
case bucketLifecycleConfig:
bucketLifecycle, err := lifecycle.ParseLifecycleConfig(io.LimitReader(reader, sz))
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
// Validate the received bucket policy document
if err = bucketLifecycle.Validate(); err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
// Validate the transition storage ARNs
if err = validateTransitionTier(bucketLifecycle); err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
configData, err := xml.Marshal(bucketLifecycle)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketMap[bucket].LifecycleConfigXML = configData
bucketMap[bucket].LifecycleConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
case bucketSSEConfig:
// Parse bucket encryption xml
encConfig, err := validateBucketSSEConfig(io.LimitReader(reader, maxBucketSSEConfigSize))
if err != nil {
rpt.SetStatus(bucket, fileName, fmt.Errorf("%s (%s)", errorCodes[ErrMalformedXML].Description, err))
continue
}
// Return error if KMS is not initialized
if GlobalKMS == nil {
rpt.SetStatus(bucket, fileName, fmt.Errorf("%s", errorCodes[ErrKMSNotConfigured].Description))
continue
}
kmsKey := encConfig.KeyID()
if kmsKey != "" {
kmsContext := kms.Context{"MinIO admin API": "ServerInfoHandler"} // Context for a test key operation
_, err := GlobalKMS.GenerateKey(ctx, kmsKey, kmsContext)
if err != nil {
if errors.Is(err, kes.ErrKeyNotFound) {
rpt.SetStatus(bucket, fileName, errKMSKeyNotFound)
continue
}
rpt.SetStatus(bucket, fileName, err)
continue
}
}
configData, err := xml.Marshal(encConfig)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketMap[bucket].EncryptionConfigXML = configData
bucketMap[bucket].EncryptionConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
case bucketTaggingConfig:
tags, err := tags.ParseBucketXML(io.LimitReader(reader, sz))
if err != nil {
rpt.SetStatus(bucket, fileName, fmt.Errorf("%s (%s)", errorCodes[ErrMalformedXML].Description, err))
continue
}
configData, err := xml.Marshal(tags)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketMap[bucket].TaggingConfigXML = configData
bucketMap[bucket].TaggingConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
case bucketQuotaConfigFile:
data, err := io.ReadAll(reader)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
_, err = parseBucketQuota(bucket, data)
if err != nil {
rpt.SetStatus(bucket, fileName, err)
continue
}
bucketMap[bucket].QuotaConfigJSON = data
bucketMap[bucket].QuotaConfigUpdatedAt = updatedAt
rpt.SetStatus(bucket, fileName, nil)
}
}
enc := func(b []byte) *string {
if b == nil {
return nil
}
v := base64.StdEncoding.EncodeToString(b)
return &v
}
for bucket, meta := range bucketMap {
err := globalBucketMetadataSys.save(ctx, *meta)
if err != nil {
rpt.SetStatus(bucket, "", err)
continue
}
// Call site replication hook.
if err = globalSiteReplicationSys.BucketMetaHook(ctx, madmin.SRBucketMeta{
Bucket: bucket,
Quota: meta.QuotaConfigJSON,
Policy: meta.PolicyConfigJSON,
Versioning: enc(meta.VersioningConfigXML),
Tags: enc(meta.TaggingConfigXML),
ObjectLockConfig: enc(meta.ObjectLockConfigXML),
SSEConfig: enc(meta.EncryptionConfigXML),
UpdatedAt: updatedAt,
}); err != nil {
rpt.SetStatus(bucket, "", err)
continue
}
}
rptData, err := json.Marshal(rpt.BucketMetaImportErrs)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
writeSuccessResponseJSON(w, rptData)
}
// ReplicationDiffHandler - POST returns info on unreplicated versions for a remote target ARN
// to the connected HTTP client.
func (a adminAPIHandlers) ReplicationDiffHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
vars := mux.Vars(r)
bucket := vars["bucket"]
objectAPI, _ := validateAdminReq(ctx, w, r, policy.ReplicationDiff)
if objectAPI == nil {
return
}
// Check if bucket exists.
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
opts := extractReplicateDiffOpts(r.Form)
if opts.ARN != "" {
tgt := globalBucketTargetSys.GetRemoteBucketTargetByArn(ctx, bucket, opts.ARN)
if tgt.Empty() {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErrWithErr(ErrInvalidRequest, fmt.Errorf("invalid arn : '%s'", opts.ARN)), r.URL)
return
}
}
keepAliveTicker := time.NewTicker(500 * time.Millisecond)
defer keepAliveTicker.Stop()
diffCh, err := getReplicationDiff(ctx, objectAPI, bucket, opts)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
enc := json.NewEncoder(w)
for {
select {
case entry, ok := <-diffCh:
if !ok {
return
}
if err := enc.Encode(entry); err != nil {
return
}
if len(diffCh) == 0 {
// Flush if nothing is queued
w.(http.Flusher).Flush()
}
case <-keepAliveTicker.C:
if len(diffCh) > 0 {
continue
}
if _, err := w.Write([]byte(" ")); err != nil {
return
}
w.(http.Flusher).Flush()
case <-ctx.Done():
return
}
}
}
// ReplicationMRFHandler - POST returns info on entries in the MRF backlog for a node or all nodes
func (a adminAPIHandlers) ReplicationMRFHandler(w http.ResponseWriter, r *http.Request) {
ctx := r.Context()
vars := mux.Vars(r)
bucket := vars["bucket"]
objectAPI, _ := validateAdminReq(ctx, w, r, policy.ReplicationDiff)
if objectAPI == nil {
return
}
// Check if bucket exists.
if bucket != "" {
if _, err := objectAPI.GetBucketInfo(ctx, bucket, BucketOptions{}); err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
}
q := r.Form
node := q.Get("node")
keepAliveTicker := time.NewTicker(500 * time.Millisecond)
defer keepAliveTicker.Stop()
mrfCh, err := globalNotificationSys.GetReplicationMRF(ctx, bucket, node)
if err != nil {
writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
return
}
enc := json.NewEncoder(w)
for {
select {
case entry, ok := <-mrfCh:
if !ok {
return
}
if err := enc.Encode(entry); err != nil {
return
}
if len(mrfCh) == 0 {
// Flush if nothing is queued
w.(http.Flusher).Flush()
}
case <-keepAliveTicker.C:
if len(mrfCh) > 0 {
continue
}
if _, err := w.Write([]byte(" ")); err != nil {
return
}
w.(http.Flusher).Flush()
case <-ctx.Done():
return
}
}
}