minio/cmd/admin-handlers.go
Anis Elleuch e05886561d
lock: Fix Refresh logic with multi resources lock (#13092)
A multi resources lock is a single lock UID with multiple associated
resources. This is created for example by multi objects delete
operation. This commit changes the behavior of Refresh() to iterate over
all locks having the same UID and refresh them.

Bonus: Fix showing top locks for multi delete objects
2021-08-27 13:07:55 -07:00

2280 lines
65 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 (
"context"
crand "crypto/rand"
"crypto/subtle"
"crypto/tls"
"encoding/json"
"fmt"
"hash/crc32"
"io"
"math/rand"
"net/http"
"net/url"
"os"
"path"
"regexp"
"runtime"
"sort"
"strconv"
"strings"
"time"
humanize "github.com/dustin/go-humanize"
"github.com/gorilla/mux"
"github.com/klauspost/compress/zip"
"github.com/minio/madmin-go"
"github.com/minio/minio/internal/dsync"
"github.com/minio/minio/internal/handlers"
xhttp "github.com/minio/minio/internal/http"
"github.com/minio/minio/internal/kms"
"github.com/minio/minio/internal/logger"
"github.com/minio/minio/internal/logger/message/log"
iampolicy "github.com/minio/pkg/iam/policy"
xnet "github.com/minio/pkg/net"
"github.com/secure-io/sio-go"
)
const (
maxEConfigJSONSize = 262272
)
// Only valid query params for mgmt admin APIs.
const (
mgmtBucket = "bucket"
mgmtPrefix = "prefix"
mgmtClientToken = "clientToken"
mgmtForceStart = "forceStart"
mgmtForceStop = "forceStop"
)
func updateServer(u *url.URL, sha256Sum []byte, lrTime time.Time, releaseInfo string, mode string) (us madmin.ServerUpdateStatus, err error) {
if err = doUpdate(u, lrTime, sha256Sum, releaseInfo, mode); err != nil {
return us, err
}
us.CurrentVersion = Version
us.UpdatedVersion = lrTime.Format(minioReleaseTagTimeLayout)
return us, nil
}
// ServerUpdateHandler - POST /minio/admin/v3/update?updateURL={updateURL}
// ----------
// updates all minio servers and restarts them gracefully.
func (a adminAPIHandlers) ServerUpdateHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "ServerUpdate")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.ServerUpdateAdminAction)
if objectAPI == nil {
return
}
if globalInplaceUpdateDisabled {
// if MINIO_UPDATE=off - inplace update is disabled, mostly in containers.
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrMethodNotAllowed), r.URL)
return
}
vars := mux.Vars(r)
updateURL := vars["updateURL"]
mode := getMinioMode()
if updateURL == "" {
updateURL = minioReleaseInfoURL
if runtime.GOOS == globalWindowsOSName {
updateURL = minioReleaseWindowsInfoURL
}
}
u, err := url.Parse(updateURL)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
content, err := downloadReleaseURL(u, updateTimeout, mode)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
sha256Sum, lrTime, releaseInfo, err := parseReleaseData(content)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
u.Path = path.Dir(u.Path) + SlashSeparator + releaseInfo
crTime, err := GetCurrentReleaseTime()
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
if lrTime.Sub(crTime) <= 0 {
updateStatus := madmin.ServerUpdateStatus{
CurrentVersion: Version,
UpdatedVersion: Version,
}
// Marshal API response
jsonBytes, err := json.Marshal(updateStatus)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
writeSuccessResponseJSON(w, jsonBytes)
return
}
for _, nerr := range globalNotificationSys.ServerUpdate(ctx, u, sha256Sum, lrTime, releaseInfo) {
if nerr.Err != nil {
logger.GetReqInfo(ctx).SetTags("peerAddress", nerr.Host.String())
logger.LogIf(ctx, nerr.Err)
err = fmt.Errorf("Server update failed, please do not restart the servers yet: failed with %w", nerr.Err)
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
}
updateStatus, err := updateServer(u, sha256Sum, lrTime, releaseInfo, mode)
if err != nil {
err = fmt.Errorf("Server update failed, please do not restart the servers yet: failed with %w", err)
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Marshal API response
jsonBytes, err := json.Marshal(updateStatus)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
writeSuccessResponseJSON(w, jsonBytes)
// Notify all other MinIO peers signal service.
for _, nerr := range globalNotificationSys.SignalService(serviceRestart) {
if nerr.Err != nil {
logger.GetReqInfo(ctx).SetTags("peerAddress", nerr.Host.String())
logger.LogIf(ctx, nerr.Err)
}
}
globalServiceSignalCh <- serviceRestart
}
// ServiceHandler - POST /minio/admin/v3/service?action={action}
// ----------
// restarts/stops minio server gracefully. In a distributed setup,
func (a adminAPIHandlers) ServiceHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "Service")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
vars := mux.Vars(r)
action := vars["action"]
var serviceSig serviceSignal
switch madmin.ServiceAction(action) {
case madmin.ServiceActionRestart:
serviceSig = serviceRestart
case madmin.ServiceActionStop:
serviceSig = serviceStop
default:
logger.LogIf(ctx, fmt.Errorf("Unrecognized service action %s requested", action), logger.Application)
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrMalformedPOSTRequest), r.URL)
return
}
var objectAPI ObjectLayer
switch serviceSig {
case serviceRestart:
objectAPI, _ = validateAdminReq(ctx, w, r, iampolicy.ServiceRestartAdminAction)
case serviceStop:
objectAPI, _ = validateAdminReq(ctx, w, r, iampolicy.ServiceStopAdminAction)
}
if objectAPI == nil {
return
}
// Notify all other MinIO peers signal service.
for _, nerr := range globalNotificationSys.SignalService(serviceSig) {
if nerr.Err != nil {
logger.GetReqInfo(ctx).SetTags("peerAddress", nerr.Host.String())
logger.LogIf(ctx, nerr.Err)
}
}
// Reply to the client before restarting, stopping MinIO server.
writeSuccessResponseHeadersOnly(w)
globalServiceSignalCh <- serviceSig
}
// ServerProperties holds some server information such as, version, region
// uptime, etc..
type ServerProperties struct {
Uptime int64 `json:"uptime"`
Version string `json:"version"`
CommitID string `json:"commitID"`
DeploymentID string `json:"deploymentID"`
Region string `json:"region"`
SQSARN []string `json:"sqsARN"`
}
// ServerConnStats holds transferred bytes from/to the server
type ServerConnStats struct {
TotalInputBytes uint64 `json:"transferred"`
TotalOutputBytes uint64 `json:"received"`
Throughput uint64 `json:"throughput,omitempty"`
S3InputBytes uint64 `json:"transferredS3"`
S3OutputBytes uint64 `json:"receivedS3"`
}
// ServerHTTPAPIStats holds total number of HTTP operations from/to the server,
// including the average duration the call was spent.
type ServerHTTPAPIStats struct {
APIStats map[string]int `json:"apiStats"`
}
// ServerHTTPStats holds all type of http operations performed to/from the server
// including their average execution time.
type ServerHTTPStats struct {
S3RequestsInQueue int32 `json:"s3RequestsInQueue"`
CurrentS3Requests ServerHTTPAPIStats `json:"currentS3Requests"`
TotalS3Requests ServerHTTPAPIStats `json:"totalS3Requests"`
TotalS3Errors ServerHTTPAPIStats `json:"totalS3Errors"`
TotalS3Canceled ServerHTTPAPIStats `json:"totalS3Canceled"`
TotalS3RejectedAuth uint64 `json:"totalS3RejectedAuth"`
TotalS3RejectedTime uint64 `json:"totalS3RejectedTime"`
TotalS3RejectedHeader uint64 `json:"totalS3RejectedHeader"`
TotalS3RejectedInvalid uint64 `json:"totalS3RejectedInvalid"`
}
// StorageInfoHandler - GET /minio/admin/v3/storageinfo
// ----------
// Get server information
func (a adminAPIHandlers) StorageInfoHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "StorageInfo")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.StorageInfoAdminAction)
if objectAPI == nil {
return
}
// ignores any errors here.
storageInfo, _ := objectAPI.StorageInfo(ctx)
// Collect any disk healing.
healing, _ := getAggregatedBackgroundHealState(ctx, nil)
healDisks := make(map[string]struct{}, len(healing.HealDisks))
for _, disk := range healing.HealDisks {
healDisks[disk] = struct{}{}
}
// find all disks which belong to each respective endpoints
for i, disk := range storageInfo.Disks {
if _, ok := healDisks[disk.Endpoint]; ok {
storageInfo.Disks[i].Healing = true
}
}
// Marshal API response
jsonBytes, err := json.Marshal(storageInfo)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Reply with storage information (across nodes in a
// distributed setup) as json.
writeSuccessResponseJSON(w, jsonBytes)
}
// DataUsageInfoHandler - GET /minio/admin/v3/datausage
// ----------
// Get server/cluster data usage info
func (a adminAPIHandlers) DataUsageInfoHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "DataUsageInfo")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.DataUsageInfoAdminAction)
if objectAPI == nil {
return
}
dataUsageInfo, err := loadDataUsageFromBackend(ctx, objectAPI)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
dataUsageInfoJSON, err := json.Marshal(dataUsageInfo)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
writeSuccessResponseJSON(w, dataUsageInfoJSON)
}
func lriToLockEntry(l lockRequesterInfo, resource, server string) *madmin.LockEntry {
entry := &madmin.LockEntry{
Timestamp: l.Timestamp,
Resource: resource,
ServerList: []string{server},
Source: l.Source,
Owner: l.Owner,
ID: l.UID,
Quorum: l.Quorum,
}
if l.Writer {
entry.Type = "WRITE"
} else {
entry.Type = "READ"
}
return entry
}
func topLockEntries(peerLocks []*PeerLocks, stale bool) madmin.LockEntries {
entryMap := make(map[string]*madmin.LockEntry)
for _, peerLock := range peerLocks {
if peerLock == nil {
continue
}
for k, v := range peerLock.Locks {
for _, lockReqInfo := range v {
if val, ok := entryMap[lockReqInfo.Name]; ok {
val.ServerList = append(val.ServerList, peerLock.Addr)
} else {
entryMap[lockReqInfo.Name] = lriToLockEntry(lockReqInfo, k, peerLock.Addr)
}
}
}
}
var lockEntries madmin.LockEntries
for _, v := range entryMap {
if stale {
lockEntries = append(lockEntries, *v)
continue
}
if len(v.ServerList) >= v.Quorum {
lockEntries = append(lockEntries, *v)
}
}
sort.Sort(lockEntries)
return lockEntries
}
// PeerLocks holds server information result of one node
type PeerLocks struct {
Addr string
Locks map[string][]lockRequesterInfo
}
// ForceUnlockHandler force unlocks requested resource
func (a adminAPIHandlers) ForceUnlockHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "ForceUnlock")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.ForceUnlockAdminAction)
if objectAPI == nil {
return
}
z, ok := objectAPI.(*erasureServerPools)
if !ok {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
return
}
vars := mux.Vars(r)
var args dsync.LockArgs
var lockers []dsync.NetLocker
for _, path := range strings.Split(vars["paths"], ",") {
if path == "" {
continue
}
args.Resources = append(args.Resources, path)
}
for _, lks := range z.serverPools[0].erasureLockers {
lockers = append(lockers, lks...)
}
for _, locker := range lockers {
locker.ForceUnlock(ctx, args)
}
}
// TopLocksHandler Get list of locks in use
func (a adminAPIHandlers) TopLocksHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "TopLocks")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.TopLocksAdminAction)
if objectAPI == nil {
return
}
count := 10 // by default list only top 10 entries
if countStr := r.Form.Get("count"); countStr != "" {
var err error
count, err = strconv.Atoi(countStr)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
}
stale := r.Form.Get("stale") == "true" // list also stale locks
peerLocks := globalNotificationSys.GetLocks(ctx, r)
topLocks := topLockEntries(peerLocks, stale)
// Marshal API response upto requested count.
if len(topLocks) > count && count > 0 {
topLocks = topLocks[:count]
}
jsonBytes, err := json.Marshal(topLocks)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Reply with storage information (across nodes in a
// distributed setup) as json.
writeSuccessResponseJSON(w, jsonBytes)
}
// StartProfilingResult contains the status of the starting
// profiling action in a given server
type StartProfilingResult struct {
NodeName string `json:"nodeName"`
Success bool `json:"success"`
Error string `json:"error"`
}
// StartProfilingHandler - POST /minio/admin/v3/profiling/start?profilerType={profilerType}
// ----------
// Enable server profiling
func (a adminAPIHandlers) StartProfilingHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "StartProfiling")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
// Validate request signature.
_, adminAPIErr := checkAdminRequestAuth(ctx, r, iampolicy.ProfilingAdminAction, "")
if adminAPIErr != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(adminAPIErr), r.URL)
return
}
if globalNotificationSys == nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
return
}
vars := mux.Vars(r)
profiles := strings.Split(vars["profilerType"], ",")
thisAddr, err := xnet.ParseHost(globalLocalNodeName)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
globalProfilerMu.Lock()
defer globalProfilerMu.Unlock()
if globalProfiler == nil {
globalProfiler = make(map[string]minioProfiler, 10)
}
// Stop profiler of all types if already running
for k, v := range globalProfiler {
for _, p := range profiles {
if p == k {
v.Stop()
delete(globalProfiler, k)
}
}
}
// Start profiling on remote servers.
var hostErrs []NotificationPeerErr
for _, profiler := range profiles {
hostErrs = append(hostErrs, globalNotificationSys.StartProfiling(profiler)...)
// Start profiling locally as well.
prof, err := startProfiler(profiler)
if err != nil {
hostErrs = append(hostErrs, NotificationPeerErr{
Host: *thisAddr,
Err: err,
})
} else {
globalProfiler[profiler] = prof
hostErrs = append(hostErrs, NotificationPeerErr{
Host: *thisAddr,
})
}
}
var startProfilingResult []StartProfilingResult
for _, nerr := range hostErrs {
result := StartProfilingResult{NodeName: nerr.Host.String()}
if nerr.Err != nil {
result.Error = nerr.Err.Error()
} else {
result.Success = true
}
startProfilingResult = append(startProfilingResult, result)
}
// Create JSON result and send it to the client
startProfilingResultInBytes, err := json.Marshal(startProfilingResult)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
writeSuccessResponseJSON(w, startProfilingResultInBytes)
}
// dummyFileInfo represents a dummy representation of a profile data file
// present only in memory, it helps to generate the zip stream.
type dummyFileInfo struct {
name string
size int64
mode os.FileMode
modTime time.Time
isDir bool
sys interface{}
}
func (f dummyFileInfo) Name() string { return f.name }
func (f dummyFileInfo) Size() int64 { return f.size }
func (f dummyFileInfo) Mode() os.FileMode { return f.mode }
func (f dummyFileInfo) ModTime() time.Time { return f.modTime }
func (f dummyFileInfo) IsDir() bool { return f.isDir }
func (f dummyFileInfo) Sys() interface{} { return f.sys }
// DownloadProfilingHandler - POST /minio/admin/v3/profiling/download
// ----------
// Download profiling information of all nodes in a zip format
func (a adminAPIHandlers) DownloadProfilingHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "DownloadProfiling")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
// Validate request signature.
_, adminAPIErr := checkAdminRequestAuth(ctx, r, iampolicy.ProfilingAdminAction, "")
if adminAPIErr != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(adminAPIErr), r.URL)
return
}
if globalNotificationSys == nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrServerNotInitialized), r.URL)
return
}
if !globalNotificationSys.DownloadProfilingData(ctx, w) {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrAdminProfilerNotEnabled), r.URL)
return
}
}
type healInitParams struct {
bucket, objPrefix string
hs madmin.HealOpts
clientToken string
forceStart, forceStop bool
}
// extractHealInitParams - Validates params for heal init API.
func extractHealInitParams(vars map[string]string, qParms url.Values, r io.Reader) (hip healInitParams, err APIErrorCode) {
hip.bucket = vars[mgmtBucket]
hip.objPrefix = vars[mgmtPrefix]
if hip.bucket == "" {
if hip.objPrefix != "" {
// Bucket is required if object-prefix is given
err = ErrHealMissingBucket
return
}
} else if isReservedOrInvalidBucket(hip.bucket, false) {
err = ErrInvalidBucketName
return
}
// empty prefix is valid.
if !IsValidObjectPrefix(hip.objPrefix) {
err = ErrInvalidObjectName
return
}
if len(qParms[mgmtClientToken]) > 0 {
hip.clientToken = qParms[mgmtClientToken][0]
}
if _, ok := qParms[mgmtForceStart]; ok {
hip.forceStart = true
}
if _, ok := qParms[mgmtForceStop]; ok {
hip.forceStop = true
}
// Invalid request conditions:
//
// Cannot have both forceStart and forceStop in the same
// request; If clientToken is provided, request can only be
// to continue receiving logs, so it cannot be start or
// stop;
if (hip.forceStart && hip.forceStop) ||
(hip.clientToken != "" && (hip.forceStart || hip.forceStop)) {
err = ErrInvalidRequest
return
}
// ignore body if clientToken is provided
if hip.clientToken == "" {
jerr := json.NewDecoder(r).Decode(&hip.hs)
if jerr != nil {
logger.LogIf(GlobalContext, jerr, logger.Application)
err = ErrRequestBodyParse
return
}
}
err = ErrNone
return
}
// HealHandler - POST /minio/admin/v3/heal/
// -----------
// Start heal processing and return heal status items.
//
// On a successful heal sequence start, a unique client token is
// returned. Subsequent requests to this endpoint providing the client
// token will receive heal status records from the running heal
// sequence.
//
// If no client token is provided, and a heal sequence is in progress
// an error is returned with information about the running heal
// sequence. However, if the force-start flag is provided, the server
// aborts the running heal sequence and starts a new one.
func (a adminAPIHandlers) HealHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "Heal")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.HealAdminAction)
if objectAPI == nil {
return
}
// Check if this setup has an erasure coded backend.
if !globalIsErasure {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrHealNotImplemented), r.URL)
return
}
hip, errCode := extractHealInitParams(mux.Vars(r), r.Form, r.Body)
if errCode != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(errCode), r.URL)
return
}
// Analyze the heal token and route the request accordingly
token, success := proxyRequestByToken(ctx, w, r, hip.clientToken)
if success {
return
}
hip.clientToken = token
// if request was not successful, try this server locally if token
// is not found the call will fail anyways. if token is empty
// try this server to generate a new token.
type healResp struct {
respBytes []byte
apiErr APIError
errBody string
}
// Define a closure to start sending whitespace to client
// after 10s unless a response item comes in
keepConnLive := func(w http.ResponseWriter, r *http.Request, respCh chan healResp) {
ticker := time.NewTicker(time.Second * 10)
defer ticker.Stop()
started := false
forLoop:
for {
select {
case <-r.Context().Done():
return
case <-ticker.C:
if !started {
// Start writing response to client
started = true
setCommonHeaders(w)
setEventStreamHeaders(w)
// Set 200 OK status
w.WriteHeader(200)
}
// Send whitespace and keep connection open
w.Write([]byte(" "))
w.(http.Flusher).Flush()
case hr := <-respCh:
switch hr.apiErr {
case noError:
if started {
w.Write(hr.respBytes)
w.(http.Flusher).Flush()
} else {
writeSuccessResponseJSON(w, hr.respBytes)
}
default:
var errorRespJSON []byte
if hr.errBody == "" {
errorRespJSON = encodeResponseJSON(getAPIErrorResponse(ctx, hr.apiErr,
r.URL.Path, w.Header().Get(xhttp.AmzRequestID),
globalDeploymentID))
} else {
errorRespJSON = encodeResponseJSON(APIErrorResponse{
Code: hr.apiErr.Code,
Message: hr.errBody,
Resource: r.URL.Path,
RequestID: w.Header().Get(xhttp.AmzRequestID),
HostID: globalDeploymentID,
})
}
if !started {
setCommonHeaders(w)
w.Header().Set(xhttp.ContentType, string(mimeJSON))
w.WriteHeader(hr.apiErr.HTTPStatusCode)
}
w.Write(errorRespJSON)
w.(http.Flusher).Flush()
}
break forLoop
}
}
}
healPath := pathJoin(hip.bucket, hip.objPrefix)
if hip.clientToken == "" && !hip.forceStart && !hip.forceStop {
nh, exists := globalAllHealState.getHealSequence(healPath)
if exists && !nh.hasEnded() && len(nh.currentStatus.Items) > 0 {
clientToken := nh.clientToken
if globalIsDistErasure {
clientToken = fmt.Sprintf("%s@%d", nh.clientToken, GetProxyEndpointLocalIndex(globalProxyEndpoints))
}
b, err := json.Marshal(madmin.HealStartSuccess{
ClientToken: clientToken,
ClientAddress: nh.clientAddress,
StartTime: nh.startTime,
})
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Client token not specified but a heal sequence exists on a path,
// Send the token back to client.
writeSuccessResponseJSON(w, b)
return
}
}
if hip.clientToken != "" && !hip.forceStart && !hip.forceStop {
// Since clientToken is given, fetch heal status from running
// heal sequence.
respBytes, errCode := globalAllHealState.PopHealStatusJSON(
healPath, hip.clientToken)
if errCode != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(errCode), r.URL)
} else {
writeSuccessResponseJSON(w, respBytes)
}
return
}
respCh := make(chan healResp)
switch {
case hip.forceStop:
go func() {
respBytes, apiErr := globalAllHealState.stopHealSequence(healPath)
hr := healResp{respBytes: respBytes, apiErr: apiErr}
respCh <- hr
}()
case hip.clientToken == "":
nh := newHealSequence(GlobalContext, hip.bucket, hip.objPrefix, handlers.GetSourceIP(r), hip.hs, hip.forceStart)
go func() {
respBytes, apiErr, errMsg := globalAllHealState.LaunchNewHealSequence(nh, objectAPI)
hr := healResp{respBytes, apiErr, errMsg}
respCh <- hr
}()
}
// Due to the force-starting functionality, the Launch
// call above can take a long time - to keep the
// connection alive, we start sending whitespace
keepConnLive(w, r, respCh)
}
// getAggregatedBackgroundHealState returns the heal state of disks.
// If no ObjectLayer is provided no set status is returned.
func getAggregatedBackgroundHealState(ctx context.Context, o ObjectLayer) (madmin.BgHealState, error) {
// Get local heal status first
bgHealStates, ok := getBackgroundHealStatus(ctx, o)
if !ok {
return bgHealStates, errServerNotInitialized
}
if globalIsDistErasure {
// Get heal status from other peers
peersHealStates, nerrs := globalNotificationSys.BackgroundHealStatus()
var errCount int
for _, nerr := range nerrs {
if nerr.Err != nil {
logger.LogIf(ctx, nerr.Err)
errCount++
}
}
if errCount == len(nerrs) {
return madmin.BgHealState{}, fmt.Errorf("all remote servers failed to report heal status, cluster is unhealthy")
}
bgHealStates.Merge(peersHealStates...)
}
return bgHealStates, nil
}
func (a adminAPIHandlers) BackgroundHealStatusHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "HealBackgroundStatus")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.HealAdminAction)
if objectAPI == nil {
return
}
// Check if this setup has an erasure coded backend.
if !globalIsErasure {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrHealNotImplemented), r.URL)
return
}
aggregateHealStateResult, err := getAggregatedBackgroundHealState(r.Context(), objectAPI)
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
if err := json.NewEncoder(w).Encode(aggregateHealStateResult); err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
}
func (a adminAPIHandlers) SpeedtestHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "SpeedtestHandler")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.HealAdminAction)
if objectAPI == nil {
return
}
if !globalIsErasure {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
return
}
sizeStr := r.Form.Get(peerRESTSize)
durationStr := r.Form.Get(peerRESTDuration)
concurrentStr := r.Form.Get(peerRESTConcurrent)
size, err := strconv.Atoi(sizeStr)
if err != nil {
size = 64 * humanize.MiByte
}
concurrent, err := strconv.Atoi(concurrentStr)
if err != nil {
concurrent = 32
}
duration, err := time.ParseDuration(durationStr)
if err != nil {
duration = time.Second * 10
}
results := globalNotificationSys.Speedtest(ctx, size, concurrent, duration)
if err := json.NewEncoder(w).Encode(results); err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
objectAPI.DeleteBucket(ctx, pathJoin(minioMetaSpeedTestBucket, minioMetaSpeedTestBucketPrefix), true)
w.(http.Flusher).Flush()
}
// Admin API errors
const (
AdminUpdateUnexpectedFailure = "XMinioAdminUpdateUnexpectedFailure"
AdminUpdateURLNotReachable = "XMinioAdminUpdateURLNotReachable"
AdminUpdateApplyFailure = "XMinioAdminUpdateApplyFailure"
)
// Returns true if the madmin.TraceInfo should be traced,
// false if certain conditions are not met.
// - input entry is not of the type *madmin.TraceInfo*
// - errOnly entries are to be traced, not status code 2xx, 3xx.
// - madmin.TraceInfo type is asked by opts
func mustTrace(entry interface{}, opts madmin.ServiceTraceOpts) (shouldTrace bool) {
trcInfo, ok := entry.(madmin.TraceInfo)
if !ok {
return false
}
// Override shouldTrace decision with errOnly filtering
defer func() {
if shouldTrace && opts.OnlyErrors {
shouldTrace = trcInfo.RespInfo.StatusCode >= http.StatusBadRequest
}
}()
if opts.Threshold > 0 {
var latency time.Duration
switch trcInfo.TraceType {
case madmin.TraceOS:
latency = trcInfo.OSStats.Duration
case madmin.TraceStorage:
latency = trcInfo.StorageStats.Duration
case madmin.TraceHTTP:
latency = trcInfo.CallStats.Latency
}
if latency < opts.Threshold {
return false
}
}
if opts.Internal && trcInfo.TraceType == madmin.TraceHTTP && HasPrefix(trcInfo.ReqInfo.Path, minioReservedBucketPath+SlashSeparator) {
return true
}
if opts.S3 && trcInfo.TraceType == madmin.TraceHTTP && !HasPrefix(trcInfo.ReqInfo.Path, minioReservedBucketPath+SlashSeparator) {
return true
}
if opts.Storage && trcInfo.TraceType == madmin.TraceStorage {
return true
}
return opts.OS && trcInfo.TraceType == madmin.TraceOS
}
func extractTraceOptions(r *http.Request) (opts madmin.ServiceTraceOpts, err error) {
q := r.Form
opts.OnlyErrors = q.Get("err") == "true"
opts.S3 = q.Get("s3") == "true"
opts.Internal = q.Get("internal") == "true"
opts.Storage = q.Get("storage") == "true"
opts.OS = q.Get("os") == "true"
// Support deprecated 'all' query
if q.Get("all") == "true" {
opts.S3 = true
opts.Internal = true
opts.Storage = true
opts.OS = true
}
if t := q.Get("threshold"); t != "" {
d, err := time.ParseDuration(t)
if err != nil {
return opts, err
}
opts.Threshold = d
}
return
}
// TraceHandler - POST /minio/admin/v3/trace
// ----------
// The handler sends http trace to the connected HTTP client.
func (a adminAPIHandlers) TraceHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "HTTPTrace")
// Validate request signature.
_, adminAPIErr := checkAdminRequestAuth(ctx, r, iampolicy.TraceAdminAction, "")
if adminAPIErr != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(adminAPIErr), r.URL)
return
}
traceOpts, err := extractTraceOptions(r)
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
return
}
setEventStreamHeaders(w)
// Trace Publisher and peer-trace-client uses nonblocking send and hence does not wait for slow receivers.
// Use buffered channel to take care of burst sends or slow w.Write()
traceCh := make(chan interface{}, 4000)
peers, _ := newPeerRestClients(globalEndpoints)
globalTrace.Subscribe(traceCh, ctx.Done(), func(entry interface{}) bool {
return mustTrace(entry, traceOpts)
})
for _, peer := range peers {
if peer == nil {
continue
}
peer.Trace(traceCh, ctx.Done(), traceOpts)
}
keepAliveTicker := time.NewTicker(500 * time.Millisecond)
defer keepAliveTicker.Stop()
enc := json.NewEncoder(w)
for {
select {
case entry := <-traceCh:
if err := enc.Encode(entry); err != nil {
return
}
if len(traceCh) == 0 {
// Flush if nothing is queued
w.(http.Flusher).Flush()
}
case <-keepAliveTicker.C:
if len(traceCh) > 0 {
continue
}
if _, err := w.Write([]byte(" ")); err != nil {
return
}
w.(http.Flusher).Flush()
case <-ctx.Done():
return
}
}
}
// The handler sends console logs to the connected HTTP client.
func (a adminAPIHandlers) ConsoleLogHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "ConsoleLog")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.ConsoleLogAdminAction)
if objectAPI == nil {
return
}
node := r.Form.Get("node")
// limit buffered console entries if client requested it.
limitStr := r.Form.Get("limit")
limitLines, err := strconv.Atoi(limitStr)
if err != nil {
limitLines = 10
}
logKind := r.Form.Get("logType")
if logKind == "" {
logKind = string(logger.All)
}
logKind = strings.ToUpper(logKind)
// Avoid reusing tcp connection if read timeout is hit
// This is needed to make r.Context().Done() work as
// expected in case of read timeout
w.Header().Set("Connection", "close")
setEventStreamHeaders(w)
logCh := make(chan interface{}, 4000)
peers, _ := newPeerRestClients(globalEndpoints)
globalConsoleSys.Subscribe(logCh, ctx.Done(), node, limitLines, logKind, nil)
for _, peer := range peers {
if peer == nil {
continue
}
if node == "" || strings.EqualFold(peer.host.Name, node) {
peer.ConsoleLog(logCh, ctx.Done())
}
}
enc := json.NewEncoder(w)
keepAliveTicker := time.NewTicker(500 * time.Millisecond)
defer keepAliveTicker.Stop()
for {
select {
case entry := <-logCh:
log, ok := entry.(log.Info)
if ok && log.SendLog(node, logKind) {
if err := enc.Encode(log); err != nil {
return
}
}
if len(logCh) == 0 {
// Flush if nothing is queued
w.(http.Flusher).Flush()
}
case <-keepAliveTicker.C:
if len(logCh) > 0 {
continue
}
if _, err := w.Write([]byte(" ")); err != nil {
return
}
w.(http.Flusher).Flush()
case <-ctx.Done():
return
}
}
}
// KMSCreateKeyHandler - POST /minio/admin/v3/kms/key/create?key-id=<master-key-id>
func (a adminAPIHandlers) KMSCreateKeyHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "KMSCreateKey")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.KMSCreateKeyAdminAction)
if objectAPI == nil {
return
}
if GlobalKMS == nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrKMSNotConfigured), r.URL)
return
}
if err := GlobalKMS.CreateKey(r.Form.Get("key-id")); err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
writeSuccessResponseHeadersOnly(w)
}
// KMSKeyStatusHandler - GET /minio/admin/v3/kms/status
func (a adminAPIHandlers) KMSStatusHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "KMSStatus")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.KMSKeyStatusAdminAction)
if objectAPI == nil {
return
}
if GlobalKMS == nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrKMSNotConfigured), r.URL)
return
}
stat, err := GlobalKMS.Stat()
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
status := madmin.KMSStatus{
Name: stat.Name,
DefaultKeyID: stat.DefaultKey,
Endpoints: make(map[string]madmin.ItemState, len(stat.Endpoints)),
}
for _, endpoint := range stat.Endpoints {
status.Endpoints[endpoint] = madmin.ItemOnline // TODO(aead): Implement an online check for mTLS
}
resp, err := json.Marshal(status)
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
writeSuccessResponseJSON(w, resp)
}
// KMSKeyStatusHandler - GET /minio/admin/v3/kms/key/status?key-id=<master-key-id>
func (a adminAPIHandlers) KMSKeyStatusHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "KMSKeyStatus")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.KMSKeyStatusAdminAction)
if objectAPI == nil {
return
}
if GlobalKMS == nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrKMSNotConfigured), r.URL)
return
}
stat, err := GlobalKMS.Stat()
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
keyID := r.Form.Get("key-id")
if keyID == "" {
keyID = stat.DefaultKey
}
var response = madmin.KMSKeyStatus{
KeyID: keyID,
}
kmsContext := kms.Context{"MinIO admin API": "KMSKeyStatusHandler"} // Context for a test key operation
// 1. Generate a new key using the KMS.
key, err := GlobalKMS.GenerateKey(keyID, kmsContext)
if err != nil {
response.EncryptionErr = err.Error()
resp, err := json.Marshal(response)
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
writeSuccessResponseJSON(w, resp)
return
}
// 2. Verify that we can indeed decrypt the (encrypted) key
decryptedKey, err := GlobalKMS.DecryptKey(key.KeyID, key.Ciphertext, kmsContext)
if err != nil {
response.DecryptionErr = err.Error()
resp, err := json.Marshal(response)
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
writeSuccessResponseJSON(w, resp)
return
}
// 3. Compare generated key with decrypted key
if subtle.ConstantTimeCompare(key.Plaintext, decryptedKey) != 1 {
response.DecryptionErr = "The generated and the decrypted data key do not match"
resp, err := json.Marshal(response)
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
writeSuccessResponseJSON(w, resp)
return
}
resp, err := json.Marshal(response)
if err != nil {
writeCustomErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), err.Error(), r.URL)
return
}
writeSuccessResponseJSON(w, resp)
}
func getServerInfo(ctx context.Context, r *http.Request) madmin.InfoMessage {
kmsStat := fetchKMSStatus()
ldap := madmin.LDAP{}
if globalLDAPConfig.Enabled {
ldapConn, err := globalLDAPConfig.Connect()
if err != nil {
ldap.Status = string(madmin.ItemOffline)
} else if ldapConn == nil {
ldap.Status = "Not Configured"
} else {
// Close ldap connection to avoid leaks.
ldapConn.Close()
ldap.Status = string(madmin.ItemOnline)
}
}
log, audit := fetchLoggerInfo()
// Get the notification target info
notifyTarget := fetchLambdaInfo()
local := getLocalServerProperty(globalEndpoints, r)
servers := globalNotificationSys.ServerInfo()
servers = append(servers, local)
assignPoolNumbers(servers)
var backend interface{}
mode := madmin.ItemInitializing
buckets := madmin.Buckets{}
objects := madmin.Objects{}
usage := madmin.Usage{}
objectAPI := newObjectLayerFn()
if objectAPI != nil {
mode = madmin.ItemOnline
// Load data usage
dataUsageInfo, err := loadDataUsageFromBackend(ctx, objectAPI)
if err == nil {
buckets = madmin.Buckets{Count: dataUsageInfo.BucketsCount}
objects = madmin.Objects{Count: dataUsageInfo.ObjectsTotalCount}
usage = madmin.Usage{Size: dataUsageInfo.ObjectsTotalSize}
} else {
buckets = madmin.Buckets{Error: err.Error()}
objects = madmin.Objects{Error: err.Error()}
usage = madmin.Usage{Error: err.Error()}
}
// Fetching the backend information
backendInfo := objectAPI.BackendInfo()
if backendInfo.Type == madmin.Erasure {
// Calculate the number of online/offline disks of all nodes
var allDisks []madmin.Disk
for _, s := range servers {
allDisks = append(allDisks, s.Disks...)
}
onlineDisks, offlineDisks := getOnlineOfflineDisksStats(allDisks)
backend = madmin.ErasureBackend{
Type: madmin.ErasureType,
OnlineDisks: onlineDisks.Sum(),
OfflineDisks: offlineDisks.Sum(),
StandardSCParity: backendInfo.StandardSCParity,
RRSCParity: backendInfo.RRSCParity,
}
} else {
backend = madmin.FSBackend{
Type: madmin.FsType,
}
}
}
domain := globalDomainNames
services := madmin.Services{
KMS: kmsStat,
LDAP: ldap,
Logger: log,
Audit: audit,
Notifications: notifyTarget,
}
return madmin.InfoMessage{
Mode: string(mode),
Domain: domain,
Region: globalServerRegion,
SQSARN: globalNotificationSys.GetARNList(false),
DeploymentID: globalDeploymentID,
Buckets: buckets,
Objects: objects,
Usage: usage,
Services: services,
Backend: backend,
Servers: servers,
}
}
// HealthInfoHandler - GET /minio/admin/v3/healthinfo
// ----------
// Get server health info
func (a adminAPIHandlers) HealthInfoHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "HealthInfo")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
objectAPI, _ := validateAdminReq(ctx, w, r, iampolicy.HealthInfoAdminAction)
if objectAPI == nil {
return
}
query := r.Form
healthInfo := madmin.HealthInfo{Version: madmin.HealthInfoVersion}
healthInfoCh := make(chan madmin.HealthInfo)
enc := json.NewEncoder(w)
partialWrite := func(oinfo madmin.HealthInfo) {
healthInfoCh <- oinfo
}
setCommonHeaders(w)
setEventStreamHeaders(w)
w.WriteHeader(http.StatusOK)
errResp := func(err error) {
errorResponse := getAPIErrorResponse(ctx, toAdminAPIErr(ctx, err), r.URL.String(),
w.Header().Get(xhttp.AmzRequestID), globalDeploymentID)
encodedErrorResponse := encodeResponse(errorResponse)
healthInfo.Error = string(encodedErrorResponse)
logger.LogIf(ctx, enc.Encode(healthInfo))
}
deadline := 1 * time.Hour
if dstr := r.Form.Get("deadline"); dstr != "" {
var err error
deadline, err = time.ParseDuration(dstr)
if err != nil {
errResp(err)
return
}
}
deadlinedCtx, deadlineCancel := context.WithTimeout(ctx, deadline)
defer deadlineCancel()
nsLock := objectAPI.NewNSLock(minioMetaBucket, "health-check-in-progress")
lkctx, err := nsLock.GetLock(ctx, newDynamicTimeout(deadline, deadline))
if err != nil { // returns a locked lock
errResp(err)
return
}
defer nsLock.Unlock(lkctx.Cancel)
hostAnonymizer := createHostAnonymizer()
// anonAddr - Anonymizes hosts in given input string.
anonAddr := func(addr string) string {
newAddr, found := hostAnonymizer[addr]
if found {
return newAddr
}
// If we reach here, it means that the given addr doesn't contain any of the hosts.
// Return it as is. Can happen for drive paths in non-distributed mode
return addr
}
// anonymizedAddr - Updated the addr of the node info with anonymized one
anonymizeAddr := func(info madmin.NodeInfo) {
info.SetAddr(anonAddr(info.GetAddr()))
}
getAndWriteCPUs := func() {
if query.Get("syscpu") == "true" {
localCPUInfo := madmin.GetCPUs(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localCPUInfo)
healthInfo.Sys.CPUInfo = append(healthInfo.Sys.CPUInfo, localCPUInfo)
peerCPUInfo := globalNotificationSys.GetCPUs(deadlinedCtx)
for _, cpuInfo := range peerCPUInfo {
anonymizeAddr(&cpuInfo)
healthInfo.Sys.CPUInfo = append(healthInfo.Sys.CPUInfo, cpuInfo)
}
partialWrite(healthInfo)
}
}
getAndWritePartitions := func() {
if query.Get("sysdrivehw") == "true" {
localPartitions := madmin.GetPartitions(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localPartitions)
healthInfo.Sys.Partitions = append(healthInfo.Sys.Partitions, localPartitions)
peerPartitions := globalNotificationSys.GetPartitions(deadlinedCtx)
for _, p := range peerPartitions {
anonymizeAddr(&p)
healthInfo.Sys.Partitions = append(healthInfo.Sys.Partitions, p)
}
partialWrite(healthInfo)
}
}
getAndWriteOSInfo := func() {
if query.Get("sysosinfo") == "true" {
localOSInfo := madmin.GetOSInfo(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localOSInfo)
healthInfo.Sys.OSInfo = append(healthInfo.Sys.OSInfo, localOSInfo)
peerOSInfos := globalNotificationSys.GetOSInfo(deadlinedCtx)
for _, o := range peerOSInfos {
anonymizeAddr(&o)
healthInfo.Sys.OSInfo = append(healthInfo.Sys.OSInfo, o)
}
partialWrite(healthInfo)
}
}
getAndWriteMemInfo := func() {
if query.Get("sysmem") == "true" {
localMemInfo := madmin.GetMemInfo(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localMemInfo)
healthInfo.Sys.MemInfo = append(healthInfo.Sys.MemInfo, localMemInfo)
peerMemInfos := globalNotificationSys.GetMemInfo(deadlinedCtx)
for _, m := range peerMemInfos {
anonymizeAddr(&m)
healthInfo.Sys.MemInfo = append(healthInfo.Sys.MemInfo, m)
}
partialWrite(healthInfo)
}
}
getAndWriteSysErrors := func() {
if query.Get(string(madmin.HealthDataTypeSysErrors)) == "true" {
localSysErrors := madmin.GetSysErrors(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localSysErrors)
healthInfo.Sys.SysErrs = append(healthInfo.Sys.SysErrs, localSysErrors)
partialWrite(healthInfo)
peerSysErrs := globalNotificationSys.GetSysErrors(deadlinedCtx)
for _, se := range peerSysErrs {
anonymizeAddr(&se)
healthInfo.Sys.SysErrs = append(healthInfo.Sys.SysErrs, se)
}
partialWrite(healthInfo)
}
}
getAndWriteSysConfig := func() {
if query.Get(string(madmin.HealthDataTypeSysConfig)) == "true" {
localSysConfig := madmin.GetSysConfig(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localSysConfig)
healthInfo.Sys.SysConfig = append(healthInfo.Sys.SysConfig, localSysConfig)
partialWrite(healthInfo)
peerSysConfig := globalNotificationSys.GetSysConfig(deadlinedCtx)
for _, sc := range peerSysConfig {
anonymizeAddr(&sc)
healthInfo.Sys.SysConfig = append(healthInfo.Sys.SysConfig, sc)
}
partialWrite(healthInfo)
}
}
getAndWriteSysServices := func() {
if query.Get(string(madmin.HealthDataTypeSysServices)) == "true" {
localSysServices := madmin.GetSysServices(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localSysServices)
healthInfo.Sys.SysServices = append(healthInfo.Sys.SysServices, localSysServices)
partialWrite(healthInfo)
peerSysServices := globalNotificationSys.GetSysServices(deadlinedCtx)
for _, ss := range peerSysServices {
anonymizeAddr(&ss)
healthInfo.Sys.SysServices = append(healthInfo.Sys.SysServices, ss)
}
partialWrite(healthInfo)
}
}
anonymizeCmdLine := func(cmdLine string) string {
if !globalIsDistErasure {
// FS mode - single server - hard code to `server1`
anonCmdLine := strings.Replace(cmdLine, globalLocalNodeName, "server1", -1)
return strings.Replace(anonCmdLine, globalMinioConsoleHost, "server1", -1)
}
// Server start command regex groups:
// 1 - minio server
// 2 - flags e.g. `--address :9000 --certs-dir /etc/minio/certs`
// 3 - pool args e.g. `https://node{01...16}.domain/data/disk{001...204} https://node{17...32}.domain/data/disk{001...204}`
re := regexp.MustCompile(`^(.*minio\s+server\s+)(--[^\s]+\s+[^\s]+\s+)*(.*)`)
// stays unchanged in the anonymized version
cmdLineWithoutPools := re.ReplaceAllString(cmdLine, `$1$2`)
// to be anonymized
poolsArgs := re.ReplaceAllString(cmdLine, `$3`)
var anonPools []string
if !(strings.Contains(poolsArgs, "{") && strings.Contains(poolsArgs, "}")) {
// No ellipses pattern. Anonymize host name from every pool arg
pools := strings.Fields(poolsArgs)
anonPools = make([]string, len(pools))
for _, arg := range pools {
anonPools = append(anonPools, anonAddr(arg))
}
return cmdLineWithoutPools + strings.Join(anonPools, " ")
}
// Ellipses pattern in pool args. Regex groups:
// 1 - server prefix
// 2 - number sequence for servers
// 3 - server suffix
// 4 - drive prefix (starting with /)
// 5 - number sequence for drives
// 6 - drive suffix
re = regexp.MustCompile(`([^\s^{]*)({\d+...\d+})?([^\s^{^/]*)(/[^\s^{]*)({\d+...\d+})?([^\s]*)`)
poolsMatches := re.FindAllStringSubmatch(poolsArgs, -1)
anonPools = make([]string, len(poolsMatches))
idxMap := map[int]string{
1: "spfx",
3: "ssfx",
}
for pi, poolsMatch := range poolsMatches {
// Replace the server prefix/suffix with anonymized ones
for idx, lbl := range idxMap {
if len(poolsMatch[idx]) > 0 {
poolsMatch[idx] = fmt.Sprintf("%s%d", lbl, crc32.ChecksumIEEE([]byte(poolsMatch[idx])))
}
}
// Remove the original pools args present at index 0
anonPools[pi] = strings.Join(poolsMatch[1:], "")
}
return cmdLineWithoutPools + strings.Join(anonPools, " ")
}
anonymizeProcInfo := func(p *madmin.ProcInfo) {
p.CmdLine = anonymizeCmdLine(p.CmdLine)
anonymizeAddr(p)
}
getAndWriteProcInfo := func() {
if query.Get("sysprocess") == "true" {
localProcInfo := madmin.GetProcInfo(deadlinedCtx, globalLocalNodeName)
anonymizeProcInfo(&localProcInfo)
healthInfo.Sys.ProcInfo = append(healthInfo.Sys.ProcInfo, localProcInfo)
peerProcInfos := globalNotificationSys.GetProcInfo(deadlinedCtx)
for _, p := range peerProcInfos {
anonymizeProcInfo(&p)
healthInfo.Sys.ProcInfo = append(healthInfo.Sys.ProcInfo, p)
}
partialWrite(healthInfo)
}
}
getAndWriteMinioConfig := func() {
if query.Get("minioconfig") == "true" {
config, err := readServerConfig(ctx, objectAPI)
if err != nil {
healthInfo.Minio.Config = madmin.MinioConfig{
Error: err.Error(),
}
} else {
healthInfo.Minio.Config = madmin.MinioConfig{
Config: config.RedactSensitiveInfo(),
}
}
partialWrite(healthInfo)
}
}
getAndWriteDrivePerfInfo := func() {
if query.Get("perfdrive") == "true" {
localDPI := getDrivePerfInfos(deadlinedCtx, globalLocalNodeName)
anonymizeAddr(&localDPI)
healthInfo.Perf.Drives = append(healthInfo.Perf.Drives, localDPI)
partialWrite(healthInfo)
perfCh := globalNotificationSys.GetDrivePerfInfos(deadlinedCtx)
for perfInfo := range perfCh {
anonymizeAddr(&perfInfo)
healthInfo.Perf.Drives = append(healthInfo.Perf.Drives, perfInfo)
partialWrite(healthInfo)
}
}
}
anonymizeNetPerfInfo := func(npi *madmin.NetPerfInfo) {
anonymizeAddr(npi)
rps := npi.RemotePeers
for idx, peer := range rps {
anonymizeAddr(&peer)
rps[idx] = peer
}
npi.RemotePeers = rps
}
getAndWriteNetPerfInfo := func() {
if globalIsDistErasure && query.Get("perfnet") == "true" {
localNPI := globalNotificationSys.GetNetPerfInfo(deadlinedCtx)
anonymizeNetPerfInfo(&localNPI)
healthInfo.Perf.Net = append(healthInfo.Perf.Net, localNPI)
partialWrite(healthInfo)
netInfos := globalNotificationSys.DispatchNetPerfChan(deadlinedCtx)
for netInfo := range netInfos {
anonymizeNetPerfInfo(&netInfo)
healthInfo.Perf.Net = append(healthInfo.Perf.Net, netInfo)
partialWrite(healthInfo)
}
ppi := globalNotificationSys.GetParallelNetPerfInfo(deadlinedCtx)
anonymizeNetPerfInfo(&ppi)
healthInfo.Perf.NetParallel = ppi
partialWrite(healthInfo)
}
}
anonymizeNetwork := func(network map[string]string) map[string]string {
anonNetwork := map[string]string{}
for endpoint, status := range network {
anonEndpoint := anonAddr(endpoint)
anonNetwork[anonEndpoint] = status
}
return anonNetwork
}
anonymizeDrives := func(drives []madmin.Disk) []madmin.Disk {
anonDrives := []madmin.Disk{}
for _, drive := range drives {
drive.Endpoint = anonAddr(drive.Endpoint)
anonDrives = append(anonDrives, drive)
}
return anonDrives
}
go func() {
defer close(healthInfoCh)
partialWrite(healthInfo) // Write first message with only version populated
getAndWriteCPUs()
getAndWritePartitions()
getAndWriteOSInfo()
getAndWriteMemInfo()
getAndWriteProcInfo()
getAndWriteMinioConfig()
getAndWriteDrivePerfInfo()
getAndWriteNetPerfInfo()
getAndWriteSysErrors()
getAndWriteSysServices()
getAndWriteSysConfig()
if query.Get("minioinfo") == "true" {
infoMessage := getServerInfo(ctx, r)
servers := []madmin.ServerInfo{}
for _, server := range infoMessage.Servers {
anonEndpoint := anonAddr(server.Endpoint)
servers = append(servers, madmin.ServerInfo{
State: server.State,
Endpoint: anonEndpoint,
Uptime: server.Uptime,
Version: server.Version,
CommitID: server.CommitID,
Network: anonymizeNetwork(server.Network),
Drives: anonymizeDrives(server.Disks),
PoolNumber: server.PoolNumber,
MemStats: madmin.MemStats{
Alloc: server.MemStats.Alloc,
TotalAlloc: server.MemStats.TotalAlloc,
Mallocs: server.MemStats.Mallocs,
Frees: server.MemStats.Frees,
HeapAlloc: server.MemStats.HeapAlloc,
},
})
}
healthInfo.Minio.Info = madmin.MinioInfo{
Mode: infoMessage.Mode,
Domain: infoMessage.Domain,
Region: infoMessage.Region,
SQSARN: infoMessage.SQSARN,
DeploymentID: infoMessage.DeploymentID,
Buckets: infoMessage.Buckets,
Objects: infoMessage.Objects,
Usage: infoMessage.Usage,
Services: infoMessage.Services,
Backend: infoMessage.Backend,
Servers: servers,
}
partialWrite(healthInfo)
}
}()
ticker := time.NewTicker(5 * time.Second)
defer ticker.Stop()
for {
select {
case oinfo, ok := <-healthInfoCh:
if !ok {
return
}
logger.LogIf(ctx, enc.Encode(oinfo))
w.(http.Flusher).Flush()
case <-ticker.C:
if _, err := w.Write([]byte(" ")); err != nil {
return
}
w.(http.Flusher).Flush()
case <-deadlinedCtx.Done():
w.(http.Flusher).Flush()
return
}
}
}
// BandwidthMonitorHandler - GET /minio/admin/v3/bandwidth
// ----------
// Get bandwidth consumption information
func (a adminAPIHandlers) BandwidthMonitorHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "BandwidthMonitor")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
// Validate request signature.
_, adminAPIErr := checkAdminRequestAuth(ctx, r, iampolicy.BandwidthMonitorAction, "")
if adminAPIErr != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(adminAPIErr), r.URL)
return
}
rnd := rand.New(rand.NewSource(time.Now().UnixNano()))
setEventStreamHeaders(w)
reportCh := make(chan madmin.BucketBandwidthReport)
keepAliveTicker := time.NewTicker(500 * time.Millisecond)
defer keepAliveTicker.Stop()
bucketsRequestedString := r.Form.Get("buckets")
bucketsRequested := strings.Split(bucketsRequestedString, ",")
go func() {
defer close(reportCh)
for {
select {
case <-ctx.Done():
return
case reportCh <- globalNotificationSys.GetBandwidthReports(ctx, bucketsRequested...):
time.Sleep(time.Duration(rnd.Float64() * float64(2*time.Second)))
}
}
}()
for {
select {
case report, ok := <-reportCh:
if !ok {
return
}
if err := json.NewEncoder(w).Encode(report); err != nil {
writeErrorResponseJSON(ctx, w, toAPIError(ctx, err), r.URL)
return
}
w.(http.Flusher).Flush()
case <-keepAliveTicker.C:
if _, err := w.Write([]byte(" ")); err != nil {
return
}
w.(http.Flusher).Flush()
case <-ctx.Done():
return
}
}
}
// ServerInfoHandler - GET /minio/admin/v3/info
// ----------
// Get server information
func (a adminAPIHandlers) ServerInfoHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "ServerInfo")
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
// Validate request signature.
_, adminAPIErr := checkAdminRequestAuth(ctx, r, iampolicy.ServerInfoAdminAction, "")
if adminAPIErr != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(adminAPIErr), r.URL)
return
}
// Marshal API response
jsonBytes, err := json.Marshal(getServerInfo(ctx, r))
if err != nil {
writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
return
}
// Reply with storage information (across nodes in a
// distributed setup) as json.
writeSuccessResponseJSON(w, jsonBytes)
}
func assignPoolNumbers(servers []madmin.ServerProperties) {
for i := range servers {
for idx, ge := range globalEndpoints {
for _, endpoint := range ge.Endpoints {
if servers[i].Endpoint == endpoint.Host {
servers[i].PoolNumber = idx + 1
} else if host, err := xnet.ParseHost(servers[i].Endpoint); err == nil {
if host.Name == endpoint.Hostname() {
servers[i].PoolNumber = idx + 1
}
}
}
}
}
}
func fetchLambdaInfo() []map[string][]madmin.TargetIDStatus {
lambdaMap := make(map[string][]madmin.TargetIDStatus)
for _, tgt := range globalConfigTargetList.Targets() {
targetIDStatus := make(map[string]madmin.Status)
active, _ := tgt.IsActive()
targetID := tgt.ID()
if active {
targetIDStatus[targetID.ID] = madmin.Status{Status: string(madmin.ItemOnline)}
} else {
targetIDStatus[targetID.ID] = madmin.Status{Status: string(madmin.ItemOffline)}
}
list := lambdaMap[targetID.Name]
list = append(list, targetIDStatus)
lambdaMap[targetID.Name] = list
}
for _, tgt := range globalEnvTargetList.Targets() {
targetIDStatus := make(map[string]madmin.Status)
active, _ := tgt.IsActive()
targetID := tgt.ID()
if active {
targetIDStatus[targetID.ID] = madmin.Status{Status: string(madmin.ItemOnline)}
} else {
targetIDStatus[targetID.ID] = madmin.Status{Status: string(madmin.ItemOffline)}
}
list := lambdaMap[targetID.Name]
list = append(list, targetIDStatus)
lambdaMap[targetID.Name] = list
}
notify := make([]map[string][]madmin.TargetIDStatus, len(lambdaMap))
counter := 0
for key, value := range lambdaMap {
v := make(map[string][]madmin.TargetIDStatus)
v[key] = value
notify[counter] = v
counter++
}
return notify
}
// fetchKMSStatus fetches KMS-related status information.
func fetchKMSStatus() madmin.KMS {
kmsStat := madmin.KMS{}
if GlobalKMS == nil {
kmsStat.Status = "disabled"
return kmsStat
}
stat, err := GlobalKMS.Stat()
if err != nil {
kmsStat.Status = string(madmin.ItemOffline)
return kmsStat
}
if len(stat.Endpoints) == 0 {
kmsStat.Status = stat.Name
return kmsStat
}
kmsStat.Status = string(madmin.ItemOnline)
kmsContext := kms.Context{"MinIO admin API": "ServerInfoHandler"} // Context for a test key operation
// 1. Generate a new key using the KMS.
key, err := GlobalKMS.GenerateKey("", kmsContext)
if err != nil {
kmsStat.Encrypt = fmt.Sprintf("Encryption failed: %v", err)
} else {
kmsStat.Encrypt = "success"
}
// 2. Verify that we can indeed decrypt the (encrypted) key
decryptedKey, err := GlobalKMS.DecryptKey(key.KeyID, key.Ciphertext, kmsContext)
switch {
case err != nil:
kmsStat.Decrypt = fmt.Sprintf("Decryption failed: %v", err)
case subtle.ConstantTimeCompare(key.Plaintext, decryptedKey) != 1:
kmsStat.Decrypt = "Decryption failed: decrypted key does not match generated key"
default:
kmsStat.Decrypt = "success"
}
return kmsStat
}
// fetchLoggerDetails return log info
func fetchLoggerInfo() ([]madmin.Logger, []madmin.Audit) {
var loggerInfo []madmin.Logger
var auditloggerInfo []madmin.Audit
for _, target := range logger.Targets {
if target.Endpoint() != "" {
tgt := target.String()
err := checkConnection(target.Endpoint(), 15*time.Second)
if err == nil {
mapLog := make(map[string]madmin.Status)
mapLog[tgt] = madmin.Status{Status: string(madmin.ItemOnline)}
loggerInfo = append(loggerInfo, mapLog)
} else {
mapLog := make(map[string]madmin.Status)
mapLog[tgt] = madmin.Status{Status: string(madmin.ItemOffline)}
loggerInfo = append(loggerInfo, mapLog)
}
}
}
for _, target := range logger.AuditTargets {
if target.Endpoint() != "" {
tgt := target.String()
err := checkConnection(target.Endpoint(), 15*time.Second)
if err == nil {
mapAudit := make(map[string]madmin.Status)
mapAudit[tgt] = madmin.Status{Status: string(madmin.ItemOnline)}
auditloggerInfo = append(auditloggerInfo, mapAudit)
} else {
mapAudit := make(map[string]madmin.Status)
mapAudit[tgt] = madmin.Status{Status: string(madmin.ItemOffline)}
auditloggerInfo = append(auditloggerInfo, mapAudit)
}
}
}
return loggerInfo, auditloggerInfo
}
// checkConnection - ping an endpoint , return err in case of no connection
func checkConnection(endpointStr string, timeout time.Duration) error {
ctx, cancel := context.WithTimeout(GlobalContext, timeout)
defer cancel()
client := &http.Client{Transport: &http.Transport{
Proxy: http.ProxyFromEnvironment,
DialContext: xhttp.NewCustomDialContext(timeout),
ResponseHeaderTimeout: 5 * time.Second,
TLSHandshakeTimeout: 5 * time.Second,
ExpectContinueTimeout: 5 * time.Second,
TLSClientConfig: &tls.Config{RootCAs: globalRootCAs},
// Go net/http automatically unzip if content-type is
// gzip disable this feature, as we are always interested
// in raw stream.
DisableCompression: true,
}}
defer client.CloseIdleConnections()
req, err := http.NewRequestWithContext(ctx, http.MethodHead, endpointStr, nil)
if err != nil {
return err
}
resp, err := client.Do(req)
if err != nil {
return err
}
defer xhttp.DrainBody(resp.Body)
return nil
}
// getRawDataer provides an interface for getting raw FS files.
type getRawDataer interface {
GetRawData(ctx context.Context, volume, file string, fn func(r io.Reader, host string, disk string, filename string, size int64, modtime time.Time) error) error
}
// InspectDataHandler - GET /minio/admin/v3/inspect-data
// ----------
// Download file from all nodes in a zip format
func (a adminAPIHandlers) InspectDataHandler(w http.ResponseWriter, r *http.Request) {
ctx := newContext(r, w, "InspectData")
// Validate request signature.
_, adminAPIErr := checkAdminRequestAuth(ctx, r, iampolicy.InspectDataAction, "")
if adminAPIErr != ErrNone {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(adminAPIErr), r.URL)
return
}
defer logger.AuditLog(ctx, w, r, mustGetClaimsFromToken(r))
o, ok := newObjectLayerFn().(getRawDataer)
if !ok {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrNotImplemented), r.URL)
return
}
volume := r.Form.Get("volume")
file := r.Form.Get("file")
if len(volume) == 0 {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidBucketName), r.URL)
return
}
if len(file) == 0 {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInvalidRequest), r.URL)
return
}
var key [32]byte
// MUST use crypto/rand
n, err := crand.Read(key[:])
if err != nil || n != len(key) {
logger.LogIf(ctx, err)
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), r.URL)
return
}
stream, err := sio.AES_256_GCM.Stream(key[:])
if err != nil {
logger.LogIf(ctx, err)
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), r.URL)
return
}
// Zero nonce, we only use each key once, and 32 bytes is plenty.
nonce := make([]byte, stream.NonceSize())
encw := stream.EncryptWriter(w, nonce, nil)
defer encw.Close()
// Write a version for making *incompatible* changes.
// The AdminClient will reject any version it does not know.
w.Write([]byte{1})
// Write key first (without encryption)
_, err = w.Write(key[:])
if err != nil {
writeErrorResponseJSON(ctx, w, errorCodes.ToAPIErr(ErrInternalError), r.URL)
return
}
// Initialize a zip writer which will provide a zipped content
// of profiling data of all nodes
zipWriter := zip.NewWriter(encw)
defer zipWriter.Close()
err = o.GetRawData(ctx, volume, file, func(r io.Reader, host, disk, filename string, size int64, modtime time.Time) error {
// Prefix host+disk
filename = path.Join(host, disk, filename)
header, zerr := zip.FileInfoHeader(dummyFileInfo{
name: filename,
size: size,
mode: 0600,
modTime: modtime,
isDir: false,
sys: nil,
})
if zerr != nil {
logger.LogIf(ctx, zerr)
return nil
}
header.Method = zip.Deflate
zwriter, zerr := zipWriter.CreateHeader(header)
if zerr != nil {
logger.LogIf(ctx, zerr)
return nil
}
if _, err = io.Copy(zwriter, r); err != nil {
logger.LogIf(ctx, err)
}
return nil
})
logger.LogIf(ctx, err)
}
func createHostAnonymizerForFSMode() map[string]string {
hostAnonymizer := map[string]string{
globalLocalNodeName: "server1",
}
apiEndpoints := getAPIEndpoints()
for _, ep := range apiEndpoints {
if len(ep) == 0 {
continue
}
if url, err := xnet.ParseHTTPURL(ep); err == nil {
// In FS mode the drive names don't include the host.
// So mapping just the host should be sufficient.
hostAnonymizer[url.Host] = "server1"
}
}
return hostAnonymizer
}
// anonymizeHost - Add entries related to given endpoint in the host anonymizer map
// The health report data can contain the hostname in various forms e.g. host, host:port,
// host:port/drivepath, full url (http://host:port/drivepath)
// The anonymizer map will have mappings for all these varients for efficiently replacing
// any of these strings to the anonymized versions at the time of health report generation.
func anonymizeHost(hostAnonymizer map[string]string, endpoint Endpoint, poolNum int, srvrNum int) {
if len(endpoint.Host) == 0 {
return
}
currentURL := endpoint.String()
// mapIfNotPresent - Maps the given key to the value only if the key is not present in the map
mapIfNotPresent := func(m map[string]string, key string, val string) {
_, found := m[key]
if !found {
m[key] = val
}
}
_, found := hostAnonymizer[currentURL]
if !found {
// In distributed setup, anonymized addr = 'poolNum.serverNum'
newHost := fmt.Sprintf("pool%d.server%d", poolNum, srvrNum)
// Hostname
mapIfNotPresent(hostAnonymizer, endpoint.Hostname(), newHost)
newHostPort := newHost
if len(endpoint.Port()) > 0 {
// Host + port
newHostPort = newHost + ":" + endpoint.Port()
mapIfNotPresent(hostAnonymizer, endpoint.Host, newHostPort)
}
newHostPortPath := newHostPort
if len(endpoint.Path) > 0 {
// Host + port + path
currentHostPortPath := endpoint.Host + endpoint.Path
newHostPortPath = newHostPort + endpoint.Path
mapIfNotPresent(hostAnonymizer, currentHostPortPath, newHostPortPath)
}
// Full url
hostAnonymizer[currentURL] = endpoint.Scheme + "://" + newHostPortPath
}
}
// createHostAnonymizer - Creats a map of various strings to corresponding anonymized names
func createHostAnonymizer() map[string]string {
if !globalIsDistErasure {
return createHostAnonymizerForFSMode()
}
hostAnonymizer := map[string]string{}
for poolIdx, pool := range globalEndpoints {
for srvrIdx, endpoint := range pool.Endpoints {
anonymizeHost(hostAnonymizer, endpoint, poolIdx+1, srvrIdx+1)
}
}
return hostAnonymizer
}