2024-01-13 02:51:08 -05:00
|
|
|
// Copyright (c) 2015-2023 MinIO, Inc.
|
2021-04-18 15:41:13 -04:00
|
|
|
//
|
|
|
|
// 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/>.
|
2020-03-18 19:19:29 -04:00
|
|
|
|
|
|
|
package cmd
|
|
|
|
|
|
|
|
import (
|
|
|
|
"context"
|
2020-04-27 13:06:21 -04:00
|
|
|
"errors"
|
2020-03-18 19:19:29 -04:00
|
|
|
"fmt"
|
|
|
|
"io"
|
2023-03-09 18:15:46 -05:00
|
|
|
"math/rand"
|
2020-12-31 12:45:09 -05:00
|
|
|
"net/http"
|
2020-03-18 19:19:29 -04:00
|
|
|
"path"
|
|
|
|
"path/filepath"
|
2021-05-11 21:36:15 -04:00
|
|
|
"sort"
|
2020-03-18 19:19:29 -04:00
|
|
|
"strings"
|
|
|
|
"time"
|
|
|
|
|
|
|
|
"github.com/cespare/xxhash/v2"
|
2024-01-13 02:51:08 -05:00
|
|
|
"github.com/dustin/go-humanize"
|
2020-06-12 13:28:21 -04:00
|
|
|
"github.com/klauspost/compress/zstd"
|
2023-06-19 20:53:08 -04:00
|
|
|
"github.com/minio/madmin-go/v3"
|
2021-06-01 17:59:40 -04:00
|
|
|
"github.com/minio/minio/internal/bucket/lifecycle"
|
2020-03-18 19:19:29 -04:00
|
|
|
"github.com/tinylib/msgp/msgp"
|
2023-09-14 14:53:52 -04:00
|
|
|
"github.com/valyala/bytebufferpool"
|
2020-03-18 19:19:29 -04:00
|
|
|
)
|
|
|
|
|
|
|
|
//go:generate msgp -file $GOFILE -unexported
|
|
|
|
|
|
|
|
// dataUsageHash is the hash type used.
|
2020-06-12 13:28:21 -04:00
|
|
|
type dataUsageHash string
|
2020-03-18 19:19:29 -04:00
|
|
|
|
2024-01-13 02:51:08 -05:00
|
|
|
// sizeHistogramV1 is size histogram V1, which has fewer intervals esp. between
|
|
|
|
// 1024B and 1MiB.
|
|
|
|
type sizeHistogramV1 [dataUsageBucketLenV1]uint64
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// sizeHistogram is a size histogram.
|
|
|
|
type sizeHistogram [dataUsageBucketLen]uint64
|
|
|
|
|
2023-03-10 11:53:59 -05:00
|
|
|
// versionsHistogram is a histogram of number of versions in an object.
|
|
|
|
type versionsHistogram [dataUsageVersionLen]uint64
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
type dataUsageEntry struct {
|
2021-10-23 21:38:33 -04:00
|
|
|
Children dataUsageHashMap `msg:"ch"`
|
2020-03-18 19:19:29 -04:00
|
|
|
// These fields do no include any children.
|
2021-10-23 21:38:33 -04:00
|
|
|
Size int64 `msg:"sz"`
|
|
|
|
Objects uint64 `msg:"os"`
|
|
|
|
Versions uint64 `msg:"vs"` // Versions that are not delete markers.
|
2023-07-18 13:49:40 -04:00
|
|
|
DeleteMarkers uint64 `msg:"dms"`
|
2021-10-23 21:38:33 -04:00
|
|
|
ObjSizes sizeHistogram `msg:"szs"`
|
2023-03-10 11:53:59 -05:00
|
|
|
ObjVersions versionsHistogram `msg:"vh"`
|
2021-10-23 21:38:33 -04:00
|
|
|
ReplicationStats *replicationAllStats `msg:"rs,omitempty"`
|
|
|
|
AllTierStats *allTierStats `msg:"ats,omitempty"`
|
|
|
|
Compacted bool `msg:"c"`
|
|
|
|
}
|
|
|
|
|
|
|
|
// allTierStats is a collection of per-tier stats across all configured remote
|
|
|
|
// tiers.
|
|
|
|
type allTierStats struct {
|
|
|
|
Tiers map[string]tierStats `msg:"ts"`
|
|
|
|
}
|
|
|
|
|
|
|
|
func newAllTierStats() *allTierStats {
|
|
|
|
return &allTierStats{
|
|
|
|
Tiers: make(map[string]tierStats),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-10-30 12:59:51 -04:00
|
|
|
func (ats *allTierStats) addSizes(tiers map[string]tierStats) {
|
|
|
|
for tier, st := range tiers {
|
2021-10-23 21:38:33 -04:00
|
|
|
ats.Tiers[tier] = ats.Tiers[tier].add(st)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (ats *allTierStats) merge(other *allTierStats) {
|
|
|
|
for tier, st := range other.Tiers {
|
|
|
|
ats.Tiers[tier] = ats.Tiers[tier].add(st)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-12 12:17:11 -04:00
|
|
|
func (ats *allTierStats) clone() *allTierStats {
|
|
|
|
if ats == nil {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
dst := *ats
|
2023-10-30 12:59:51 -04:00
|
|
|
dst.Tiers = make(map[string]tierStats, len(ats.Tiers))
|
|
|
|
for tier, st := range ats.Tiers {
|
|
|
|
dst.Tiers[tier] = st
|
2023-06-12 12:17:11 -04:00
|
|
|
}
|
|
|
|
return &dst
|
|
|
|
}
|
|
|
|
|
2023-10-30 12:59:51 -04:00
|
|
|
func (ats *allTierStats) populateStats(stats map[string]madmin.TierStats) {
|
2021-10-23 21:38:33 -04:00
|
|
|
if ats == nil {
|
2023-10-30 12:59:51 -04:00
|
|
|
return
|
2021-10-23 21:38:33 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// Update stats for tiers as they become available.
|
|
|
|
for tier, st := range ats.Tiers {
|
|
|
|
stats[tier] = madmin.TierStats{
|
|
|
|
TotalSize: st.TotalSize,
|
|
|
|
NumVersions: st.NumVersions,
|
|
|
|
NumObjects: st.NumObjects,
|
|
|
|
}
|
|
|
|
}
|
2023-10-30 12:59:51 -04:00
|
|
|
return
|
2021-10-23 21:38:33 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// tierStats holds per-tier stats of a remote tier.
|
|
|
|
type tierStats struct {
|
|
|
|
TotalSize uint64 `msg:"ts"`
|
|
|
|
NumVersions int `msg:"nv"`
|
|
|
|
NumObjects int `msg:"no"`
|
|
|
|
}
|
|
|
|
|
|
|
|
func (ts tierStats) add(u tierStats) tierStats {
|
2023-10-30 12:59:51 -04:00
|
|
|
return tierStats{
|
|
|
|
TotalSize: ts.TotalSize + u.TotalSize,
|
|
|
|
NumVersions: ts.NumVersions + u.NumVersions,
|
|
|
|
NumObjects: ts.NumObjects + u.NumObjects,
|
|
|
|
}
|
2021-04-03 12:03:42 -04:00
|
|
|
}
|
|
|
|
|
2021-09-21 12:01:51 -04:00
|
|
|
//msgp:tuple replicationStatsV1
|
|
|
|
type replicationStatsV1 struct {
|
|
|
|
PendingSize uint64
|
|
|
|
ReplicatedSize uint64
|
|
|
|
FailedSize uint64
|
|
|
|
ReplicaSize uint64
|
|
|
|
FailedCount uint64
|
|
|
|
PendingCount uint64
|
|
|
|
MissedThresholdSize uint64
|
|
|
|
AfterThresholdSize uint64
|
|
|
|
MissedThresholdCount uint64
|
|
|
|
AfterThresholdCount uint64
|
|
|
|
}
|
|
|
|
|
|
|
|
func (rsv1 replicationStatsV1) Empty() bool {
|
|
|
|
return rsv1.ReplicatedSize == 0 &&
|
|
|
|
rsv1.FailedSize == 0 &&
|
|
|
|
rsv1.FailedCount == 0
|
|
|
|
}
|
|
|
|
|
2021-04-03 12:03:42 -04:00
|
|
|
//msgp:tuple replicationStats
|
|
|
|
type replicationStats struct {
|
|
|
|
PendingSize uint64
|
|
|
|
ReplicatedSize uint64
|
|
|
|
FailedSize uint64
|
|
|
|
FailedCount uint64
|
|
|
|
PendingCount uint64
|
|
|
|
MissedThresholdSize uint64
|
|
|
|
AfterThresholdSize uint64
|
|
|
|
MissedThresholdCount uint64
|
|
|
|
AfterThresholdCount uint64
|
2023-08-30 04:00:59 -04:00
|
|
|
ReplicatedCount uint64
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
2021-09-18 16:31:35 -04:00
|
|
|
func (rs replicationStats) Empty() bool {
|
|
|
|
return rs.ReplicatedSize == 0 &&
|
|
|
|
rs.FailedSize == 0 &&
|
|
|
|
rs.FailedCount == 0
|
|
|
|
}
|
|
|
|
|
|
|
|
type replicationAllStats struct {
|
2023-08-30 04:00:59 -04:00
|
|
|
Targets map[string]replicationStats `msg:"t,omitempty"`
|
|
|
|
ReplicaSize uint64 `msg:"r,omitempty"`
|
|
|
|
ReplicaCount uint64 `msg:"rc,omitempty"`
|
2021-10-23 21:38:33 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
//msgp:tuple replicationAllStatsV1
|
|
|
|
type replicationAllStatsV1 struct {
|
2023-08-30 04:00:59 -04:00
|
|
|
Targets map[string]replicationStats
|
|
|
|
ReplicaSize uint64 `msg:"ReplicaSize,omitempty"`
|
|
|
|
ReplicaCount uint64 `msg:"ReplicaCount,omitempty"`
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
|
|
|
|
2024-03-28 13:13:07 -04:00
|
|
|
// empty returns true if the replicationAllStats is empty (contains no entries).
|
|
|
|
func (r *replicationAllStats) empty() bool {
|
|
|
|
if r == nil {
|
|
|
|
return true
|
|
|
|
}
|
|
|
|
if r.ReplicaSize != 0 || r.ReplicaCount != 0 {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
for _, v := range r.Targets {
|
|
|
|
if !v.Empty() {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return true
|
|
|
|
}
|
|
|
|
|
2023-06-12 12:17:11 -04:00
|
|
|
// clone creates a deep-copy clone.
|
|
|
|
func (r *replicationAllStats) clone() *replicationAllStats {
|
|
|
|
if r == nil {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
// Shallow copy
|
|
|
|
dst := *r
|
|
|
|
|
|
|
|
// Copy individual targets.
|
2023-10-30 12:59:51 -04:00
|
|
|
dst.Targets = make(map[string]replicationStats, len(r.Targets))
|
|
|
|
for k, v := range r.Targets {
|
|
|
|
dst.Targets[k] = v
|
2023-06-12 12:17:11 -04:00
|
|
|
}
|
2023-10-30 12:59:51 -04:00
|
|
|
|
2023-06-12 12:17:11 -04:00
|
|
|
return &dst
|
|
|
|
}
|
|
|
|
|
2024-01-13 02:51:08 -05:00
|
|
|
//msgp:encode ignore dataUsageEntryV2 dataUsageEntryV3 dataUsageEntryV4 dataUsageEntryV5 dataUsageEntryV6 dataUsageEntryV7
|
|
|
|
//msgp:marshal ignore dataUsageEntryV2 dataUsageEntryV3 dataUsageEntryV4 dataUsageEntryV5 dataUsageEntryV6 dataUsageEntryV7
|
2021-05-11 21:36:15 -04:00
|
|
|
|
2021-01-13 12:58:08 -05:00
|
|
|
//msgp:tuple dataUsageEntryV2
|
|
|
|
type dataUsageEntryV2 struct {
|
|
|
|
// These fields do no include any children.
|
|
|
|
Size int64
|
|
|
|
Objects uint64
|
|
|
|
ObjSizes sizeHistogram
|
|
|
|
Children dataUsageHashMap
|
|
|
|
}
|
|
|
|
|
2021-04-03 12:03:42 -04:00
|
|
|
//msgp:tuple dataUsageEntryV3
|
|
|
|
type dataUsageEntryV3 struct {
|
|
|
|
// These fields do no include any children.
|
|
|
|
Size int64
|
|
|
|
ReplicatedSize uint64
|
|
|
|
ReplicationPendingSize uint64
|
|
|
|
ReplicationFailedSize uint64
|
|
|
|
ReplicaSize uint64
|
|
|
|
Objects uint64
|
|
|
|
ObjSizes sizeHistogram
|
|
|
|
Children dataUsageHashMap
|
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
//msgp:tuple dataUsageEntryV4
|
|
|
|
type dataUsageEntryV4 struct {
|
|
|
|
Children dataUsageHashMap
|
|
|
|
// These fields do no include any children.
|
|
|
|
Size int64
|
|
|
|
Objects uint64
|
|
|
|
ObjSizes sizeHistogram
|
2021-09-21 12:01:51 -04:00
|
|
|
ReplicationStats replicationStatsV1
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
|
|
|
|
2021-09-18 16:31:35 -04:00
|
|
|
//msgp:tuple dataUsageEntryV5
|
|
|
|
type dataUsageEntryV5 struct {
|
|
|
|
Children dataUsageHashMap
|
|
|
|
// These fields do no include any children.
|
|
|
|
Size int64
|
|
|
|
Objects uint64
|
|
|
|
Versions uint64 // Versions that are not delete markers.
|
|
|
|
ObjSizes sizeHistogram
|
2021-09-21 12:01:51 -04:00
|
|
|
ReplicationStats *replicationStatsV1
|
2021-09-18 16:31:35 -04:00
|
|
|
Compacted bool
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
//msgp:tuple dataUsageEntryV6
|
|
|
|
type dataUsageEntryV6 struct {
|
|
|
|
Children dataUsageHashMap
|
|
|
|
// These fields do no include any children.
|
|
|
|
Size int64
|
|
|
|
Objects uint64
|
|
|
|
Versions uint64 // Versions that are not delete markers.
|
|
|
|
ObjSizes sizeHistogram
|
|
|
|
ReplicationStats *replicationAllStatsV1
|
|
|
|
Compacted bool
|
|
|
|
}
|
|
|
|
|
2024-01-13 02:51:08 -05:00
|
|
|
type dataUsageEntryV7 struct {
|
|
|
|
Children dataUsageHashMap `msg:"ch"`
|
|
|
|
// These fields do no include any children.
|
|
|
|
Size int64 `msg:"sz"`
|
|
|
|
Objects uint64 `msg:"os"`
|
|
|
|
Versions uint64 `msg:"vs"` // Versions that are not delete markers.
|
|
|
|
DeleteMarkers uint64 `msg:"dms"`
|
|
|
|
ObjSizes sizeHistogramV1 `msg:"szs"`
|
|
|
|
ObjVersions versionsHistogram `msg:"vh"`
|
|
|
|
ReplicationStats *replicationAllStats `msg:"rs,omitempty"`
|
|
|
|
AllTierStats *allTierStats `msg:"ats,omitempty"`
|
|
|
|
Compacted bool `msg:"c"`
|
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// dataUsageCache contains a cache of data usage entries latest version.
|
2020-06-12 13:28:21 -04:00
|
|
|
type dataUsageCache struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntry
|
|
|
|
}
|
|
|
|
|
2024-01-13 02:51:08 -05:00
|
|
|
//msgp:encode ignore dataUsageCacheV2 dataUsageCacheV3 dataUsageCacheV4 dataUsageCacheV5 dataUsageCacheV6 dataUsageCacheV7
|
|
|
|
//msgp:marshal ignore dataUsageCacheV2 dataUsageCacheV3 dataUsageCacheV4 dataUsageCacheV5 dataUsageCacheV6 dataUsageCacheV7
|
2021-05-11 21:36:15 -04:00
|
|
|
|
2021-04-03 12:03:42 -04:00
|
|
|
// dataUsageCacheV2 contains a cache of data usage entries version 2.
|
2021-01-13 12:58:08 -05:00
|
|
|
type dataUsageCacheV2 struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntryV2
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
// dataUsageCacheV3 contains a cache of data usage entries version 3.
|
2021-04-03 12:03:42 -04:00
|
|
|
type dataUsageCacheV3 struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntryV3
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
// dataUsageCacheV4 contains a cache of data usage entries version 4.
|
2021-05-11 21:36:15 -04:00
|
|
|
type dataUsageCacheV4 struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntryV4
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
// dataUsageCacheV5 contains a cache of data usage entries version 5.
|
2021-09-18 16:31:35 -04:00
|
|
|
type dataUsageCacheV5 struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntryV5
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
// dataUsageCacheV6 contains a cache of data usage entries version 6.
|
|
|
|
type dataUsageCacheV6 struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntryV6
|
|
|
|
}
|
|
|
|
|
2024-01-13 02:51:08 -05:00
|
|
|
// dataUsageCacheV7 contains a cache of data usage entries version 7.
|
|
|
|
type dataUsageCacheV7 struct {
|
|
|
|
Info dataUsageCacheInfo
|
|
|
|
Cache map[string]dataUsageEntryV7
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
//msgp:ignore dataUsageEntryInfo
|
|
|
|
type dataUsageEntryInfo struct {
|
|
|
|
Name string
|
|
|
|
Parent string
|
|
|
|
Entry dataUsageEntry
|
|
|
|
}
|
|
|
|
|
|
|
|
type dataUsageCacheInfo struct {
|
|
|
|
// Name of the bucket. Also root element.
|
2021-01-19 05:40:52 -05:00
|
|
|
Name string
|
|
|
|
NextCycle uint32
|
2021-04-03 12:03:42 -04:00
|
|
|
LastUpdate time.Time
|
2021-02-17 15:04:11 -05:00
|
|
|
// indicates if the disk is being healed and scanner
|
2021-01-19 05:40:52 -05:00
|
|
|
// should skip healing the disk
|
|
|
|
SkipHealing bool
|
2021-05-19 17:38:30 -04:00
|
|
|
|
|
|
|
// Active lifecycle, if any on the bucket
|
|
|
|
lifeCycle *lifecycle.Lifecycle `msg:"-"`
|
|
|
|
|
|
|
|
// optional updates channel.
|
|
|
|
// If set updates will be sent regularly to this channel.
|
|
|
|
// Will not be closed when returned.
|
2021-06-01 22:59:11 -04:00
|
|
|
updates chan<- dataUsageEntry `msg:"-"`
|
|
|
|
replication replicationConfig `msg:"-"`
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
2020-12-07 16:47:48 -05:00
|
|
|
func (e *dataUsageEntry) addSizes(summary sizeSummary) {
|
|
|
|
e.Size += summary.totalSize
|
2021-05-11 21:36:15 -04:00
|
|
|
e.Versions += summary.versions
|
2023-07-18 13:49:40 -04:00
|
|
|
e.DeleteMarkers += summary.deleteMarkers
|
2021-05-11 21:36:15 -04:00
|
|
|
e.ObjSizes.add(summary.totalSize)
|
2023-03-10 11:53:59 -05:00
|
|
|
e.ObjVersions.add(summary.versions)
|
2021-04-03 12:03:42 -04:00
|
|
|
|
2021-10-22 22:13:50 -04:00
|
|
|
if e.ReplicationStats == nil {
|
|
|
|
e.ReplicationStats = &replicationAllStats{
|
|
|
|
Targets: make(map[string]replicationStats),
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
2021-10-22 22:13:50 -04:00
|
|
|
} else if e.ReplicationStats.Targets == nil {
|
|
|
|
e.ReplicationStats.Targets = make(map[string]replicationStats)
|
|
|
|
}
|
|
|
|
e.ReplicationStats.ReplicaSize += uint64(summary.replicaSize)
|
2023-08-30 04:00:59 -04:00
|
|
|
e.ReplicationStats.ReplicaCount += uint64(summary.replicaCount)
|
2021-10-22 22:13:50 -04:00
|
|
|
|
2023-09-16 22:09:29 -04:00
|
|
|
for arn, st := range summary.replTargetStats {
|
|
|
|
tgtStat, ok := e.ReplicationStats.Targets[arn]
|
|
|
|
if !ok {
|
|
|
|
tgtStat = replicationStats{}
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
2023-09-16 22:09:29 -04:00
|
|
|
tgtStat.PendingSize += uint64(st.pendingSize)
|
|
|
|
tgtStat.FailedSize += uint64(st.failedSize)
|
|
|
|
tgtStat.ReplicatedSize += uint64(st.replicatedSize)
|
|
|
|
tgtStat.ReplicatedCount += uint64(st.replicatedCount)
|
|
|
|
tgtStat.FailedCount += st.failedCount
|
|
|
|
tgtStat.PendingCount += st.pendingCount
|
|
|
|
e.ReplicationStats.Targets[arn] = tgtStat
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
2023-10-30 12:59:51 -04:00
|
|
|
if len(summary.tiers) != 0 {
|
2021-10-23 21:38:33 -04:00
|
|
|
if e.AllTierStats == nil {
|
|
|
|
e.AllTierStats = newAllTierStats()
|
|
|
|
}
|
2023-10-30 12:59:51 -04:00
|
|
|
e.AllTierStats.addSizes(summary.tiers)
|
2021-10-23 21:38:33 -04:00
|
|
|
}
|
2020-12-07 16:47:48 -05:00
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// merge other data usage entry into this, excluding children.
|
|
|
|
func (e *dataUsageEntry) merge(other dataUsageEntry) {
|
|
|
|
e.Objects += other.Objects
|
2021-05-11 21:36:15 -04:00
|
|
|
e.Versions += other.Versions
|
2023-07-18 13:49:40 -04:00
|
|
|
e.DeleteMarkers += other.DeleteMarkers
|
2020-03-18 19:19:29 -04:00
|
|
|
e.Size += other.Size
|
2021-10-22 22:13:50 -04:00
|
|
|
if other.ReplicationStats != nil {
|
2021-05-11 21:36:15 -04:00
|
|
|
if e.ReplicationStats == nil {
|
2021-09-18 16:31:35 -04:00
|
|
|
e.ReplicationStats = &replicationAllStats{Targets: make(map[string]replicationStats)}
|
2021-10-22 22:13:50 -04:00
|
|
|
} else if e.ReplicationStats.Targets == nil {
|
|
|
|
e.ReplicationStats.Targets = make(map[string]replicationStats)
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
2021-10-22 22:13:50 -04:00
|
|
|
e.ReplicationStats.ReplicaSize += other.ReplicationStats.ReplicaSize
|
2023-08-30 04:00:59 -04:00
|
|
|
e.ReplicationStats.ReplicaCount += other.ReplicationStats.ReplicaCount
|
2021-10-22 22:13:50 -04:00
|
|
|
for arn, stat := range other.ReplicationStats.Targets {
|
|
|
|
st := e.ReplicationStats.Targets[arn]
|
|
|
|
e.ReplicationStats.Targets[arn] = replicationStats{
|
2023-08-30 04:00:59 -04:00
|
|
|
PendingSize: stat.PendingSize + st.PendingSize,
|
|
|
|
FailedSize: stat.FailedSize + st.FailedSize,
|
|
|
|
ReplicatedSize: stat.ReplicatedSize + st.ReplicatedSize,
|
|
|
|
PendingCount: stat.PendingCount + st.PendingCount,
|
|
|
|
FailedCount: stat.FailedCount + st.FailedCount,
|
|
|
|
ReplicatedCount: stat.ReplicatedCount + st.ReplicatedCount,
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
|
|
|
}
|
2020-12-07 16:47:48 -05:00
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
for i, v := range other.ObjSizes[:] {
|
|
|
|
e.ObjSizes[i] += v
|
|
|
|
}
|
2021-10-23 21:38:33 -04:00
|
|
|
|
2023-03-10 11:53:59 -05:00
|
|
|
for i, v := range other.ObjVersions[:] {
|
|
|
|
e.ObjVersions[i] += v
|
|
|
|
}
|
|
|
|
|
2023-10-30 12:59:51 -04:00
|
|
|
if other.AllTierStats != nil && len(other.AllTierStats.Tiers) != 0 {
|
2021-10-23 21:38:33 -04:00
|
|
|
if e.AllTierStats == nil {
|
|
|
|
e.AllTierStats = newAllTierStats()
|
|
|
|
}
|
|
|
|
e.AllTierStats.merge(other.AllTierStats)
|
|
|
|
}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// mod returns true if the hash mod cycles == cycle.
|
2020-08-24 16:47:01 -04:00
|
|
|
// If cycles is 0 false is always returned.
|
|
|
|
// If cycles is 1 true is always returned (as expected).
|
2020-04-27 13:06:21 -04:00
|
|
|
func (h dataUsageHash) mod(cycle uint32, cycles uint32) bool {
|
2020-08-24 16:47:01 -04:00
|
|
|
if cycles <= 1 {
|
|
|
|
return cycles == 1
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
return uint32(xxhash.Sum64String(string(h)))%cycles == cycle%cycles
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
2022-03-07 12:25:53 -05:00
|
|
|
// modAlt returns true if the hash mod cycles == cycle.
|
|
|
|
// This is out of sync with mod.
|
|
|
|
// If cycles is 0 false is always returned.
|
|
|
|
// If cycles is 1 true is always returned (as expected).
|
|
|
|
func (h dataUsageHash) modAlt(cycle uint32, cycles uint32) bool {
|
|
|
|
if cycles <= 1 {
|
|
|
|
return cycles == 1
|
|
|
|
}
|
|
|
|
return uint32(xxhash.Sum64String(string(h))>>32)%(cycles) == cycle%cycles
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// addChild will add a child based on its hash.
|
|
|
|
// If it already exists it will not be added again.
|
|
|
|
func (e *dataUsageEntry) addChild(hash dataUsageHash) {
|
2020-06-12 13:28:21 -04:00
|
|
|
if _, ok := e.Children[hash.Key()]; ok {
|
2020-03-18 19:19:29 -04:00
|
|
|
return
|
|
|
|
}
|
|
|
|
if e.Children == nil {
|
|
|
|
e.Children = make(dataUsageHashMap, 1)
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
e.Children[hash.Key()] = struct{}{}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
2021-08-24 10:11:38 -04:00
|
|
|
// Create a clone of the entry.
|
|
|
|
func (e dataUsageEntry) clone() dataUsageEntry {
|
|
|
|
// We operate on a copy from the receiver.
|
|
|
|
if e.Children != nil {
|
|
|
|
ch := make(dataUsageHashMap, len(e.Children))
|
|
|
|
for k, v := range e.Children {
|
|
|
|
ch[k] = v
|
|
|
|
}
|
|
|
|
e.Children = ch
|
|
|
|
}
|
|
|
|
if e.ReplicationStats != nil {
|
2023-06-12 12:17:11 -04:00
|
|
|
// Clone ReplicationStats
|
|
|
|
e.ReplicationStats = e.ReplicationStats.clone()
|
2021-08-24 10:11:38 -04:00
|
|
|
}
|
2021-10-23 21:38:33 -04:00
|
|
|
if e.AllTierStats != nil {
|
2023-06-12 12:17:11 -04:00
|
|
|
e.AllTierStats = e.AllTierStats.clone()
|
2021-10-23 21:38:33 -04:00
|
|
|
}
|
2021-08-24 10:11:38 -04:00
|
|
|
return e
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// find a path in the cache.
|
|
|
|
// Returns nil if not found.
|
|
|
|
func (d *dataUsageCache) find(path string) *dataUsageEntry {
|
2020-06-12 13:28:21 -04:00
|
|
|
due, ok := d.Cache[hashPath(path).Key()]
|
2020-03-18 19:19:29 -04:00
|
|
|
if !ok {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
return &due
|
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// isCompacted returns whether an entry is compacted.
|
|
|
|
// Returns false if not found.
|
|
|
|
func (d *dataUsageCache) isCompacted(h dataUsageHash) bool {
|
|
|
|
due, ok := d.Cache[h.Key()]
|
|
|
|
if !ok {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
return due.Compacted
|
|
|
|
}
|
|
|
|
|
2020-08-24 16:47:01 -04:00
|
|
|
// findChildrenCopy returns a copy of the children of the supplied hash.
|
|
|
|
func (d *dataUsageCache) findChildrenCopy(h dataUsageHash) dataUsageHashMap {
|
|
|
|
ch := d.Cache[h.String()].Children
|
|
|
|
res := make(dataUsageHashMap, len(ch))
|
|
|
|
for k := range ch {
|
|
|
|
res[k] = struct{}{}
|
|
|
|
}
|
|
|
|
return res
|
|
|
|
}
|
|
|
|
|
2021-05-19 17:38:30 -04:00
|
|
|
// searchParent will search for the parent of h.
|
|
|
|
// This is an O(N*N) operation if there is no parent or it cannot be guessed.
|
|
|
|
func (d *dataUsageCache) searchParent(h dataUsageHash) *dataUsageHash {
|
|
|
|
want := h.Key()
|
|
|
|
if idx := strings.LastIndexByte(want, '/'); idx >= 0 {
|
|
|
|
if v := d.find(want[:idx]); v != nil {
|
2024-03-27 11:12:14 -04:00
|
|
|
_, ok := v.Children[want]
|
|
|
|
if ok {
|
|
|
|
found := hashPath(want[:idx])
|
|
|
|
return &found
|
2021-05-19 17:38:30 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
for k, v := range d.Cache {
|
2024-03-27 11:12:14 -04:00
|
|
|
_, ok := v.Children[want]
|
|
|
|
if ok {
|
|
|
|
found := dataUsageHash(k)
|
|
|
|
return &found
|
2021-05-19 17:38:30 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// deleteRecursive will delete an entry recursively, but not change its parent.
|
2020-06-12 13:28:21 -04:00
|
|
|
func (d *dataUsageCache) deleteRecursive(h dataUsageHash) {
|
|
|
|
if existing, ok := d.Cache[h.String()]; ok {
|
|
|
|
// Delete first if there should be a loop.
|
|
|
|
delete(d.Cache, h.Key())
|
|
|
|
for child := range existing.Children {
|
|
|
|
d.deleteRecursive(dataUsageHash(child))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-04-03 12:03:42 -04:00
|
|
|
// dui converts the flattened version of the path to madmin.DataUsageInfo.
|
2020-04-27 13:06:21 -04:00
|
|
|
// As a side effect d will be flattened, use a clone if this is not ok.
|
2021-09-18 16:31:35 -04:00
|
|
|
func (d *dataUsageCache) dui(path string, buckets []BucketInfo) DataUsageInfo {
|
2020-03-18 19:19:29 -04:00
|
|
|
e := d.find(path)
|
|
|
|
if e == nil {
|
2020-05-27 09:45:43 -04:00
|
|
|
// No entry found, return empty.
|
2021-09-18 16:31:35 -04:00
|
|
|
return DataUsageInfo{}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
flat := d.flatten(*e)
|
2021-09-18 16:31:35 -04:00
|
|
|
dui := DataUsageInfo{
|
2023-07-18 13:49:40 -04:00
|
|
|
LastUpdate: d.Info.LastUpdate,
|
|
|
|
ObjectsTotalCount: flat.Objects,
|
|
|
|
VersionsTotalCount: flat.Versions,
|
|
|
|
DeleteMarkersTotalCount: flat.DeleteMarkers,
|
|
|
|
ObjectsTotalSize: uint64(flat.Size),
|
|
|
|
BucketsCount: uint64(len(e.Children)),
|
|
|
|
BucketsUsage: d.bucketsUsageInfo(buckets),
|
|
|
|
TierStats: d.tiersUsageInfo(buckets),
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
|
|
|
return dui
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// replace will add or replace an entry in the cache.
|
|
|
|
// If a parent is specified it will be added to that if not already there.
|
|
|
|
// If the parent does not exist, it will be added.
|
|
|
|
func (d *dataUsageCache) replace(path, parent string, e dataUsageEntry) {
|
|
|
|
hash := hashPath(path)
|
|
|
|
if d.Cache == nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache = make(map[string]dataUsageEntry, 100)
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache[hash.Key()] = e
|
2020-03-18 19:19:29 -04:00
|
|
|
if parent != "" {
|
|
|
|
phash := hashPath(parent)
|
2020-06-12 13:28:21 -04:00
|
|
|
p := d.Cache[phash.Key()]
|
2020-03-18 19:19:29 -04:00
|
|
|
p.addChild(hash)
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache[phash.Key()] = p
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// replaceHashed add or replaces an entry to the cache based on its hash.
|
|
|
|
// If a parent is specified it will be added to that if not already there.
|
|
|
|
// If the parent does not exist, it will be added.
|
|
|
|
func (d *dataUsageCache) replaceHashed(hash dataUsageHash, parent *dataUsageHash, e dataUsageEntry) {
|
|
|
|
if d.Cache == nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache = make(map[string]dataUsageEntry, 100)
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache[hash.Key()] = e
|
2020-03-18 19:19:29 -04:00
|
|
|
if parent != nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
p := d.Cache[parent.Key()]
|
2020-03-18 19:19:29 -04:00
|
|
|
p.addChild(hash)
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache[parent.Key()] = p
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-04-27 13:06:21 -04:00
|
|
|
// copyWithChildren will copy entry with hash from src if it exists along with any children.
|
|
|
|
// If a parent is specified it will be added to that if not already there.
|
|
|
|
// If the parent does not exist, it will be added.
|
|
|
|
func (d *dataUsageCache) copyWithChildren(src *dataUsageCache, hash dataUsageHash, parent *dataUsageHash) {
|
|
|
|
if d.Cache == nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache = make(map[string]dataUsageEntry, 100)
|
2020-04-27 13:06:21 -04:00
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
e, ok := src.Cache[hash.String()]
|
2020-04-27 13:06:21 -04:00
|
|
|
if !ok {
|
|
|
|
return
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache[hash.Key()] = e
|
2020-04-27 13:06:21 -04:00
|
|
|
for ch := range e.Children {
|
2020-06-12 13:28:21 -04:00
|
|
|
if ch == hash.Key() {
|
2024-04-04 08:04:40 -04:00
|
|
|
scannerLogIf(GlobalContext, errors.New("dataUsageCache.copyWithChildren: Circular reference"))
|
2020-04-27 13:06:21 -04:00
|
|
|
return
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
d.copyWithChildren(src, dataUsageHash(ch), &hash)
|
2020-04-27 13:06:21 -04:00
|
|
|
}
|
|
|
|
if parent != nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
p := d.Cache[parent.Key()]
|
2020-04-27 13:06:21 -04:00
|
|
|
p.addChild(hash)
|
2020-06-12 13:28:21 -04:00
|
|
|
d.Cache[parent.Key()] = p
|
2020-04-27 13:06:21 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// reduceChildrenOf will reduce the recursive number of children to the limit
|
|
|
|
// by compacting the children with the least number of objects.
|
|
|
|
func (d *dataUsageCache) reduceChildrenOf(path dataUsageHash, limit int, compactSelf bool) {
|
|
|
|
e, ok := d.Cache[path.Key()]
|
|
|
|
if !ok {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
if e.Compacted {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
// If direct children have more, compact all.
|
|
|
|
if len(e.Children) > limit && compactSelf {
|
|
|
|
flat := d.sizeRecursive(path.Key())
|
|
|
|
flat.Compacted = true
|
|
|
|
d.deleteRecursive(path)
|
|
|
|
d.replaceHashed(path, nil, *flat)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
total := d.totalChildrenRec(path.Key())
|
|
|
|
if total < limit {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
// Appears to be printed with _MINIO_SERVER_DEBUG=off
|
|
|
|
// console.Debugf(" %d children found, compacting %v\n", total, path)
|
|
|
|
|
2022-01-02 12:15:06 -05:00
|
|
|
leaves := make([]struct {
|
2021-05-11 21:36:15 -04:00
|
|
|
objects uint64
|
|
|
|
path dataUsageHash
|
|
|
|
}, total)
|
|
|
|
// Collect current leaves that have children.
|
|
|
|
leaves = leaves[:0]
|
|
|
|
remove := total - limit
|
|
|
|
var add func(path dataUsageHash)
|
|
|
|
add = func(path dataUsageHash) {
|
|
|
|
e, ok := d.Cache[path.Key()]
|
|
|
|
if !ok {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
if len(e.Children) == 0 {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
sz := d.sizeRecursive(path.Key())
|
|
|
|
leaves = append(leaves, struct {
|
|
|
|
objects uint64
|
|
|
|
path dataUsageHash
|
|
|
|
}{objects: sz.Objects, path: path})
|
|
|
|
for ch := range e.Children {
|
|
|
|
add(dataUsageHash(ch))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Add path recursively.
|
|
|
|
add(path)
|
|
|
|
sort.Slice(leaves, func(i, j int) bool {
|
|
|
|
return leaves[i].objects < leaves[j].objects
|
|
|
|
})
|
|
|
|
for remove > 0 && len(leaves) > 0 {
|
|
|
|
// Remove top entry.
|
|
|
|
e := leaves[0]
|
|
|
|
candidate := e.path
|
|
|
|
if candidate == path && !compactSelf {
|
|
|
|
// We should be the biggest,
|
|
|
|
// if we cannot compact ourself, we are done.
|
|
|
|
break
|
|
|
|
}
|
|
|
|
removing := d.totalChildrenRec(candidate.Key())
|
|
|
|
flat := d.sizeRecursive(candidate.Key())
|
|
|
|
if flat == nil {
|
|
|
|
leaves = leaves[1:]
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
// Appears to be printed with _MINIO_SERVER_DEBUG=off
|
|
|
|
// console.Debugf("compacting %v, removing %d children\n", candidate, removing)
|
|
|
|
|
|
|
|
flat.Compacted = true
|
|
|
|
d.deleteRecursive(candidate)
|
|
|
|
d.replaceHashed(candidate, nil, *flat)
|
|
|
|
|
|
|
|
// Remove top entry and subtract removed children.
|
|
|
|
remove -= removing
|
|
|
|
leaves = leaves[1:]
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-05-10 10:49:50 -04:00
|
|
|
// forceCompact will force compact the cache of the top entry.
|
|
|
|
// If the number of children is more than limit*100, it will compact self.
|
|
|
|
// When above the limit a cleanup will also be performed to remove any possible abandoned entries.
|
|
|
|
func (d *dataUsageCache) forceCompact(limit int) {
|
|
|
|
if d == nil || len(d.Cache) <= limit {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
top := hashPath(d.Info.Name).Key()
|
|
|
|
topE := d.find(top)
|
|
|
|
if topE == nil {
|
|
|
|
scannerLogIf(GlobalContext, errors.New("forceCompact: root not found"))
|
|
|
|
return
|
|
|
|
}
|
|
|
|
// If off by 2 orders of magnitude, compact self and log error.
|
|
|
|
if len(topE.Children) > dataScannerForceCompactAtFolders {
|
|
|
|
// If we still have too many children, compact self.
|
|
|
|
scannerLogOnceIf(GlobalContext, fmt.Errorf("forceCompact: %q has %d children. Force compacting. Expect reduced scanner performance", d.Info.Name, len(topE.Children)), d.Info.Name)
|
|
|
|
d.reduceChildrenOf(hashPath(d.Info.Name), limit, true)
|
|
|
|
}
|
|
|
|
if len(d.Cache) <= limit {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
// Check for abandoned entries.
|
|
|
|
found := make(map[string]struct{}, len(d.Cache))
|
|
|
|
|
|
|
|
// Mark all children recursively
|
|
|
|
var mark func(entry dataUsageEntry)
|
|
|
|
mark = func(entry dataUsageEntry) {
|
|
|
|
for k := range entry.Children {
|
|
|
|
found[k] = struct{}{}
|
|
|
|
if ch, ok := d.Cache[k]; ok {
|
|
|
|
mark(ch)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
found[top] = struct{}{}
|
|
|
|
mark(*topE)
|
|
|
|
|
|
|
|
// Delete all entries not found.
|
|
|
|
for k := range d.Cache {
|
|
|
|
if _, ok := found[k]; !ok {
|
|
|
|
delete(d.Cache, k)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// StringAll returns a detailed string representation of all entries in the cache.
|
|
|
|
func (d *dataUsageCache) StringAll() string {
|
2021-08-04 12:14:14 -04:00
|
|
|
// Remove bloom filter from print.
|
2020-03-18 19:19:29 -04:00
|
|
|
s := fmt.Sprintf("info:%+v\n", d.Info)
|
|
|
|
for k, v := range d.Cache {
|
|
|
|
s += fmt.Sprintf("\t%v: %+v\n", k, v)
|
|
|
|
}
|
|
|
|
return strings.TrimSpace(s)
|
|
|
|
}
|
|
|
|
|
2020-06-12 13:28:21 -04:00
|
|
|
// String returns a human readable representation of the string.
|
|
|
|
func (h dataUsageHash) String() string {
|
|
|
|
return string(h)
|
2020-04-27 13:06:21 -04:00
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// Key returns the key.
|
2020-06-12 13:28:21 -04:00
|
|
|
func (h dataUsageHash) Key() string {
|
|
|
|
return string(h)
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
2021-07-13 13:42:11 -04:00
|
|
|
func (d *dataUsageCache) flattenChildrens(root dataUsageEntry) (m map[string]dataUsageEntry) {
|
|
|
|
m = make(map[string]dataUsageEntry)
|
|
|
|
for id := range root.Children {
|
|
|
|
e := d.Cache[id]
|
|
|
|
if len(e.Children) > 0 {
|
|
|
|
e = d.flatten(e)
|
|
|
|
}
|
|
|
|
m[id] = e
|
|
|
|
}
|
|
|
|
return m
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// flatten all children of the root into the root element and return it.
|
|
|
|
func (d *dataUsageCache) flatten(root dataUsageEntry) dataUsageEntry {
|
|
|
|
for id := range root.Children {
|
|
|
|
e := d.Cache[id]
|
|
|
|
if len(e.Children) > 0 {
|
|
|
|
e = d.flatten(e)
|
|
|
|
}
|
|
|
|
root.merge(e)
|
|
|
|
}
|
|
|
|
root.Children = nil
|
|
|
|
return root
|
|
|
|
}
|
|
|
|
|
|
|
|
// add a size to the histogram.
|
|
|
|
func (h *sizeHistogram) add(size int64) {
|
|
|
|
// Fetch the histogram interval corresponding
|
|
|
|
// to the passed object size.
|
2023-03-10 11:53:59 -05:00
|
|
|
for i, interval := range ObjectsHistogramIntervals[:] {
|
2020-03-18 19:19:29 -04:00
|
|
|
if size >= interval.start && size <= interval.end {
|
|
|
|
h[i]++
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-01-13 02:51:08 -05:00
|
|
|
// mergeV1 is used to migrate data usage cache from sizeHistogramV1 to
|
|
|
|
// sizeHistogram
|
|
|
|
func (h *sizeHistogram) mergeV1(v sizeHistogramV1) {
|
|
|
|
var oidx, nidx int
|
|
|
|
for oidx < len(v) {
|
|
|
|
intOld, intNew := ObjectsHistogramIntervalsV1[oidx], ObjectsHistogramIntervals[nidx]
|
|
|
|
// skip intervals that aren't common to both histograms
|
|
|
|
if intOld.start != intNew.start || intOld.end != intNew.end {
|
|
|
|
nidx++
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
h[nidx] += v[oidx]
|
|
|
|
oidx++
|
|
|
|
nidx++
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-05-27 09:45:43 -04:00
|
|
|
// toMap returns the map to a map[string]uint64.
|
|
|
|
func (h *sizeHistogram) toMap() map[string]uint64 {
|
|
|
|
res := make(map[string]uint64, dataUsageBucketLen)
|
2024-01-13 02:51:08 -05:00
|
|
|
var splCount uint64
|
2020-03-18 19:19:29 -04:00
|
|
|
for i, count := range h {
|
2024-01-13 02:51:08 -05:00
|
|
|
szInt := ObjectsHistogramIntervals[i]
|
|
|
|
switch {
|
|
|
|
case humanize.KiByte == szInt.start && szInt.end == humanize.MiByte-1:
|
|
|
|
// spl interval: [1024B, 1MiB)
|
|
|
|
res[szInt.name] = splCount
|
|
|
|
case humanize.KiByte <= szInt.start && szInt.end <= humanize.MiByte-1:
|
|
|
|
// intervals that fall within the spl interval above; they
|
|
|
|
// appear earlier in this array of intervals, see
|
|
|
|
// ObjectsHistogramIntervals
|
|
|
|
splCount += count
|
|
|
|
fallthrough
|
|
|
|
default:
|
|
|
|
res[szInt.name] = count
|
|
|
|
}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
return res
|
|
|
|
}
|
|
|
|
|
2023-03-10 11:53:59 -05:00
|
|
|
// add a version count to the histogram.
|
|
|
|
func (h *versionsHistogram) add(versions uint64) {
|
|
|
|
// Fetch the histogram interval corresponding
|
|
|
|
// to the passed object size.
|
|
|
|
for i, interval := range ObjectsVersionCountIntervals[:] {
|
|
|
|
if versions >= uint64(interval.start) && versions <= uint64(interval.end) {
|
|
|
|
h[i]++
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// toMap returns the map to a map[string]uint64.
|
|
|
|
func (h *versionsHistogram) toMap() map[string]uint64 {
|
|
|
|
res := make(map[string]uint64, dataUsageVersionLen)
|
|
|
|
for i, count := range h {
|
|
|
|
res[ObjectsVersionCountIntervals[i].name] = count
|
|
|
|
}
|
|
|
|
return res
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
func (d *dataUsageCache) tiersUsageInfo(buckets []BucketInfo) *allTierStats {
|
|
|
|
dst := newAllTierStats()
|
|
|
|
for _, bucket := range buckets {
|
|
|
|
e := d.find(bucket.Name)
|
|
|
|
if e == nil {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
flat := d.flatten(*e)
|
|
|
|
if flat.AllTierStats == nil {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
dst.merge(flat.AllTierStats)
|
|
|
|
}
|
|
|
|
if len(dst.Tiers) == 0 {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
return dst
|
|
|
|
}
|
|
|
|
|
2020-05-27 09:45:43 -04:00
|
|
|
// bucketsUsageInfo returns the buckets usage info as a map, with
|
|
|
|
// key as bucket name
|
2021-09-18 16:31:35 -04:00
|
|
|
func (d *dataUsageCache) bucketsUsageInfo(buckets []BucketInfo) map[string]BucketUsageInfo {
|
2022-01-02 12:15:06 -05:00
|
|
|
dst := make(map[string]BucketUsageInfo, len(buckets))
|
2020-03-18 19:19:29 -04:00
|
|
|
for _, bucket := range buckets {
|
|
|
|
e := d.find(bucket.Name)
|
|
|
|
if e == nil {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
flat := d.flatten(*e)
|
2021-09-18 16:31:35 -04:00
|
|
|
bui := BucketUsageInfo{
|
2023-03-10 11:53:59 -05:00
|
|
|
Size: uint64(flat.Size),
|
|
|
|
VersionsCount: flat.Versions,
|
|
|
|
ObjectsCount: flat.Objects,
|
2023-07-18 13:49:40 -04:00
|
|
|
DeleteMarkersCount: flat.DeleteMarkers,
|
2023-03-10 11:53:59 -05:00
|
|
|
ObjectSizesHistogram: flat.ObjSizes.toMap(),
|
|
|
|
ObjectVersionsHistogram: flat.ObjVersions.toMap(),
|
2020-05-27 09:45:43 -04:00
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
if flat.ReplicationStats != nil {
|
|
|
|
bui.ReplicaSize = flat.ReplicationStats.ReplicaSize
|
2023-08-30 04:00:59 -04:00
|
|
|
bui.ReplicaCount = flat.ReplicationStats.ReplicaCount
|
|
|
|
|
2021-09-18 16:31:35 -04:00
|
|
|
bui.ReplicationInfo = make(map[string]BucketTargetUsageInfo, len(flat.ReplicationStats.Targets))
|
|
|
|
for arn, stat := range flat.ReplicationStats.Targets {
|
|
|
|
bui.ReplicationInfo[arn] = BucketTargetUsageInfo{
|
|
|
|
ReplicationPendingSize: stat.PendingSize,
|
|
|
|
ReplicatedSize: stat.ReplicatedSize,
|
|
|
|
ReplicationFailedSize: stat.FailedSize,
|
|
|
|
ReplicationPendingCount: stat.PendingCount,
|
|
|
|
ReplicationFailedCount: stat.FailedCount,
|
2023-08-30 04:00:59 -04:00
|
|
|
ReplicatedCount: stat.ReplicatedCount,
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
|
|
|
dst[bucket.Name] = bui
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
return dst
|
|
|
|
}
|
|
|
|
|
|
|
|
// sizeRecursive returns the path as a flattened entry.
|
|
|
|
func (d *dataUsageCache) sizeRecursive(path string) *dataUsageEntry {
|
|
|
|
root := d.find(path)
|
|
|
|
if root == nil || len(root.Children) == 0 {
|
|
|
|
return root
|
|
|
|
}
|
|
|
|
flat := d.flatten(*root)
|
2024-03-28 13:13:07 -04:00
|
|
|
if flat.ReplicationStats.empty() {
|
|
|
|
flat.ReplicationStats = nil
|
|
|
|
}
|
2020-03-18 19:19:29 -04:00
|
|
|
return &flat
|
|
|
|
}
|
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// totalChildrenRec returns the total number of children recorded.
|
|
|
|
func (d *dataUsageCache) totalChildrenRec(path string) int {
|
|
|
|
root := d.find(path)
|
|
|
|
if root == nil || len(root.Children) == 0 {
|
|
|
|
return 0
|
|
|
|
}
|
|
|
|
n := len(root.Children)
|
|
|
|
for ch := range root.Children {
|
|
|
|
n += d.totalChildrenRec(ch)
|
|
|
|
}
|
|
|
|
return n
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// root returns the root of the cache.
|
|
|
|
func (d *dataUsageCache) root() *dataUsageEntry {
|
|
|
|
return d.find(d.Info.Name)
|
|
|
|
}
|
|
|
|
|
|
|
|
// rootHash returns the root of the cache.
|
|
|
|
func (d *dataUsageCache) rootHash() dataUsageHash {
|
|
|
|
return hashPath(d.Info.Name)
|
|
|
|
}
|
|
|
|
|
|
|
|
// clone returns a copy of the cache with no references to the existing.
|
|
|
|
func (d *dataUsageCache) clone() dataUsageCache {
|
|
|
|
clone := dataUsageCache{
|
|
|
|
Info: d.Info,
|
2020-06-12 13:28:21 -04:00
|
|
|
Cache: make(map[string]dataUsageEntry, len(d.Cache)),
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
for k, v := range d.Cache {
|
2021-08-24 10:11:38 -04:00
|
|
|
clone.Cache[k] = v.clone()
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
return clone
|
|
|
|
}
|
|
|
|
|
|
|
|
// merge root of other into d.
|
|
|
|
// children of root will be flattened before being merged.
|
|
|
|
// Last update time will be set to the last updated.
|
|
|
|
func (d *dataUsageCache) merge(other dataUsageCache) {
|
|
|
|
existingRoot := d.root()
|
|
|
|
otherRoot := other.root()
|
|
|
|
if existingRoot == nil && otherRoot == nil {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
if otherRoot == nil {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
if existingRoot == nil {
|
|
|
|
*d = other.clone()
|
|
|
|
return
|
|
|
|
}
|
|
|
|
if other.Info.LastUpdate.After(d.Info.LastUpdate) {
|
|
|
|
d.Info.LastUpdate = other.Info.LastUpdate
|
|
|
|
}
|
|
|
|
existingRoot.merge(*otherRoot)
|
|
|
|
eHash := d.rootHash()
|
|
|
|
for key := range otherRoot.Children {
|
|
|
|
entry := other.Cache[key]
|
|
|
|
flat := other.flatten(entry)
|
|
|
|
existing := d.Cache[key]
|
|
|
|
// If not found, merging simply adds.
|
|
|
|
existing.merge(flat)
|
2020-06-12 13:28:21 -04:00
|
|
|
d.replaceHashed(dataUsageHash(key), &eHash, existing)
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-09-10 12:18:19 -04:00
|
|
|
type objectIO interface {
|
2023-04-17 15:16:37 -04:00
|
|
|
GetObjectNInfo(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, opts ObjectOptions) (reader *GetObjectReader, err error)
|
2020-09-10 12:18:19 -04:00
|
|
|
PutObject(ctx context.Context, bucket, object string, data *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error)
|
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// load the cache content with name from minioMetaBackgroundOpsBucket.
|
|
|
|
// Only backend errors are returned as errors.
|
2023-03-09 18:15:46 -05:00
|
|
|
// The loader is optimistic and has no locking, but tries 5 times before giving up.
|
2023-10-02 22:22:35 -04:00
|
|
|
// If the object is not found, a nil error with empty data usage cache is returned.
|
2020-09-10 12:18:19 -04:00
|
|
|
func (d *dataUsageCache) load(ctx context.Context, store objectIO, name string) error {
|
2024-01-18 02:03:17 -05:00
|
|
|
// By default, empty data usage cache
|
2023-09-15 16:11:08 -04:00
|
|
|
*d = dataUsageCache{}
|
|
|
|
|
2023-09-14 14:53:52 -04:00
|
|
|
load := func(name string, timeout time.Duration) (bool, error) {
|
|
|
|
// Abandon if more than time.Minute, so we don't hold up scanner.
|
|
|
|
// drive timeout by default is 2 minutes, we do not need to wait longer.
|
|
|
|
ctx, cancel := context.WithTimeout(ctx, timeout)
|
|
|
|
defer cancel()
|
2022-01-03 13:22:58 -05:00
|
|
|
|
2024-04-22 13:49:30 -04:00
|
|
|
r, err := store.GetObjectNInfo(ctx, minioMetaBucket, pathJoin(bucketMetaPrefix, name), nil, http.Header{}, ObjectOptions{NoLock: true})
|
2023-03-09 18:15:46 -05:00
|
|
|
if err != nil {
|
|
|
|
switch err.(type) {
|
|
|
|
case ObjectNotFound, BucketNotFound:
|
2024-04-22 13:49:30 -04:00
|
|
|
r, err = store.GetObjectNInfo(ctx, dataUsageBucket, name, nil, http.Header{}, ObjectOptions{NoLock: true})
|
|
|
|
if err != nil {
|
|
|
|
switch err.(type) {
|
|
|
|
case ObjectNotFound, BucketNotFound:
|
|
|
|
return false, nil
|
|
|
|
case InsufficientReadQuorum, StorageErr:
|
|
|
|
return true, nil
|
|
|
|
}
|
|
|
|
return false, err
|
|
|
|
}
|
|
|
|
err = d.deserialize(r)
|
|
|
|
r.Close()
|
|
|
|
return err != nil, nil
|
2023-03-09 18:15:46 -05:00
|
|
|
case InsufficientReadQuorum, StorageErr:
|
2023-09-14 14:53:52 -04:00
|
|
|
return true, nil
|
2023-03-09 18:15:46 -05:00
|
|
|
}
|
2023-09-15 16:11:08 -04:00
|
|
|
return false, err
|
2023-09-14 14:53:52 -04:00
|
|
|
}
|
|
|
|
err = d.deserialize(r)
|
|
|
|
r.Close()
|
|
|
|
return err != nil, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
// Caches are read+written without locks,
|
|
|
|
retries := 0
|
|
|
|
for retries < 5 {
|
|
|
|
retry, err := load(name, time.Minute)
|
|
|
|
if err != nil {
|
2023-09-15 16:11:08 -04:00
|
|
|
return toObjectErr(err, dataUsageBucket, name)
|
2023-03-09 18:15:46 -05:00
|
|
|
}
|
2023-09-15 16:11:08 -04:00
|
|
|
if !retry {
|
|
|
|
break
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2023-10-02 22:22:35 -04:00
|
|
|
retry, err = load(name+".bkp", 30*time.Second)
|
|
|
|
if err == nil && !retry {
|
|
|
|
// Only return when we have valid data from the backup
|
2023-09-15 16:11:08 -04:00
|
|
|
break
|
|
|
|
}
|
|
|
|
retries++
|
|
|
|
time.Sleep(time.Duration(rand.Int63n(int64(time.Second))))
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2023-10-02 22:22:35 -04:00
|
|
|
|
|
|
|
if retries == 5 {
|
2024-04-04 08:04:40 -04:00
|
|
|
scannerLogOnceIf(ctx, fmt.Errorf("maximum retry reached to load the data usage cache `%s`", name), "retry-loading-data-usage-cache")
|
2023-10-02 22:22:35 -04:00
|
|
|
}
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
2023-03-09 18:15:46 -05:00
|
|
|
// Maximum running concurrent saves on server.
|
|
|
|
var maxConcurrentScannerSaves = make(chan struct{}, 4)
|
|
|
|
|
2020-03-18 19:19:29 -04:00
|
|
|
// save the content of the cache to minioMetaBackgroundOpsBucket with the provided name.
|
2023-03-09 18:15:46 -05:00
|
|
|
// Note that no locking is done when saving.
|
2020-09-10 12:18:19 -04:00
|
|
|
func (d *dataUsageCache) save(ctx context.Context, store objectIO, name string) error {
|
2023-09-14 14:53:52 -04:00
|
|
|
select {
|
|
|
|
case <-ctx.Done():
|
|
|
|
return ctx.Err()
|
|
|
|
case maxConcurrentScannerSaves <- struct{}{}:
|
|
|
|
}
|
|
|
|
|
|
|
|
buf := bytebufferpool.Get()
|
|
|
|
defer func() {
|
2023-11-01 12:09:28 -04:00
|
|
|
<-maxConcurrentScannerSaves
|
2023-09-14 14:53:52 -04:00
|
|
|
buf.Reset()
|
|
|
|
bytebufferpool.Put(buf)
|
|
|
|
}()
|
|
|
|
|
|
|
|
if err := d.serializeTo(buf); err != nil {
|
|
|
|
return err
|
2022-12-05 16:01:11 -05:00
|
|
|
}
|
2020-12-31 12:45:09 -05:00
|
|
|
|
2023-09-14 14:53:52 -04:00
|
|
|
save := func(name string, timeout time.Duration) error {
|
|
|
|
// Abandon if more than a minute, so we don't hold up scanner.
|
|
|
|
ctx, cancel := context.WithTimeout(ctx, timeout)
|
|
|
|
defer cancel()
|
|
|
|
|
2024-04-22 13:49:30 -04:00
|
|
|
return saveConfig(ctx, store, pathJoin(bucketMetaPrefix, name), buf.Bytes())
|
2020-04-28 04:16:57 -04:00
|
|
|
}
|
2023-11-01 12:09:28 -04:00
|
|
|
defer save(name+".bkp", 5*time.Second) // Keep a backup as well
|
2023-09-14 14:53:52 -04:00
|
|
|
|
|
|
|
// drive timeout by default is 2 minutes, we do not need to wait longer.
|
|
|
|
return save(name, time.Minute)
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// dataUsageCacheVer indicates the cache version.
|
|
|
|
// Bumping the cache version will drop data from previous versions
|
|
|
|
// and write new data with the new version.
|
2021-01-13 12:58:08 -05:00
|
|
|
const (
|
2024-01-13 02:51:08 -05:00
|
|
|
dataUsageCacheVerCurrent = 8
|
|
|
|
dataUsageCacheVerV7 = 7
|
2021-10-23 21:38:33 -04:00
|
|
|
dataUsageCacheVerV6 = 6
|
2021-09-18 16:31:35 -04:00
|
|
|
dataUsageCacheVerV5 = 5
|
2021-05-11 21:36:15 -04:00
|
|
|
dataUsageCacheVerV4 = 4
|
|
|
|
dataUsageCacheVerV3 = 3
|
|
|
|
dataUsageCacheVerV2 = 2
|
|
|
|
dataUsageCacheVerV1 = 1
|
2021-01-13 12:58:08 -05:00
|
|
|
)
|
2020-03-18 19:19:29 -04:00
|
|
|
|
|
|
|
// serialize the contents of the cache.
|
2020-12-10 16:03:22 -05:00
|
|
|
func (d *dataUsageCache) serializeTo(dst io.Writer) error {
|
|
|
|
// Add version and compress.
|
2021-05-11 21:36:15 -04:00
|
|
|
_, err := dst.Write([]byte{dataUsageCacheVerCurrent})
|
2020-12-10 16:03:22 -05:00
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
enc, err := zstd.NewWriter(dst,
|
2020-06-12 13:28:21 -04:00
|
|
|
zstd.WithEncoderLevel(zstd.SpeedFastest),
|
|
|
|
zstd.WithWindowSize(1<<20),
|
|
|
|
zstd.WithEncoderConcurrency(2))
|
2020-03-18 19:19:29 -04:00
|
|
|
if err != nil {
|
2020-12-10 16:03:22 -05:00
|
|
|
return err
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
mEnc := msgp.NewWriter(enc)
|
|
|
|
err = d.EncodeMsg(mEnc)
|
|
|
|
if err != nil {
|
2020-12-10 16:03:22 -05:00
|
|
|
return err
|
|
|
|
}
|
|
|
|
err = mEnc.Flush()
|
|
|
|
if err != nil {
|
|
|
|
return err
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
err = enc.Close()
|
|
|
|
if err != nil {
|
2020-12-10 16:03:22 -05:00
|
|
|
return err
|
2020-06-12 13:28:21 -04:00
|
|
|
}
|
2020-12-10 16:03:22 -05:00
|
|
|
return nil
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// deserialize the supplied byte slice into the cache.
|
2020-06-12 13:28:21 -04:00
|
|
|
func (d *dataUsageCache) deserialize(r io.Reader) error {
|
|
|
|
var b [1]byte
|
|
|
|
n, _ := r.Read(b[:])
|
|
|
|
if n != 1 {
|
2020-03-18 19:19:29 -04:00
|
|
|
return io.ErrUnexpectedEOF
|
|
|
|
}
|
2021-05-12 11:07:02 -04:00
|
|
|
ver := int(b[0])
|
2021-05-11 21:36:15 -04:00
|
|
|
switch ver {
|
2021-01-13 12:58:08 -05:00
|
|
|
case dataUsageCacheVerV1:
|
2020-06-12 13:28:21 -04:00
|
|
|
return errors.New("cache version deprecated (will autoupdate)")
|
2021-01-13 12:58:08 -05:00
|
|
|
case dataUsageCacheVerV2:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
2020-03-18 19:19:29 -04:00
|
|
|
|
2021-01-13 12:58:08 -05:00
|
|
|
dold := &dataUsageCacheV2{}
|
|
|
|
if err = dold.DecodeMsg(msgp.NewReader(dec)); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
d.Info = dold.Info
|
|
|
|
d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
|
|
|
|
for k, v := range dold.Cache {
|
|
|
|
d.Cache[k] = dataUsageEntry{
|
2021-05-11 21:36:15 -04:00
|
|
|
Size: v.Size,
|
|
|
|
Objects: v.Objects,
|
|
|
|
ObjSizes: v.ObjSizes,
|
|
|
|
Children: v.Children,
|
|
|
|
Compacted: len(v.Children) == 0 && k != d.Info.Name,
|
2021-01-13 12:58:08 -05:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return nil
|
|
|
|
case dataUsageCacheVerV3:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
2021-04-03 12:03:42 -04:00
|
|
|
dold := &dataUsageCacheV3{}
|
|
|
|
if err = dold.DecodeMsg(msgp.NewReader(dec)); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
d.Info = dold.Info
|
|
|
|
d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
|
|
|
|
for k, v := range dold.Cache {
|
2021-05-11 21:36:15 -04:00
|
|
|
due := dataUsageEntry{
|
2021-04-03 12:03:42 -04:00
|
|
|
Size: v.Size,
|
|
|
|
Objects: v.Objects,
|
|
|
|
ObjSizes: v.ObjSizes,
|
|
|
|
Children: v.Children,
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
|
|
|
if v.ReplicatedSize > 0 || v.ReplicaSize > 0 || v.ReplicationFailedSize > 0 || v.ReplicationPendingSize > 0 {
|
2021-11-19 17:46:14 -05:00
|
|
|
cfg, _ := getReplicationConfig(GlobalContext, d.Info.Name)
|
|
|
|
if cfg != nil && cfg.RoleArn != "" {
|
|
|
|
due.ReplicationStats = &replicationAllStats{
|
|
|
|
Targets: make(map[string]replicationStats),
|
|
|
|
}
|
|
|
|
due.ReplicationStats.ReplicaSize = v.ReplicaSize
|
|
|
|
due.ReplicationStats.Targets[cfg.RoleArn] = replicationStats{
|
|
|
|
ReplicatedSize: v.ReplicatedSize,
|
|
|
|
FailedSize: v.ReplicationFailedSize,
|
|
|
|
PendingSize: v.ReplicationPendingSize,
|
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
2021-04-03 12:03:42 -04:00
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
due.Compacted = len(due.Children) == 0 && k != d.Info.Name
|
|
|
|
|
|
|
|
d.Cache[k] = due
|
2021-04-03 12:03:42 -04:00
|
|
|
}
|
|
|
|
return nil
|
|
|
|
case dataUsageCacheVerV4:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
2021-05-11 21:36:15 -04:00
|
|
|
dold := &dataUsageCacheV4{}
|
|
|
|
if err = dold.DecodeMsg(msgp.NewReader(dec)); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
d.Info = dold.Info
|
|
|
|
d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
|
|
|
|
for k, v := range dold.Cache {
|
|
|
|
due := dataUsageEntry{
|
|
|
|
Size: v.Size,
|
|
|
|
Objects: v.Objects,
|
|
|
|
ObjSizes: v.ObjSizes,
|
|
|
|
Children: v.Children,
|
|
|
|
}
|
2021-09-21 12:01:51 -04:00
|
|
|
empty := replicationStatsV1{}
|
2021-09-18 16:31:35 -04:00
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
if v.ReplicationStats != empty {
|
2021-11-19 17:46:14 -05:00
|
|
|
cfg, _ := getReplicationConfig(GlobalContext, d.Info.Name)
|
|
|
|
if cfg != nil && cfg.RoleArn != "" {
|
|
|
|
due.ReplicationStats = &replicationAllStats{
|
|
|
|
Targets: make(map[string]replicationStats),
|
|
|
|
}
|
|
|
|
due.ReplicationStats.Targets[cfg.RoleArn] = replicationStats{
|
|
|
|
ReplicatedSize: v.ReplicationStats.ReplicatedSize,
|
|
|
|
FailedSize: v.ReplicationStats.FailedSize,
|
|
|
|
FailedCount: v.ReplicationStats.FailedCount,
|
|
|
|
PendingSize: v.ReplicationStats.PendingSize,
|
|
|
|
PendingCount: v.ReplicationStats.PendingCount,
|
|
|
|
}
|
|
|
|
due.ReplicationStats.ReplicaSize = v.ReplicationStats.ReplicaSize
|
2021-09-21 12:01:51 -04:00
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
}
|
|
|
|
due.Compacted = len(due.Children) == 0 && k != d.Info.Name
|
|
|
|
|
|
|
|
d.Cache[k] = due
|
|
|
|
}
|
2020-03-18 19:19:29 -04:00
|
|
|
|
2021-05-11 21:36:15 -04:00
|
|
|
// Populate compacted value and remove unneeded replica stats.
|
|
|
|
for k, e := range d.Cache {
|
2021-09-18 16:31:35 -04:00
|
|
|
if e.ReplicationStats != nil && len(e.ReplicationStats.Targets) == 0 {
|
2021-05-11 21:36:15 -04:00
|
|
|
e.ReplicationStats = nil
|
|
|
|
}
|
2021-09-18 16:31:35 -04:00
|
|
|
d.Cache[k] = e
|
|
|
|
}
|
|
|
|
return nil
|
|
|
|
case dataUsageCacheVerV5:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
|
|
|
dold := &dataUsageCacheV5{}
|
|
|
|
if err = dold.DecodeMsg(msgp.NewReader(dec)); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
d.Info = dold.Info
|
|
|
|
d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
|
|
|
|
for k, v := range dold.Cache {
|
|
|
|
due := dataUsageEntry{
|
|
|
|
Size: v.Size,
|
|
|
|
Objects: v.Objects,
|
|
|
|
ObjSizes: v.ObjSizes,
|
|
|
|
Children: v.Children,
|
|
|
|
}
|
|
|
|
if v.ReplicationStats != nil && !v.ReplicationStats.Empty() {
|
2021-11-19 17:46:14 -05:00
|
|
|
cfg, _ := getReplicationConfig(GlobalContext, d.Info.Name)
|
|
|
|
if cfg != nil && cfg.RoleArn != "" {
|
|
|
|
due.ReplicationStats = &replicationAllStats{
|
|
|
|
Targets: make(map[string]replicationStats),
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
|
|
|
d.Info.replication = replicationConfig{Config: cfg}
|
|
|
|
|
2021-11-19 17:46:14 -05:00
|
|
|
due.ReplicationStats.Targets[cfg.RoleArn] = replicationStats{
|
2021-09-21 12:01:51 -04:00
|
|
|
ReplicatedSize: v.ReplicationStats.ReplicatedSize,
|
|
|
|
FailedSize: v.ReplicationStats.FailedSize,
|
|
|
|
FailedCount: v.ReplicationStats.FailedCount,
|
|
|
|
PendingSize: v.ReplicationStats.PendingSize,
|
2021-09-22 13:48:45 -04:00
|
|
|
PendingCount: v.ReplicationStats.PendingCount,
|
2021-09-21 12:01:51 -04:00
|
|
|
}
|
|
|
|
due.ReplicationStats.ReplicaSize = v.ReplicationStats.ReplicaSize
|
2021-09-18 16:31:35 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
due.Compacted = len(due.Children) == 0 && k != d.Info.Name
|
|
|
|
|
|
|
|
d.Cache[k] = due
|
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
|
2021-09-18 16:31:35 -04:00
|
|
|
// Populate compacted value and remove unneeded replica stats.
|
|
|
|
for k, e := range d.Cache {
|
|
|
|
if e.ReplicationStats != nil && len(e.ReplicationStats.Targets) == 0 {
|
|
|
|
e.ReplicationStats = nil
|
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
d.Cache[k] = e
|
|
|
|
}
|
2021-05-12 11:07:02 -04:00
|
|
|
return nil
|
2021-10-23 21:38:33 -04:00
|
|
|
case dataUsageCacheVerV6:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
|
|
|
dold := &dataUsageCacheV6{}
|
|
|
|
if err = dold.DecodeMsg(msgp.NewReader(dec)); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
d.Info = dold.Info
|
|
|
|
d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
|
|
|
|
for k, v := range dold.Cache {
|
|
|
|
var replicationStats *replicationAllStats
|
|
|
|
if v.ReplicationStats != nil {
|
|
|
|
replicationStats = &replicationAllStats{
|
2023-08-30 04:00:59 -04:00
|
|
|
Targets: v.ReplicationStats.Targets,
|
|
|
|
ReplicaSize: v.ReplicationStats.ReplicaSize,
|
|
|
|
ReplicaCount: v.ReplicationStats.ReplicaCount,
|
2021-10-23 21:38:33 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
due := dataUsageEntry{
|
|
|
|
Children: v.Children,
|
|
|
|
Size: v.Size,
|
|
|
|
Objects: v.Objects,
|
|
|
|
Versions: v.Versions,
|
|
|
|
ObjSizes: v.ObjSizes,
|
|
|
|
ReplicationStats: replicationStats,
|
|
|
|
Compacted: v.Compacted,
|
|
|
|
}
|
|
|
|
d.Cache[k] = due
|
|
|
|
}
|
2024-01-13 02:51:08 -05:00
|
|
|
return nil
|
|
|
|
case dataUsageCacheVerV7:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
|
|
|
dold := &dataUsageCacheV7{}
|
|
|
|
if err = dold.DecodeMsg(msgp.NewReader(dec)); err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
d.Info = dold.Info
|
|
|
|
d.Cache = make(map[string]dataUsageEntry, len(dold.Cache))
|
|
|
|
for k, v := range dold.Cache {
|
|
|
|
var szHist sizeHistogram
|
|
|
|
szHist.mergeV1(v.ObjSizes)
|
|
|
|
d.Cache[k] = dataUsageEntry{
|
|
|
|
Children: v.Children,
|
|
|
|
Size: v.Size,
|
|
|
|
Objects: v.Objects,
|
|
|
|
Versions: v.Versions,
|
|
|
|
ObjSizes: szHist,
|
|
|
|
ReplicationStats: v.ReplicationStats,
|
|
|
|
Compacted: v.Compacted,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-10-23 21:38:33 -04:00
|
|
|
return nil
|
2021-05-11 21:36:15 -04:00
|
|
|
case dataUsageCacheVerCurrent:
|
|
|
|
// Zstd compressed.
|
|
|
|
dec, err := zstd.NewReader(r, zstd.WithDecoderConcurrency(2))
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer dec.Close()
|
2021-01-13 12:58:08 -05:00
|
|
|
return d.DecodeMsg(msgp.NewReader(dec))
|
2021-05-12 11:07:02 -04:00
|
|
|
default:
|
|
|
|
return fmt.Errorf("dataUsageCache: unknown version: %d", ver)
|
2021-01-13 12:58:08 -05:00
|
|
|
}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// Trim this from start+end of hashes.
|
|
|
|
var hashPathCutSet = dataUsageRoot
|
|
|
|
|
|
|
|
func init() {
|
|
|
|
if dataUsageRoot != string(filepath.Separator) {
|
|
|
|
hashPathCutSet = dataUsageRoot + string(filepath.Separator)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// hashPath calculates a hash of the provided string.
|
|
|
|
func hashPath(data string) dataUsageHash {
|
|
|
|
if data != dataUsageRoot {
|
|
|
|
data = strings.Trim(data, hashPathCutSet)
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
return dataUsageHash(path.Clean(data))
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
|
2020-06-12 13:28:21 -04:00
|
|
|
//msgp:ignore dataUsageHashMap
|
|
|
|
type dataUsageHashMap map[string]struct{}
|
2020-03-18 19:19:29 -04:00
|
|
|
|
2020-06-12 13:28:21 -04:00
|
|
|
// DecodeMsg implements msgp.Decodable
|
|
|
|
func (z *dataUsageHashMap) DecodeMsg(dc *msgp.Reader) (err error) {
|
|
|
|
var zb0002 uint32
|
|
|
|
zb0002, err = dc.ReadArrayHeader()
|
2020-03-18 19:19:29 -04:00
|
|
|
if err != nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
err = msgp.WrapError(err)
|
2020-03-18 19:19:29 -04:00
|
|
|
return
|
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
if zb0002 == 0 {
|
|
|
|
*z = nil
|
|
|
|
return
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
*z = make(dataUsageHashMap, zb0002)
|
|
|
|
for i := uint32(0); i < zb0002; i++ {
|
|
|
|
{
|
|
|
|
var zb0003 string
|
|
|
|
zb0003, err = dc.ReadString()
|
|
|
|
if err != nil {
|
|
|
|
err = msgp.WrapError(err)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
(*z)[zb0003] = struct{}{}
|
|
|
|
}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2020-06-12 13:28:21 -04:00
|
|
|
// EncodeMsg implements msgp.Encodable
|
|
|
|
func (z dataUsageHashMap) EncodeMsg(en *msgp.Writer) (err error) {
|
|
|
|
err = en.WriteArrayHeader(uint32(len(z)))
|
2020-03-18 19:19:29 -04:00
|
|
|
if err != nil {
|
|
|
|
err = msgp.WrapError(err)
|
|
|
|
return
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
for zb0004 := range z {
|
|
|
|
err = en.WriteString(zb0004)
|
2020-03-18 19:19:29 -04:00
|
|
|
if err != nil {
|
2020-06-12 13:28:21 -04:00
|
|
|
err = msgp.WrapError(err, zb0004)
|
2020-03-18 19:19:29 -04:00
|
|
|
return
|
|
|
|
}
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
// MarshalMsg implements msgp.Marshaler
|
|
|
|
func (z dataUsageHashMap) MarshalMsg(b []byte) (o []byte, err error) {
|
|
|
|
o = msgp.Require(b, z.Msgsize())
|
|
|
|
o = msgp.AppendArrayHeader(o, uint32(len(z)))
|
|
|
|
for zb0004 := range z {
|
|
|
|
o = msgp.AppendString(o, zb0004)
|
|
|
|
}
|
|
|
|
return
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
|
|
|
|
// UnmarshalMsg implements msgp.Unmarshaler
|
|
|
|
func (z *dataUsageHashMap) UnmarshalMsg(bts []byte) (o []byte, err error) {
|
|
|
|
var zb0002 uint32
|
|
|
|
zb0002, bts, err = msgp.ReadArrayHeaderBytes(bts)
|
2020-03-18 19:19:29 -04:00
|
|
|
if err != nil {
|
|
|
|
err = msgp.WrapError(err)
|
|
|
|
return
|
|
|
|
}
|
2021-05-11 21:36:15 -04:00
|
|
|
if zb0002 == 0 {
|
|
|
|
*z = nil
|
|
|
|
return bts, nil
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
*z = make(dataUsageHashMap, zb0002)
|
|
|
|
for i := uint32(0); i < zb0002; i++ {
|
|
|
|
{
|
|
|
|
var zb0003 string
|
|
|
|
zb0003, bts, err = msgp.ReadStringBytes(bts)
|
|
|
|
if err != nil {
|
|
|
|
err = msgp.WrapError(err)
|
|
|
|
return
|
|
|
|
}
|
|
|
|
(*z)[zb0003] = struct{}{}
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
|
|
|
}
|
2020-06-12 13:28:21 -04:00
|
|
|
o = bts
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
// Msgsize returns an upper bound estimate of the number of bytes occupied by the serialized message
|
|
|
|
func (z dataUsageHashMap) Msgsize() (s int) {
|
|
|
|
s = msgp.ArrayHeaderSize
|
|
|
|
for zb0004 := range z {
|
|
|
|
s += msgp.StringPrefixSize + len(zb0004)
|
|
|
|
}
|
|
|
|
return
|
2020-03-18 19:19:29 -04:00
|
|
|
}
|
2022-07-05 17:45:49 -04:00
|
|
|
|
|
|
|
//msgp:encode ignore currentScannerCycle
|
|
|
|
//msgp:decode ignore currentScannerCycle
|
|
|
|
|
|
|
|
type currentScannerCycle struct {
|
|
|
|
current uint64
|
|
|
|
next uint64
|
|
|
|
started time.Time
|
|
|
|
cycleCompleted []time.Time
|
|
|
|
}
|
|
|
|
|
|
|
|
// clone returns a clone.
|
|
|
|
func (z currentScannerCycle) clone() currentScannerCycle {
|
|
|
|
z.cycleCompleted = append(make([]time.Time, 0, len(z.cycleCompleted)), z.cycleCompleted...)
|
|
|
|
return z
|
|
|
|
}
|