analyze: analyze table after restore

cmd/backup.go

@@ -5,7 +5,6 @@ package cmd
 import (
 	"github.com/pingcap/log"
 	"github.com/pingcap/tidb/ddl"
-	"github.com/pingcap/tidb/session"
 	"github.com/spf13/cobra"
 	"go.uber.org/zap"
 
@@ -56,8 +55,6 @@ func NewBackupCommand() *cobra.Command {
 
 			// Do not run ddl worker in BR.
 			ddl.RunWorker = false
-			// Do not run stat worker in BR.
-			session.DisableStats4Test()
 
 			summary.SetUnit(summary.BackupUnit)
 			return nil

cmd/restore.go

@@ -4,7 +4,6 @@ package cmd
 
 import (
 	"github.com/pingcap/log"
-	"github.com/pingcap/tidb/session"
 	"github.com/spf13/cobra"
 	"go.uber.org/zap"
 
@@ -82,9 +81,6 @@ func NewRestoreCommand() *cobra.Command {
 			utils.LogBRInfo()
 			task.LogArguments(c)
 
-			// Do not run stat worker in BR.
-			session.DisableStats4Test()
-
 			summary.SetUnit(summary.RestoreUnit)
 			return nil
 		},

go.mod1

@@ -16,10 +16,10 @@ require (
 	github.com/pingcap/check v0.0.0-20200212061837-5e12011dc712
 	github.com/pingcap/errors v0.11.5-0.20201029093017-5a7df2af2ac7
 	github.com/pingcap/failpoint v0.0.0-20200702092429-9f69995143ce
-	github.com/pingcap/kvproto v0.0.0-20201113092725-08f2872278eb
+	github.com/pingcap/kvproto v0.0.0-20201120081251-756b1447ba12
 	github.com/pingcap/log v0.0.0-20201112100606-8f1e84a3abc8
-	github.com/pingcap/parser v0.0.0-20201112105242-773b8b74f44e
-	github.com/pingcap/tidb v1.1.0-beta.0.20201119091753-38bbb0dd2197
+	github.com/pingcap/parser v0.0.0-20201123080035-8f4c6ab94e11
+	github.com/pingcap/tidb v1.1.0-beta.0.20201125030655-07ff41d0c43e
 	github.com/pingcap/tidb-tools v4.0.5-0.20200820092506-34ea90c93237+incompatible
 	github.com/pingcap/tipb v0.0.0-20201026044621-45e60c77588f
 	github.com/prometheus/client_golang v1.5.1

go.sum1

@@ -753,8 +753,9 @@ github.com/pingcap/kvproto v0.0.0-20200411081810-b85805c9476c/go.mod h1:IOdRDPLy
 github.com/pingcap/kvproto v0.0.0-20200417092353-efbe03bcffbd/go.mod h1:IOdRDPLyda8GX2hE/jO7gqaCV/PNFh8BZQCQZXfIOqI=
 github.com/pingcap/kvproto v0.0.0-20200420075417-e0c6e8842f22/go.mod h1:IOdRDPLyda8GX2hE/jO7gqaCV/PNFh8BZQCQZXfIOqI=
 github.com/pingcap/kvproto v0.0.0-20200810113304-6157337686b1/go.mod h1:IOdRDPLyda8GX2hE/jO7gqaCV/PNFh8BZQCQZXfIOqI=
-github.com/pingcap/kvproto v0.0.0-20201113092725-08f2872278eb h1:K3r4KjVQeD4nLnfj44ibdLIXnUh58aQpkgVNWuBO9z0=
 github.com/pingcap/kvproto v0.0.0-20201113092725-08f2872278eb/go.mod h1:IOdRDPLyda8GX2hE/jO7gqaCV/PNFh8BZQCQZXfIOqI=
+github.com/pingcap/kvproto v0.0.0-20201120081251-756b1447ba12 h1:f33y/pngBI525jqytoSZevpmmq43XiIHoeElx3BppNQ=
+github.com/pingcap/kvproto v0.0.0-20201120081251-756b1447ba12/go.mod h1:IOdRDPLyda8GX2hE/jO7gqaCV/PNFh8BZQCQZXfIOqI=
 github.com/pingcap/log v0.0.0-20191012051959-b742a5d432e9 h1:AJD9pZYm72vMgPcQDww9rkZ1DnWfl0pXV3BOWlkYIjA=
 github.com/pingcap/log v0.0.0-20191012051959-b742a5d432e9/go.mod h1:4rbK1p9ILyIfb6hU7OG2CiWSqMXnp3JMbiaVJ6mvoY8=
 github.com/pingcap/log v0.0.0-20200117041106-d28c14d3b1cd h1:CV3VsP3Z02MVtdpTMfEgRJ4T9NGgGTxdHpJerent7rM=
@@ -765,8 +766,8 @@ github.com/pingcap/log v0.0.0-20201112100606-8f1e84a3abc8 h1:M+DNpOu/I3uDmwee6vc
 github.com/pingcap/log v0.0.0-20201112100606-8f1e84a3abc8/go.mod h1:4rbK1p9ILyIfb6hU7OG2CiWSqMXnp3JMbiaVJ6mvoY8=
 github.com/pingcap/parser v0.0.0-20190506092653-e336082eb825/go.mod h1:1FNvfp9+J0wvc4kl8eGNh7Rqrxveg15jJoWo/a0uHwA=
 github.com/pingcap/parser v0.0.0-20200422082501-7329d80eaf2c/go.mod h1:9v0Edh8IbgjGYW2ArJr19E+bvL8zKahsFp+ixWeId+4=
-github.com/pingcap/parser v0.0.0-20201112105242-773b8b74f44e h1:TCSZlqvNP2Ipt0xUwZ5Z6ehmKo1dlk5KCU1SOLXUYL4=
-github.com/pingcap/parser v0.0.0-20201112105242-773b8b74f44e/go.mod h1:GbEr2PgY72/4XqPZzmzstlOU/+il/wrjeTNFs6ihsSE=
+github.com/pingcap/parser v0.0.0-20201123080035-8f4c6ab94e11 h1:CpYV3/Mp42Szvt8ne6g2ojLtBmsUq6o86JazeAj4pIo=
+github.com/pingcap/parser v0.0.0-20201123080035-8f4c6ab94e11/go.mod h1:GbEr2PgY72/4XqPZzmzstlOU/+il/wrjeTNFs6ihsSE=
 github.com/pingcap/pd v2.1.5+incompatible/go.mod h1:nD3+EoYes4+aNNODO99ES59V83MZSI+dFbhyr667a0E=
 github.com/pingcap/pd/v4 v4.0.0-rc.1.0.20200422143320-428acd53eba2/go.mod h1:s+utZtXDznOiL24VK0qGmtoHjjXNsscJx3m1n8cC56s=
 github.com/pingcap/sysutil v0.0.0-20200206130906-2bfa6dc40bcd/go.mod h1:EB/852NMQ+aRKioCpToQ94Wl7fktV+FNnxf3CX/TTXI=
@@ -776,8 +777,8 @@ github.com/pingcap/sysutil v0.0.0-20200715082929-4c47bcac246a/go.mod h1:EB/852NM
 github.com/pingcap/sysutil v0.0.0-20201021075216-f93ced2829e2 h1:b2G/eqDeywtdJF3w9nIUdqMmXChsmpLvf4FzUxJ9Vmk=
 github.com/pingcap/sysutil v0.0.0-20201021075216-f93ced2829e2/go.mod h1:EB/852NMQ+aRKioCpToQ94Wl7fktV+FNnxf3CX/TTXI=
 github.com/pingcap/tidb v1.1.0-beta.0.20200423094549-0ad7ce6cdae6/go.mod h1:KJXj2xHYfl1x4zcusC2JEANzVci+ietFOMh/CAmrYdw=
-github.com/pingcap/tidb v1.1.0-beta.0.20201119091753-38bbb0dd2197 h1:J6vZT3DALFLPmJSKYi5kFkLvayNqq7Oo6ongGQSpTc8=
-github.com/pingcap/tidb v1.1.0-beta.0.20201119091753-38bbb0dd2197/go.mod h1:yAq/uipWkDY0PDXdRCgs0uQs8lV9isdWcIdRYWtk+mU=
+github.com/pingcap/tidb v1.1.0-beta.0.20201125030655-07ff41d0c43e h1:EHjW6Jjcy7rGUqbOiGfoyzhWVxry38zC3xFQFUw27EU=
+github.com/pingcap/tidb v1.1.0-beta.0.20201125030655-07ff41d0c43e/go.mod h1:nmENlVbrMlaj+Tw0nPI/aQThVFficGJ/citl9OA/B3A=
 github.com/pingcap/tidb-insight v0.3.1/go.mod h1:HZ6cGBYxViftTKLbl/V4moRol79bifZzI9ImGavJms0=
 github.com/pingcap/tidb-lightning v4.0.9-0.20201106041742-a1ac97827a27+incompatible h1:y6cwuJJo8tLeHjgjODpwCSqNAv1g+9WUtcsFOHu/7ag=
 github.com/pingcap/tidb-lightning v4.0.9-0.20201106041742-a1ac97827a27+incompatible/go.mod h1:UrN2l06neiEfkass87Tjvv5JFIMsOMb2DT7P7s+fwMc=

pkg/backup/client.go

@@ -250,12 +250,15 @@ func BuildBackupRangeAndSchema(
 	storage kv.Storage,
 	tableFilter filter.Filter,
 	backupTS uint64,
+	ignoreStats bool,
 ) ([]rtree.Range, *Schemas, error) {
 	info, err := dom.GetSnapshotInfoSchema(backupTS)
 	if err != nil {
 		return nil, nil, errors.Trace(err)
 	}
 
+	h := dom.StatsHandle()
+
 	ranges := make([]rtree.Range, 0)
 	backupSchemas := newBackupSchemas()
 	for _, dbInfo := range info.AllSchemas() {
@@ -332,9 +335,22 @@ func BuildBackupRangeAndSchema(
 				return nil, nil, errors.Trace(err)
 			}
 
+			var stats []byte
+			if !ignoreStats {
+				jsonTable, err := h.DumpStatsToJSON(dbInfo.Name.String(), tableInfo, nil)
+				if err != nil {
+					return nil, nil, errors.Trace(err)
+				}
+				stats, err = json.Marshal(jsonTable)
+				if err != nil {
+					return nil, nil, errors.Trace(err)
+				}
+			}
+
 			schema := kvproto.Schema{
 				Db:    dbData,
 				Table: tableData,
+				Stats: stats,
 			}
 			backupSchemas.pushPending(schema, dbInfo.Name.L, tableInfo.Name.L)
 

pkg/backup/schema_test.go

@@ -61,23 +61,23 @@ func (s *testBackupSchemaSuite) TestBuildBackupRangeAndSchema(c *C) {
 	testFilter, err := filter.Parse([]string{"test.t1"})
 	c.Assert(err, IsNil)
 	_, backupSchemas, err := backup.BuildBackupRangeAndSchema(
-		s.mock.Domain, s.mock.Storage, testFilter, math.MaxUint64)
+		s.mock.Domain, s.mock.Storage, testFilter, math.MaxUint64, false)
 	c.Assert(err, IsNil)
 	c.Assert(backupSchemas, IsNil)
 
 	// Database is not exist.
 	fooFilter, err := filter.Parse([]string{"foo.t1"})
 	c.Assert(err, IsNil)
 	_, backupSchemas, err = backup.BuildBackupRangeAndSchema(
-		s.mock.Domain, s.mock.Storage, fooFilter, math.MaxUint64)
+		s.mock.Domain, s.mock.Storage, fooFilter, math.MaxUint64, false)
 	c.Assert(err, IsNil)
 	c.Assert(backupSchemas, IsNil)
 
 	// Empty database.
 	noFilter, err := filter.Parse([]string{"*.*"})
 	c.Assert(err, IsNil)
 	_, backupSchemas, err = backup.BuildBackupRangeAndSchema(
-		s.mock.Domain, s.mock.Storage, noFilter, math.MaxUint64)
+		s.mock.Domain, s.mock.Storage, noFilter, math.MaxUint64, false)
 	c.Assert(err, IsNil)
 	c.Assert(backupSchemas, IsNil)
 
@@ -87,7 +87,7 @@ func (s *testBackupSchemaSuite) TestBuildBackupRangeAndSchema(c *C) {
 	tk.MustExec("insert into t1 values (10);")
 
 	_, backupSchemas, err = backup.BuildBackupRangeAndSchema(
-		s.mock.Domain, s.mock.Storage, testFilter, math.MaxUint64)
+		s.mock.Domain, s.mock.Storage, testFilter, math.MaxUint64, false)
 	c.Assert(err, IsNil)
 	c.Assert(backupSchemas.Len(), Equals, 1)
 	updateCh := new(simpleProgress)
@@ -107,7 +107,7 @@ func (s *testBackupSchemaSuite) TestBuildBackupRangeAndSchema(c *C) {
 	tk.MustExec("insert into t2 values (11);")
 
 	_, backupSchemas, err = backup.BuildBackupRangeAndSchema(
-		s.mock.Domain, s.mock.Storage, noFilter, math.MaxUint64)
+		s.mock.Domain, s.mock.Storage, noFilter, math.MaxUint64, false)
 	c.Assert(err, IsNil)
 	c.Assert(backupSchemas.Len(), Equals, 2)
 	updateCh.reset()

pkg/conn/conn.go

@@ -10,8 +10,6 @@ import (
 	"time"
 
 	"github.com/pingcap/errors"
-	"github.com/pingcap/kvproto/pkg/backup"
-	"github.com/pingcap/kvproto/pkg/metapb"
 	"github.com/pingcap/log"
 	"github.com/pingcap/tidb/domain"
 	"github.com/pingcap/tidb/store/tikv"
@@ -22,6 +20,9 @@ import (
 	"google.golang.org/grpc/credentials"
 	"google.golang.org/grpc/keepalive"
 
+	"github.com/pingcap/kvproto/pkg/backup"
+	"github.com/pingcap/kvproto/pkg/metapb"
+
 	berrors "github.com/pingcap/br/pkg/errors"
 	"github.com/pingcap/br/pkg/glue"
 	"github.com/pingcap/br/pkg/logutil"

pkg/gluetidb/glue.go

@@ -49,7 +49,20 @@ type tidbSession struct {
 
 // GetDomain implements glue.Glue.
 func (Glue) GetDomain(store kv.Storage) (*domain.Domain, error) {
-	return session.GetDomain(store)
+	se, err := session.CreateSession(store)
+	if err != nil {
+		return nil, err
+	}
+	dom, err := session.GetDomain(store)
+	if err != nil {
+		return nil, err
+	}
+	// create stats handler for backup and restore.
+	err = dom.UpdateTableStatsLoop(se)
+	if err != nil {
+		return nil, err
+	}
+	return dom, nil
 }
 
 // CreateSession implements glue.Glue.

pkg/restore/batcher_test.go

@@ -68,52 +68,62 @@ func newDrySender() *drySender {
 	}
 }
 
-type recordCurrentTableManager map[int64]bool
+type recordCurrentTableManager struct {
+	lock sync.Mutex
+	m    map[int64]bool
+}
 
-func (manager recordCurrentTableManager) Close(ctx context.Context) {
-	if len(manager) > 0 {
+func (manager *recordCurrentTableManager) Close(ctx context.Context) {
+	if len(manager.m) > 0 {
 		log.Panic("When closing, there are still some tables doesn't be sent",
-			zap.Any("tables", manager))
+			zap.Any("tables", manager.m))
 	}
 }
 
-func newMockManager() recordCurrentTableManager {
-	return make(recordCurrentTableManager)
+func newMockManager() *recordCurrentTableManager {
+	return &recordCurrentTableManager{
+		m: make(map[int64]bool),
+	}
 }
 
-func (manager recordCurrentTableManager) Enter(_ context.Context, tables []restore.CreatedTable) error {
+func (manager *recordCurrentTableManager) Enter(_ context.Context, tables []restore.CreatedTable) error {
 	for _, t := range tables {
 		log.Info("entering", zap.Int64("table ID", t.Table.ID))
-		manager[t.Table.ID] = true
+		manager.lock.Lock()
+		manager.m[t.Table.ID] = true
+		manager.lock.Unlock()
 	}
 	return nil
 }
 
-func (manager recordCurrentTableManager) Leave(_ context.Context, tables []restore.CreatedTable) error {
+func (manager *recordCurrentTableManager) Leave(_ context.Context, tables []restore.CreatedTable) error {
 	for _, t := range tables {
-		if !manager[t.Table.ID] {
+		manager.lock.Lock()
+		if !manager.m[t.Table.ID] {
+			manager.lock.Unlock()
 			return errors.Errorf("Table %d is removed before added", t.Table.ID)
 		}
 		log.Info("leaving", zap.Int64("table ID", t.Table.ID))
-		delete(manager, t.Table.ID)
+		delete(manager.m, t.Table.ID)
+		manager.lock.Unlock()
 	}
 	return nil
 }
 
-func (manager recordCurrentTableManager) Has(tables ...restore.TableWithRange) bool {
+func (manager *recordCurrentTableManager) Has(tables ...restore.TableWithRange) bool {
 	ids := make([]int64, 0, len(tables))
-	currentIDs := make([]int64, 0, len(manager))
+	currentIDs := make([]int64, 0, len(manager.m))
 	for _, t := range tables {
 		ids = append(ids, t.Table.ID)
 	}
-	for id, contains := range manager {
+	for id, contains := range manager.m {
 		if contains {
 			currentIDs = append(currentIDs, id)
 		}
 	}
 	log.Info("testing", zap.Int64s("should has ID", ids), zap.Int64s("has ID", currentIDs))
 	for _, i := range ids {
-		if !manager[i] {
+		if !manager.m[i] {
 			return false
 		}
 	}

pkg/restore/client.go

@@ -21,6 +21,7 @@ import (
 	"github.com/pingcap/parser/model"
 	"github.com/pingcap/tidb/domain"
 	"github.com/pingcap/tidb/kv"
+	"github.com/pingcap/tidb/statistics/handle"
 	"github.com/pingcap/tidb/store/tikv/oracle"
 	"github.com/pingcap/tidb/tablecodec"
 	"github.com/pingcap/tidb/util/codec"
@@ -80,6 +81,12 @@ type Client struct {
 	backend            *backup.StorageBackend
 	switchModeInterval time.Duration
 	switchCh           chan struct{}
+
+	// statHandler and dom are used for analyze table after restore.
+	// it will backup stats with #dump.DumpStatsToJSON
+	// and restore stats with #dump.LoadStatsFromJSON
+	statsHandler *handle.Handle
+	dom          *domain.Domain
 }
 
 // NewRestoreClient returns a new RestoreClient.
@@ -94,6 +101,16 @@ func NewRestoreClient(
 	if err != nil {
 		return nil, errors.Trace(err)
 	}
+	dom, err := g.GetDomain(store)
+	if err != nil {
+		return nil, errors.Trace(err)
+	}
+
+	var statsHandle *handle.Handle
+	// tikv.Glue will return nil, tidb.Glue will return available domain
+	if dom != nil {
+		statsHandle = dom.StatsHandle()
+	}
 
 	return &Client{
 		pdClient:      pdClient,
@@ -102,6 +119,8 @@ func NewRestoreClient(
 		tlsConf:       tlsConf,
 		keepaliveConf: keepaliveConf,
 		switchCh:      make(chan struct{}),
+		dom:           dom,
+		statsHandler:  statsHandle,
 	}, nil
 }
 
@@ -775,6 +794,17 @@ func (rc *Client) execChecksum(ctx context.Context, tbl CreatedTable, kvClient k
 		)
 		return errors.Annotate(berrors.ErrRestoreChecksumMismatch, "failed to validate checksum")
 	}
+	if table.Stats != nil {
+		log.Info("start loads analyze after validate checksum",
+			zap.Stringer("db name", tbl.OldTable.DB.Name),
+			zap.Stringer("name", tbl.OldTable.Info.Name),
+			zap.Int64("old id", tbl.OldTable.Info.ID),
+			zap.Int64("new id", tbl.Table.ID),
+		)
+		if err := rc.statsHandler.LoadStatsFromJSON(rc.dom.InfoSchema(), table.Stats); err != nil {
+			log.Error("analyze table failed, ", zap.Any("table", table.Stats), zap.Error(err))
+		}
+	}
 	return nil
 }
 

pkg/task/backup.go

@@ -35,6 +35,7 @@ const (
 	flagCompressionType  = "compression"
 	flagCompressionLevel = "compression-level"
 	flagRemoveSchedulers = "remove-schedulers"
+	flagIgnoreStats      = "ignore-stats"
 
 	flagGCTTL = "gcttl"
 
@@ -57,6 +58,7 @@ type BackupConfig struct {
 	LastBackupTS     uint64        `json:"last-backup-ts" toml:"last-backup-ts"`
 	GCTTL            int64         `json:"gc-ttl" toml:"gc-ttl"`
 	RemoveSchedulers bool          `json:"remove-schedulers" toml:"remove-schedulers"`
+	IgnoreStats      bool          `json:"ignore-stats" toml:"ignore-stats"`
 	CompressionConfig
 }
 
@@ -80,6 +82,11 @@ func DefineBackupFlags(flags *pflag.FlagSet) {
 		"disable the balance, shuffle and region-merge schedulers in PD to speed up backup")
 	// This flag can impact the online cluster, so hide it in case of abuse.
 	_ = flags.MarkHidden(flagRemoveSchedulers)
+
+	flags.Bool(flagIgnoreStats, false,
+		"ignore backup stats, used for test")
+	// This flag is used for test. we should backup stats all the time.
+	_ = flags.MarkHidden(flagIgnoreStats)
 }
 
 // ParseFromFlags parses the backup-related flags from the flag set.
@@ -119,8 +126,11 @@ func (cfg *BackupConfig) ParseFromFlags(flags *pflag.FlagSet) error {
 	if err = cfg.Config.ParseFromFlags(flags); err != nil {
 		return errors.Trace(err)
 	}
-
 	cfg.RemoveSchedulers, err = flags.GetBool(flagRemoveSchedulers)
+	if err != nil {
+		return errors.Trace(err)
+	}
+	cfg.IgnoreStats, err = flags.GetBool(flagIgnoreStats)
 	return err
 }
 
@@ -245,7 +255,7 @@ func RunBackup(c context.Context, g glue.Glue, cmdName string, cfg *BackupConfig
 	}
 
 	ranges, backupSchemas, err := backup.BuildBackupRangeAndSchema(
-		mgr.GetDomain(), mgr.GetTiKV(), cfg.TableFilter, backupTS)
+		mgr.GetDomain(), mgr.GetTiKV(), cfg.TableFilter, backupTS, cfg.IgnoreStats)
 	if err != nil {
 		return err
 	}