stmtsummary: add persistence implementation

bindinfo/BUILD.bazel

@@ -34,7 +34,6 @@ go_library(
         "//util/memory",
         "//util/parser",
         "//util/sqlexec",
-        "//util/stmtsummary",
         "//util/table-filter",
         "//util/timeutil",
         "@org_golang_x_exp//maps",

bindinfo/handle.go

@@ -41,7 +41,6 @@ import (
 	"github.com/pingcap/tidb/util/logutil"
 	utilparser "github.com/pingcap/tidb/util/parser"
 	"github.com/pingcap/tidb/util/sqlexec"
-	"github.com/pingcap/tidb/util/stmtsummary"
 	tablefilter "github.com/pingcap/tidb/util/table-filter"
 	"github.com/pingcap/tidb/util/timeutil"
 	"go.uber.org/zap"
@@ -892,7 +891,7 @@ func (h *BindHandle) CaptureBaselines() {
 	parser4Capture := parser.New()
 	captureFilter := h.extractCaptureFilterFromStorage()
 	emptyCaptureFilter := captureFilter.isEmpty()
-	bindableStmts := stmtsummary.StmtSummaryByDigestMap.GetMoreThanCntBindableStmt(captureFilter.frequency)
+	bindableStmts := h.sctx.GetSessionVars().StmtSummary.GetBindableStmts(captureFilter.frequency)
 	for _, bindableStmt := range bindableStmts {
 		stmt, err := parser4Capture.ParseOneStmt(bindableStmt.Query, bindableStmt.Charset, bindableStmt.Collation)
 		if err != nil {

config/config.go

@@ -502,6 +502,20 @@ type Instance struct {
 	DDLSlowOprThreshold uint32 `toml:"ddl_slow_threshold" json:"ddl_slow_threshold"`
 	// ExpensiveQueryTimeThreshold indicates the time threshold of expensive query.
 	ExpensiveQueryTimeThreshold uint64 `toml:"tidb_expensive_query_time_threshold" json:"tidb_expensive_query_time_threshold"`
+	// StmtSummaryEnablePersistent indicates whether to enable file persistence for stmtsummary.
+	StmtSummaryEnablePersistent bool `toml:"tidb_stmt_summary_enable_persistent" json:"tidb_stmt_summary_enable_persistent"`
+	// StmtSummaryFilename indicates the file name written by stmtsummary
+	// when StmtSummaryEnablePersistent is true.
+	StmtSummaryFilename string `toml:"tidb_stmt_summary_filename" json:"tidb_stmt_summary_filename"`
+	// StmtSummaryFileMaxDays indicates how many days the files written by
+	// stmtsummary will be kept when StmtSummaryEnablePersistent is true.
+	StmtSummaryFileMaxDays int `toml:"tidb_stmt_summary_file_max_days" json:"tidb_stmt_summary_file_max_days"`
+	// StmtSummaryFileMaxSize indicates the maximum size (in mb) of a single file
+	// written by stmtsummary when StmtSummaryEnablePersistent is true.
+	StmtSummaryFileMaxSize int `toml:"tidb_stmt_summary_file_max_size" json:"tidb_stmt_summary_file_max_size"`
+	// StmtSummaryFileMaxBackups indicates the maximum number of files written
+	// by stmtsummary when StmtSummaryEnablePersistent is true.
+	StmtSummaryFileMaxBackups int `toml:"tidb_stmt_summary_file_max_backups" json:"tidb_stmt_summary_file_max_backups"`
 
 	// These variables exist in both 'instance' section and another place.
 	// The configuration in 'instance' section takes precedence.
@@ -901,6 +915,11 @@ var defaultConf = Config{
 		EnablePProfSQLCPU:           false,
 		DDLSlowOprThreshold:         DefDDLSlowOprThreshold,
 		ExpensiveQueryTimeThreshold: DefExpensiveQueryTimeThreshold,
+		StmtSummaryEnablePersistent: false,
+		StmtSummaryFilename:         "tidb-statements.log",
+		StmtSummaryFileMaxDays:      3,
+		StmtSummaryFileMaxSize:      64,
+		StmtSummaryFileMaxBackups:   0,
 		EnableSlowLog:               *NewAtomicBool(logutil.DefaultTiDBEnableSlowLog),
 		SlowThreshold:               logutil.DefaultSlowThreshold,
 		RecordPlanInSlowLog:         logutil.DefaultRecordPlanInSlowLog,

executor/BUILD.bazel

@@ -85,6 +85,7 @@ go_library(
         "slow_query.go",
         "sort.go",
         "split.go",
+        "stmtsummary.go",
         "table_reader.go",
         "trace.go",
         "union_scan.go",
@@ -194,6 +195,7 @@ go_library(
         "//util/size",
         "//util/sqlexec",
         "//util/stmtsummary",
+        "//util/stmtsummary/v2:stmtsummary",
         "//util/stringutil",
         "//util/table-filter",
         "//util/timeutil",
@@ -326,6 +328,7 @@ go_test(
         "split_test.go",
         "stale_txn_test.go",
         "statement_context_test.go",
+        "stmtsummary_test.go",
         "table_readers_required_rows_test.go",
         "temporary_table_test.go",
         "tikv_regions_peers_table_test.go",
@@ -419,6 +422,7 @@ go_test(
         "//util/rowcodec",
         "//util/set",
         "//util/sqlexec",
+        "//util/stmtsummary/v2:stmtsummary",
         "//util/stringutil",
         "//util/tableutil",
         "//util/timeutil",

executor/adapter.go

@@ -1802,8 +1802,10 @@ func (a *ExecStmt) SummaryStmt(succ bool) {
 		userString = sessVars.User.Username
 	}
 
+	enabled := sessVars.StmtSummary.Enabled()
+	enabledInternalQuery := sessVars.StmtSummary.EnabledInternal()
 	// Internal SQLs must also be recorded to keep the consistency of `PrevStmt` and `PrevStmtDigest`.
-	if !stmtsummary.StmtSummaryByDigestMap.Enabled() || ((sessVars.InRestrictedSQL || len(userString) == 0) && !stmtsummary.StmtSummaryByDigestMap.EnabledInternal()) {
+	if !enabled || ((sessVars.InRestrictedSQL || len(userString) == 0) && !enabledInternalQuery) {
 		sessVars.SetPrevStmtDigest("")
 		return
 	}
@@ -1920,7 +1922,7 @@ func (a *ExecStmt) SummaryStmt(succ bool) {
 	if a.retryCount > 0 {
 		stmtExecInfo.ExecRetryTime = costTime - sessVars.DurationParse - sessVars.DurationCompile - time.Since(a.retryStartTime)
 	}
-	stmtsummary.StmtSummaryByDigestMap.AddStatement(stmtExecInfo)
+	sessVars.StmtSummary.Add(stmtExecInfo)
 }
 
 // GetTextToLog return the query text to log.

executor/builder.go

@@ -1937,8 +1937,6 @@ func (b *executorBuilder) buildMemTable(v *plannercore.PhysicalMemTable) Executo
 			strings.ToLower(infoschema.TableTiFlashReplica),
 			strings.ToLower(infoschema.TableTiDBServersInfo),
 			strings.ToLower(infoschema.TableTiKVStoreStatus),
-			strings.ToLower(infoschema.TableStatementsSummaryEvicted),
-			strings.ToLower(infoschema.ClusterTableStatementsSummaryEvicted),
 			strings.ToLower(infoschema.TableClientErrorsSummaryGlobal),
 			strings.ToLower(infoschema.TableClientErrorsSummaryByUser),
 			strings.ToLower(infoschema.TableClientErrorsSummaryByHost),
@@ -1993,16 +1991,18 @@ func (b *executorBuilder) buildMemTable(v *plannercore.PhysicalMemTable) Executo
 			}
 		case strings.ToLower(infoschema.TableStatementsSummary),
 			strings.ToLower(infoschema.TableStatementsSummaryHistory),
+			strings.ToLower(infoschema.TableStatementsSummaryEvicted),
+			strings.ToLower(infoschema.ClusterTableStatementsSummary),
 			strings.ToLower(infoschema.ClusterTableStatementsSummaryHistory),
-			strings.ToLower(infoschema.ClusterTableStatementsSummary):
+			strings.ToLower(infoschema.ClusterTableStatementsSummaryEvicted):
+			var extractor *plannercore.StatementsSummaryExtractor
+			if v.Extractor != nil {
+				extractor = v.Extractor.(*plannercore.StatementsSummaryExtractor)
+			}
 			return &MemTableReaderExec{
 				baseExecutor: newBaseExecutor(b.ctx, v.Schema(), v.ID()),
 				table:        v.Table,
-				retriever: &stmtSummaryTableRetriever{
-					table:     v.Table,
-					columns:   v.Columns,
-					extractor: v.Extractor.(*plannercore.StatementsSummaryExtractor),
-				},
+				retriever:    buildStmtSummaryRetriever(b.ctx, v.Table, v.Columns, extractor),
 			}
 		case strings.ToLower(infoschema.TableColumns):
 			return &MemTableReaderExec{
@@ -2016,7 +2016,6 @@ func (b *executorBuilder) buildMemTable(v *plannercore.PhysicalMemTable) Executo
 					viewOutputNamesMap: make(map[int64]types.NameSlice),
 				},
 			}
-
 		case strings.ToLower(infoschema.TableSlowQuery), strings.ToLower(infoschema.ClusterTableSlowLog):
 			memTracker := memory.NewTracker(v.ID(), -1)
 			memTracker.AttachTo(b.ctx.GetSessionVars().StmtCtx.MemTracker)

executor/infoschema_reader.go

@@ -70,7 +70,6 @@ import (
 	"github.com/pingcap/tidb/util/servermemorylimit"
 	"github.com/pingcap/tidb/util/set"
 	"github.com/pingcap/tidb/util/sqlexec"
-	"github.com/pingcap/tidb/util/stmtsummary"
 	"github.com/pingcap/tidb/util/stringutil"
 	"github.com/tikv/client-go/v2/txnkv/txnlock"
 	"go.uber.org/zap"
@@ -158,9 +157,6 @@ func (e *memtableRetriever) retrieve(ctx context.Context, sctx sessionctx.Contex
 			e.dataForTableTiFlashReplica(sctx, dbs)
 		case infoschema.TableTiKVStoreStatus:
 			err = e.dataForTiKVStoreStatus(sctx)
-		case infoschema.TableStatementsSummaryEvicted,
-			infoschema.ClusterTableStatementsSummaryEvicted:
-			err = e.setDataForStatementsSummaryEvicted(sctx)
 		case infoschema.TableClientErrorsSummaryGlobal,
 			infoschema.TableClientErrorsSummaryByUser,
 			infoschema.TableClientErrorsSummaryByHost:
@@ -2306,22 +2302,6 @@ func (e *memtableRetriever) dataForTableTiFlashReplica(ctx sessionctx.Context, s
 	e.rows = rows
 }
 
-func (e *memtableRetriever) setDataForStatementsSummaryEvicted(ctx sessionctx.Context) error {
-	if !hasPriv(ctx, mysql.ProcessPriv) {
-		return plannercore.ErrSpecificAccessDenied.GenWithStackByArgs("PROCESS")
-	}
-	e.rows = stmtsummary.StmtSummaryByDigestMap.ToEvictedCountDatum()
-	switch e.table.Name.O {
-	case infoschema.ClusterTableStatementsSummaryEvicted:
-		rows, err := infoschema.AppendHostInfoToRows(ctx, e.rows)
-		if err != nil {
-			return err
-		}
-		e.rows = rows
-	}
-	return nil
-}
-
 func (e *memtableRetriever) setDataForClientErrorsSummary(ctx sessionctx.Context, tableName string) error {
 	// Seeing client errors should require the PROCESS privilege, with the exception of errors for your own user.
 	// This is similar to information_schema.processlist, which is the closest comparison.
@@ -2476,50 +2456,6 @@ func (e *memtableRetriever) setDataForClusterMemoryUsageOpsHistory(ctx sessionct
 	return nil
 }
 
-type stmtSummaryTableRetriever struct {
-	dummyCloser
-	table     *model.TableInfo
-	columns   []*model.ColumnInfo
-	retrieved bool
-	extractor *plannercore.StatementsSummaryExtractor
-}
-
-// retrieve implements the infoschemaRetriever interface
-func (e *stmtSummaryTableRetriever) retrieve(ctx context.Context, sctx sessionctx.Context) ([][]types.Datum, error) {
-	if e.extractor.SkipRequest || e.retrieved {
-		return nil, nil
-	}
-	e.retrieved = true
-
-	var err error
-	var instanceAddr string
-	switch e.table.Name.O {
-	case infoschema.ClusterTableStatementsSummary,
-		infoschema.ClusterTableStatementsSummaryHistory:
-		instanceAddr, err = infoschema.GetInstanceAddr(sctx)
-		if err != nil {
-			return nil, err
-		}
-	}
-	user := sctx.GetSessionVars().User
-	reader := stmtsummary.NewStmtSummaryReader(user, hasPriv(sctx, mysql.ProcessPriv), e.columns, instanceAddr, sctx.GetSessionVars().StmtCtx.TimeZone)
-	if e.extractor.Enable {
-		checker := stmtsummary.NewStmtSummaryChecker(e.extractor.Digests)
-		reader.SetChecker(checker)
-	}
-	var rows [][]types.Datum
-	switch e.table.Name.O {
-	case infoschema.TableStatementsSummary,
-		infoschema.ClusterTableStatementsSummary:
-		rows = reader.GetStmtSummaryCurrentRows()
-	case infoschema.TableStatementsSummaryHistory,
-		infoschema.ClusterTableStatementsSummaryHistory:
-		rows = reader.GetStmtSummaryHistoryRows()
-	}
-
-	return rows, nil
-}
-
 // tidbTrxTableRetriever is the memtable retriever for the TIDB_TRX and CLUSTER_TIDB_TRX table.
 type tidbTrxTableRetriever struct {
 	dummyCloser
@@ -3018,9 +2954,6 @@ func (e *hugeMemTableRetriever) retrieve(ctx context.Context, sctx sessionctx.Co
 }
 
 func adjustColumns(input [][]types.Datum, outColumns []*model.ColumnInfo, table *model.TableInfo) [][]types.Datum {
-	if table.Name.O == infoschema.TableStatementsSummary {
-		return input
-	}
 	if len(outColumns) == len(table.Columns) {
 		return input
 	}