*: support automatically create sql baselines

bindinfo/bind_test.go

@@ -100,8 +100,8 @@ func (s *testSuite) TearDownTest(c *C) {
 }
 
 func (s *testSuite) cleanBindingEnv(tk *testkit.TestKit) {
-	tk.MustExec("drop table if exists mysql.bind_info")
-	tk.MustExec(session.CreateBindInfoTable)
+	tk.MustExec("truncate table mysql.bind_info")
+	s.domain.BindHandle().Clear()
 }
 
 func (s *testSuite) TestBindParse(c *C) {
@@ -486,3 +486,26 @@ func (s *testSuite) TestPreparedStmt(c *C) {
 	tk.MustExec("execute stmt1")
 	c.Assert(len(tk.Se.GetSessionVars().StmtCtx.IndexNames), Equals, 0)
 }
+
+func (s *testSuite) TestCapturePlanBaseline(c *C) {
+	tk := testkit.NewTestKit(c, s.store)
+	s.cleanBindingEnv(tk)
+	tk.MustExec("set @@tidb_enable_stmt_summary = on")
+	tk.MustExec(" set @@tidb_capture_plan_baselines = on")
+	defer func() {
+		tk.MustExec("set @@tidb_enable_stmt_summary = off")
+		tk.MustExec(" set @@tidb_capture_plan_baselines = off")
+	}()
+	tk.MustExec("use test")
+	tk.MustExec("drop table if exists t")
+	tk.MustExec("create table t(a int)")
+	s.domain.BindHandle().CaptureBaselines(s.domain.InfoSchema())
+	tk.MustQuery("show global bindings").Check(testkit.Rows())
+	tk.MustExec("select * from t")
+	tk.MustExec("select * from t")
+	s.domain.BindHandle().CaptureBaselines(s.domain.InfoSchema())
+	rows := tk.MustQuery("show global bindings").Rows()
+	c.Assert(len(rows), Equals, 1)
+	c.Assert(rows[0][0], Equals, "select * from t")
+	c.Assert(rows[0][1], Equals, "select /*+ USE_INDEX(@`sel_1` `t` )*/ * from t")
+}

bindinfo/handle.go

@@ -16,20 +16,24 @@ package bindinfo
 import (
 	"context"
 	"fmt"
-	"go.uber.org/zap"
+	"strings"
 	"sync"
 	"sync/atomic"
 	"time"
 
 	"github.com/pingcap/parser"
+	"github.com/pingcap/parser/ast"
 	"github.com/pingcap/parser/mysql"
 	"github.com/pingcap/parser/terror"
+	"github.com/pingcap/tidb/infoschema"
 	"github.com/pingcap/tidb/metrics"
 	"github.com/pingcap/tidb/sessionctx"
 	"github.com/pingcap/tidb/store/tikv/oracle"
 	"github.com/pingcap/tidb/types"
 	"github.com/pingcap/tidb/util/logutil"
 	"github.com/pingcap/tidb/util/sqlexec"
+	"github.com/pingcap/tidb/util/stmtsummary"
+	"go.uber.org/zap"
 )
 
 // BindHandle is used to handle all global sql bind operations.
@@ -70,6 +74,8 @@ type BindHandle struct {
 	}
 
 	lastUpdateTime types.Time
+
+	parser4Baseline *parser.Parser
 }
 
 // Lease influences the duration of loading bind info and handling invalid bind.
@@ -86,6 +92,7 @@ func NewBindHandle(ctx sessionctx.Context) *BindHandle {
 	handle.sctx.Context = ctx
 	handle.bindInfo.Value.Store(make(cache, 32))
 	handle.bindInfo.parser = parser.New()
+	handle.parser4Baseline = parser.New()
 	handle.invalidBindRecordMap.Value.Store(make(map[string]*invalidBindRecordMap))
 	return handle
 }
@@ -444,3 +451,51 @@ func (h *BindHandle) logicalDeleteBindInfoSQL(normdOrigSQL, db string, updateTs
 		normdOrigSQL,
 		db)
 }
+
+// GenHintsFromSQL is used to generate hints from SQL.
+// It is used to avoid the circle dependence with planner package.
+var GenHintsFromSQL func(ctx context.Context, sctx sessionctx.Context, node ast.Node, is infoschema.InfoSchema) (string, error)
+
+// CaptureBaselines is used to automatically capture plan baselines.
+func (h *BindHandle) CaptureBaselines(is infoschema.InfoSchema) {
+	schemas, sqls := stmtsummary.StmtSummaryByDigestMap.GetMoreThanOnceSelect()
+	for i := range sqls {
+		stmt, err := h.parser4Baseline.ParseOneStmt(sqls[i], "", "")
+		if err != nil {
+			logutil.BgLogger().Debug("parse SQL failed", zap.String("SQL", sqls[i]), zap.Error(err))
+			continue
+		}
+		normalizedSQL, digiest := parser.NormalizeDigest(sqls[i])
+		if r := h.GetBindRecord(digiest, normalizedSQL, schemas[i]); r != nil && r.Status == Using {
+			continue
+		}
+		h.sctx.Lock()
+		err = h.sctx.RefreshTxnCtx(context.TODO())
+		var hints string
+		if err == nil {
+			h.sctx.GetSessionVars().CurrentDB = schemas[i]
+			hints, err = GenHintsFromSQL(context.TODO(), h.sctx.Context, stmt, is)
+		}
+		h.sctx.Unlock()
+		if err != nil {
+			logutil.BgLogger().Info("generate hints failed", zap.String("SQL", sqls[i]), zap.Error(err))
+			continue
+		}
+		// We can skip simple query like point get.
+		if hints == "" {
+			continue
+		}
+		bindsql := strings.Replace(normalizedSQL, "select", fmt.Sprintf("select /*+ %s*/", hints), 1)
+		err = h.AddBindRecord(&BindRecord{OriginalSQL: sqls[i], BindSQL: bindsql, Db: schemas[i], Status: Using})
+		if err != nil {
+			logutil.BgLogger().Info("capture baseline failed", zap.String("SQL", sqls[i]), zap.Error(err))
+		}
+	}
+}
+
+// Clear resets the bind handle. It is used for test.
+func (h *BindHandle) Clear() {
+	h.bindInfo.Store(make(cache))
+	h.invalidBindRecordMap.Store(make(map[string]*invalidBindRecordMap))
+	h.lastUpdateTime = types.ZeroTimestamp
+}

domain/domain.go

@@ -827,25 +827,34 @@ func (do *Domain) LoadBindInfoLoop(ctx sessionctx.Context) error {
 		return err
 	}
 
-	do.loadBindInfoLoop()
+	do.globalBindHandleWorkerLoop()
 	do.handleInvalidBindTaskLoop()
 	return nil
 }
 
-func (do *Domain) loadBindInfoLoop() {
+func (do *Domain) globalBindHandleWorkerLoop() {
 	do.wg.Add(1)
 	go func() {
 		defer do.wg.Done()
-		defer recoverInDomain("loadBindInfoLoop", false)
+		defer recoverInDomain("globalBindHandleWorkerLoop", false)
+		loadTicker := time.NewTicker(bindinfo.Lease)
+		defer loadTicker.Stop()
+		captureBaselineTicker := time.NewTicker(bindinfo.Lease)
+		defer captureBaselineTicker.Stop()
 		for {
 			select {
 			case <-do.exit:
 				return
-			case <-time.After(bindinfo.Lease):
-			}
-			err := do.bindHandle.Update(false)
-			if err != nil {
-				logutil.BgLogger().Error("update bindinfo failed", zap.Error(err))
+			case <-loadTicker.C:
+				err := do.bindHandle.Update(false)
+				if err != nil {
+					logutil.BgLogger().Error("update bindinfo failed", zap.Error(err))
+				}
+			case <-captureBaselineTicker.C:
+				if !variable.TiDBOptOn(variable.CapturePlanBaseline.GetVal()) {
+					continue
+				}
+				do.bindHandle.CaptureBaselines(do.InfoSchema())
 			}
 		}
 	}()

domain/global_vars_cache.go

@@ -44,7 +44,7 @@ func (gvc *GlobalVariableCache) Update(rows []chunk.Row, fields []*ast.ResultFie
 	gvc.fields = fields
 	gvc.Unlock()
 
-	checkEnableStmtSummary(rows, fields)
+	checkEnableServerGlobalVar(rows)
 }
 
 // Get gets the global variables from cache.
@@ -67,24 +67,22 @@ func (gvc *GlobalVariableCache) Disable() {
 	return
 }
 
-// checkEnableStmtSummary looks for TiDBEnableStmtSummary and notifies StmtSummary
-func checkEnableStmtSummary(rows []chunk.Row, fields []*ast.ResultField) {
+// checkEnableServerGlobalVar processes variables that acts in server and global level.
+func checkEnableServerGlobalVar(rows []chunk.Row) {
 	for _, row := range rows {
-		varName := row.GetString(0)
-		if varName == variable.TiDBEnableStmtSummary {
-			varVal := row.GetDatum(1, &fields[1].Column.FieldType)
-
+		switch row.GetString(0) {
+		case variable.TiDBEnableStmtSummary:
 			sVal := ""
-			if !varVal.IsNull() {
-				var err error
-				sVal, err = varVal.ToString()
-				if err != nil {
-					return
-				}
+			if !row.IsNull(1) {
+				sVal = row.GetString(1)
 			}
-
 			stmtsummary.StmtSummaryByDigestMap.SetEnabled(sVal, false)
-			break
+		case variable.TiDBCapturePlanBaseline:
+			sVal := ""
+			if !row.IsNull(1) {
+				sVal = row.GetString(1)
+			}
+			variable.CapturePlanBaseline.Set(sVal, false)
 		}
 	}
 }

executor/set.go

@@ -199,6 +199,8 @@ func (e *SetExecutor) setSysVariable(name string, v *expression.VarAssignment) e
 
 	if name == variable.TiDBEnableStmtSummary {
 		stmtsummary.StmtSummaryByDigestMap.SetEnabled(valStr, !v.IsGlobal)
+	} else if name == variable.TiDBCapturePlanBaseline {
+		variable.CapturePlanBaseline.Set(valStr, !v.IsGlobal)
 	}
 
 	return nil

planner/optimize.go

@@ -240,6 +240,20 @@ func OptimizeExecStmt(ctx context.Context, sctx sessionctx.Context,
 	return nil, err
 }
 
+// GenHintsFromSQL is used to generate hints from SQL and inject the hints into original SQL.
+func GenHintsFromSQL(ctx context.Context, sctx sessionctx.Context, node ast.Node, is infoschema.InfoSchema) (string, error) {
+	err := plannercore.Preprocess(sctx, node, is)
+	if err != nil {
+		return "", err
+	}
+	p, err := Optimize(ctx, sctx, node, is)
+	if err != nil {
+		return "", err
+	}
+	return plannercore.GenHintsFromPhysicalPlan(p), nil
+}
+
 func init() {
 	plannercore.OptimizeAstNode = Optimize
+	bindinfo.GenHintsFromSQL = GenHintsFromSQL
 }

session/session.go

@@ -1835,6 +1835,7 @@ var builtinGlobalVariable = []string{
 	variable.TiDBTxnMode,
 	variable.TiDBEnableStmtSummary,
 	variable.TiDBMaxDeltaSchemaCount,
+	variable.TiDBCapturePlanBaseline,
 }
 
 var (

sessionctx/variable/sysvar.go

@@ -726,6 +726,7 @@ var defaultSysVars = []*SysVar{
 	{ScopeSession, TiDBReplicaRead, "leader"},
 	{ScopeSession, TiDBAllowRemoveAutoInc, BoolToIntStr(DefTiDBAllowRemoveAutoInc)},
 	{ScopeGlobal | ScopeSession, TiDBEnableStmtSummary, "0"},
+	{ScopeGlobal | ScopeSession, TiDBCapturePlanBaseline, "0"},
 }
 
 // SynonymsSysVariables is synonyms of system variables.

sessionctx/variable/tidb_vars.go

@@ -322,6 +322,9 @@ const (
 
 	// TiDBEnableStmtSummary indicates whether the statement summary is enabled.
 	TiDBEnableStmtSummary = "tidb_enable_stmt_summary"
+
+	// TiDBCapturePlanBaseline indicates whether the capture of plan baselines is enabled.
+	TiDBCapturePlanBaseline = "tidb_capture_plan_baselines"
 )
 
 // Default TiDB system variable values.
@@ -416,4 +419,5 @@ var (
 	MaxOfMaxAllowedPacket          uint64 = 1073741824
 	ExpensiveQueryTimeThreshold    uint64 = DefTiDBExpensiveQueryTimeThreshold
 	MinExpensiveQueryTimeThreshold uint64 = 10 //10s
+	CapturePlanBaseline                   = serverGlobalVariable{globalVal: "0"}
 )

sessionctx/variable/varsutil.go

@@ -19,6 +19,7 @@ import (
 	"math"
 	"strconv"
 	"strings"
+	"sync"
 	"sync/atomic"
 	"time"
 
@@ -611,7 +612,7 @@ func ValidateSetSystemVar(vars *SessionVars, name string, value string) (string,
 			return "off", nil
 		}
 		return value, ErrWrongValueForVar.GenWithStackByArgs(name, value)
-	case TiDBEnableStmtSummary:
+	case TiDBEnableStmtSummary, TiDBCapturePlanBaseline:
 		switch {
 		case strings.EqualFold(value, "ON") || value == "1":
 			return "1", nil
@@ -737,3 +738,31 @@ func setAnalyzeTime(s *SessionVars, val string) (string, error) {
 	}
 	return t.Format(AnalyzeFullTimeFormat), nil
 }
+
+// serverGlobalVariable is used to handle variables that acts in server and global scope.
+type serverGlobalVariable struct {
+	sync.Mutex
+	serverVal string
+	globalVal string
+}
+
+// Set sets the value according to variable scope.
+func (v *serverGlobalVariable) Set(val string, isServer bool) {
+	v.Lock()
+	if isServer {
+		v.serverVal = val
+	} else {
+		v.globalVal = val
+	}
+	v.Unlock()
+}
+
+// GetVal gets the value.
+func (v *serverGlobalVariable) GetVal() string {
+	v.Lock()
+	defer v.Unlock()
+	if v.serverVal != "" {
+		return v.serverVal
+	}
+	return v.globalVal
+}

util/stmtsummary/statement_summary.go

@@ -230,6 +230,26 @@ func (ssMap *stmtSummaryByDigestMap) ToDatum() [][]types.Datum {
 	return rows
 }
 
+// GetMoreThanOnceSelect gets select SQLs that occurred more than once.
+func (ssMap *stmtSummaryByDigestMap) GetMoreThanOnceSelect() ([]string, []string) {
+	ssMap.Lock()
+	values := ssMap.summaryMap.Values()
+	ssMap.Unlock()
+
+	schemas := make([]string, 0, len(values))
+	sqls := make([]string, 0, len(values))
+	for _, value := range values {
+		summary := value.(*stmtSummaryByDigest)
+		summary.Lock()
+		if strings.HasPrefix(summary.normalizedSQL, "select") && summary.execCount > 1 {
+			schemas = append(schemas, summary.schemaName)
+			sqls = append(sqls, summary.sampleSQL)
+		}
+		summary.Unlock()
+	}
+	return schemas, sqls
+}
+
 // SetEnabled enables or disables statement summary in global(cluster) or session(server) scope.
 func (ssMap *stmtSummaryByDigestMap) SetEnabled(value string, inSession bool) {
 	value = ssMap.normalizeEnableValue(value)