planner: include both indexes and columns in job info

pkg/executor/analyze.go

@@ -124,7 +124,7 @@ func (e *AnalyzeExec) Next(ctx context.Context, _ *chunk.Chunk) error {
 		return e.handleResultsError(ctx, concurrency, needGlobalStats, globalStatsMap, resultsCh, len(tasks))
 	})
 	for _, task := range tasks {
-		prepareV2AnalyzeJobInfo(task.colExec, false)
+		prepareV2AnalyzeJobInfo(task.colExec)
 		AddNewAnalyzeJob(e.Ctx(), task.job)
 	}
 	failpoint.Inject("mockKillPendingAnalyzeJob", func() {

pkg/executor/analyze_col.go

@@ -384,29 +384,23 @@ func hasPkHist(handleCols plannerutil.HandleCols) bool {
 	return handleCols != nil && handleCols.IsInt()
 }
 
-func prepareV2AnalyzeJobInfo(e *AnalyzeColumnsExec, retry bool) {
-	if e == nil || e.StatsVersion != statistics.Version2 {
-		return
-	}
-	opts := e.opts
+// prepareColumns prepares the columns for the analyze job.
+func prepareColumns(e *AnalyzeColumnsExec, b *strings.Builder) {
 	cols := e.colsInfo
-	if e.V2Options != nil {
-		opts = e.V2Options.FilledOpts
-	}
-	sampleRate := *e.analyzePB.ColReq.SampleRate
-	var b strings.Builder
-	if retry {
-		b.WriteString("retry ")
-	}
-	if e.ctx.GetSessionVars().InRestrictedSQL {
-		b.WriteString("auto ")
-	}
-	b.WriteString("analyze table")
+	// Ignore the _row_id column.
 	if len(cols) > 0 && cols[len(cols)-1].ID == model.ExtraHandleID {
 		cols = cols[:len(cols)-1]
 	}
+	// If there are no columns, skip the process.
+	if len(cols) == 0 {
+		return
+	}
 	if len(cols) < len(e.tableInfo.Columns) {
-		b.WriteString(" columns ")
+		if len(cols) > 1 {
+			b.WriteString(" columns ")
+		} else {
+			b.WriteString(" column ")
+		}
 		for i, col := range cols {
 			if i > 0 {
 				b.WriteString(", ")
@@ -416,6 +410,58 @@ func prepareV2AnalyzeJobInfo(e *AnalyzeColumnsExec, retry bool) {
 	} else {
 		b.WriteString(" all columns")
 	}
+}
+
+// prepareIndexes prepares the indexes for the analyze job.
+func prepareIndexes(e *AnalyzeColumnsExec, b *strings.Builder) {
+	indexes := e.indexes
+
+	// If there are no indexes, skip the process.
+	if len(indexes) == 0 {
+		return
+	}
+	if len(indexes) < len(e.tableInfo.Indices) {
+		if len(indexes) > 1 {
+			b.WriteString(" indexes ")
+		} else {
+			b.WriteString(" index ")
+		}
+		for i, index := range indexes {
+			if i > 0 {
+				b.WriteString(", ")
+			}
+			b.WriteString(index.Name.O)
+		}
+	} else {
+		b.WriteString(" all indexes")
+	}
+}
+
+// prepareV2AnalyzeJobInfo prepares the job info for the analyze job.
+func prepareV2AnalyzeJobInfo(e *AnalyzeColumnsExec) {
+	// For v1, we analyze all columns in a single job, so we don't need to set the job info.
+	if e == nil || e.StatsVersion != statistics.Version2 {
+		return
+	}
+
+	opts := e.opts
+	if e.V2Options != nil {
+		opts = e.V2Options.FilledOpts
+	}
+	sampleRate := *e.analyzePB.ColReq.SampleRate
+	var b strings.Builder
+	// If it is an internal SQL, it means it is triggered by the system itself(auto-analyze).
+	if e.ctx.GetSessionVars().InRestrictedSQL {
+		b.WriteString("auto ")
+	}
+	b.WriteString("analyze table")
+
+	prepareIndexes(e, &b)
+	if len(e.indexes) > 0 && len(e.colsInfo) > 0 {
+		b.WriteString(",")
+	}
+	prepareColumns(e, &b)
+
 	var needComma bool
 	b.WriteString(" with ")
 	printOption := func(optType ast.AnalyzeOptionType) {

pkg/executor/show_stats_test.go

@@ -407,7 +407,7 @@ func TestShowAnalyzeStatus(t *testing.T) {
 	require.Equal(t, "test", rows[0][0])
 	require.Equal(t, "t", rows[0][1])
 	require.Equal(t, "", rows[0][2])
-	require.Equal(t, "analyze table all columns with 256 buckets, 100 topn, 1 samplerate", rows[0][3])
+	require.Equal(t, "analyze table all indexes, all columns with 256 buckets, 100 topn, 1 samplerate", rows[0][3])
 	require.Equal(t, "2", rows[0][4])
 	checkTime := func(val any) {
 		str, ok := val.(string)

pkg/executor/test/analyzetest/analyze_test.go

@@ -2156,19 +2156,19 @@ func TestShowAanalyzeStatusJobInfo(t *testing.T) {
 		require.Equal(t, expected, rows[0][3])
 		tk.MustExec("delete from mysql.analyze_jobs")
 	}
-	checkJobInfo("analyze table columns b, c, d with 2 buckets, 2 topn, 1 samplerate")
+	checkJobInfo("analyze table all indexes, columns b, c, d with 2 buckets, 2 topn, 1 samplerate")
 	tk.MustExec("set global tidb_persist_analyze_options = 1")
 	tk.MustExec("select * from t where c > 1")
 	h := dom.StatsHandle()
 	require.NoError(t, h.DumpColStatsUsageToKV())
 	tk.MustExec("analyze table t predicate columns with 2 topn, 2 buckets")
-	checkJobInfo("analyze table columns b, c, d with 2 buckets, 2 topn, 1 samplerate")
+	checkJobInfo("analyze table all indexes, columns b, c, d with 2 buckets, 2 topn, 1 samplerate")
 	tk.MustExec("analyze table t")
-	checkJobInfo("analyze table columns b, c, d with 2 buckets, 2 topn, 1 samplerate")
+	checkJobInfo("analyze table all indexes, columns b, c, d with 2 buckets, 2 topn, 1 samplerate")
 	tk.MustExec("analyze table t columns a with 1 topn, 3 buckets")
-	checkJobInfo("analyze table columns a, b, d with 3 buckets, 1 topn, 1 samplerate")
+	checkJobInfo("analyze table all indexes, columns a, b, d with 3 buckets, 1 topn, 1 samplerate")
 	tk.MustExec("analyze table t")
-	checkJobInfo("analyze table columns a, b, d with 3 buckets, 1 topn, 1 samplerate")
+	checkJobInfo("analyze table all indexes, columns a, b, d with 3 buckets, 1 topn, 1 samplerate")
 }
 
 func TestAnalyzePartitionTableWithDynamicMode(t *testing.T) {
@@ -2775,7 +2775,7 @@ func TestAnalyzeColumnsSkipMVIndexJsonCol(t *testing.T) {
 	tk.MustQuery("select job_info from mysql.analyze_jobs where table_schema = 'test' and table_name = 't'").Sort().Check(
 		testkit.Rows(
 			"analyze index idx_c",
-			"analyze table columns a, b with 256 buckets, 100 topn, 1 samplerate",
+			"analyze table index idx_b, columns a, b with 256 buckets, 100 topn, 1 samplerate",
 		))
 
 	is := dom.InfoSchema()
@@ -2914,7 +2914,7 @@ func TestAnalyzeMVIndex(t *testing.T) {
 	// 2. analyze and check analyze jobs
 	tk.MustExec("analyze table t with 1 samplerate, 3 topn")
 	tk.MustQuery("select id, table_schema, table_name, partition_name, job_info, processed_rows, state from mysql.analyze_jobs order by id").
-		Check(testkit.Rows("1 test t  analyze table columns a with 256 buckets, 3 topn, 1 samplerate 27 finished",
+		Check(testkit.Rows("1 test t  analyze table index ia, column a with 256 buckets, 3 topn, 1 samplerate 27 finished",
 			"2 test t  analyze index ij_signed 190 finished",
 			"3 test t  analyze index ij_unsigned 135 finished",
 			"4 test t  analyze index ij_double 154 finished",

pkg/statistics/handle/autoanalyze/autoanalyze_test.go

@@ -125,7 +125,7 @@ func TestAutoAnalyzeWithPredicateColumns(t *testing.T) {
 
 	// Check analyze jobs.
 	tk.MustQuery("select table_name, job_info from mysql.analyze_jobs order by id desc limit 1").Check(
-		testkit.Rows("t auto analyze table columns a with 256 buckets, 100 topn, 1 samplerate"),
+		testkit.Rows("t auto analyze table column a with 256 buckets, 100 topn, 1 samplerate"),
 	)
 }
 
@@ -342,7 +342,7 @@ func TestAutoAnalyzeSkipColumnTypes(t *testing.T) {
 		exec.AutoAnalyzeMinCnt = originalVal
 	}()
 	require.True(t, h.HandleAutoAnalyze())
-	tk.MustQuery("select job_info from mysql.analyze_jobs where job_info like '%auto analyze table%'").Check(testkit.Rows("auto analyze table columns a, b, d with 256 buckets, 100 topn, 1 samplerate"))
+	tk.MustQuery("select job_info from mysql.analyze_jobs where job_info like '%auto analyze table%'").Check(testkit.Rows("auto analyze table all indexes, columns a, b, d with 256 buckets, 100 topn, 1 samplerate"))
 }
 
 func TestAutoAnalyzeOnEmptyTable(t *testing.T) {

pkg/statistics/handle/usage/predicate_column_test.go

@@ -74,7 +74,7 @@ func TestAnalyzeTableWithPredicateColumns(t *testing.T) {
 	// Analyze table and check analyze jobs.
 	tk.MustExec("analyze table t")
 	tk.MustQuery("select table_name, job_info from mysql.analyze_jobs order by id desc limit 1").Check(
-		testkit.Rows("t analyze table columns a with 256 buckets, 100 topn, 1 samplerate"),
+		testkit.Rows("t analyze table column a with 256 buckets, 100 topn, 1 samplerate"),
 	)
 
 	// More columns.
@@ -175,7 +175,7 @@ func TestAnalyzeTableWithTiDBPersistAnalyzeOptionsDisabled(t *testing.T) {
 	// Analyze again, it should use the predicate columns.
 	tk.MustExec("analyze table t")
 	tk.MustQuery("select table_name, job_info from mysql.analyze_jobs order by id desc limit 1").Check(
-		testkit.Rows("t analyze table columns a with 256 buckets, 100 topn, 1 samplerate"),
+		testkit.Rows("t analyze table column a with 256 buckets, 100 topn, 1 samplerate"),
 	)
 }
 
@@ -242,9 +242,8 @@ func TestAnalyzeNoPredicateColumnsWithIndexes(t *testing.T) {
 			"Note 1105 Analyze use auto adjusted sample rate 1.000000 for table test.t, reason to use this rate is \"use min(1, 110000/10000) as the sample-rate=1\"",
 		),
 	)
-	// TODO: we should also include indexes in the job info.
 	tk.MustQuery("select table_name, job_info from mysql.analyze_jobs order by id desc limit 1").Check(
-		testkit.Rows("t analyze table columns a, b with 256 buckets, 100 topn, 1 samplerate"),
+		testkit.Rows("t analyze table all indexes, columns a, b with 256 buckets, 100 topn, 1 samplerate"),
 	)
 }
 
@@ -263,12 +262,18 @@ func TestAnalyzeWithNoPredicateColumnsAndNoIndexes(t *testing.T) {
 	err := h.DumpColStatsUsageToKV()
 	require.NoError(t, err)
 
+	// Check stats_meta first.
+	rows := tk.MustQuery("select * from mysql.stats_meta where version != 0").Rows()
+	require.Len(t, rows, 0, "haven't been analyzed yet")
 	// Analyze table.
 	tk.MustExec("analyze table t")
-	// FIXME: We should correct the job info or skip this kind of job.
 	tk.MustQuery("select table_name, job_info from mysql.analyze_jobs order by id desc limit 1").Check(
-		testkit.Rows("t analyze table columns  with 256 buckets, 100 topn, 1 samplerate"),
+		testkit.Rows("t analyze table with 256 buckets, 100 topn, 1 samplerate"),
 	)
+
+	// Check stats_meta again.
+	rows = tk.MustQuery("select * from mysql.stats_meta where version != 0 and modify_count = 0").Rows()
+	require.Len(t, rows, 1, "modify_count should be flushed")
 }
 
 func TestAnalyzeNoPredicateColumnsWithPrimaryKey(t *testing.T) {
@@ -290,6 +295,6 @@ func TestAnalyzeNoPredicateColumnsWithPrimaryKey(t *testing.T) {
 	// Analyze table.
 	tk.MustExec("analyze table t")
 	tk.MustQuery("select table_name, job_info from mysql.analyze_jobs order by id desc limit 1").Check(
-		testkit.Rows("t analyze table columns a, b with 256 buckets, 100 topn, 1 samplerate"),
+		testkit.Rows("t analyze table all indexes, columns a, b with 256 buckets, 100 topn, 1 samplerate"),
 	)
 }

tests/integrationtest/r/executor/analyze.result

@@ -824,12 +824,12 @@ delete from mysql.analyze_jobs;
 analyze table t;
 select job_info from mysql.analyze_jobs where job_info like '%analyze table%';
 job_info
-analyze table columns a, b, d with 256 buckets, 100 topn, 1 samplerate
+analyze table all indexes, columns a, b, d with 256 buckets, 100 topn, 1 samplerate
 delete from mysql.analyze_jobs;
 analyze table t columns a, e;
 select job_info from mysql.analyze_jobs where job_info like '%analyze table%';
 job_info
-analyze table columns a, d with 256 buckets, 100 topn, 1 samplerate
+analyze table all indexes, columns a, d with 256 buckets, 100 topn, 1 samplerate
 set @@session.tidb_analyze_skip_column_types = default;
 DROP TABLE IF EXISTS Issue34228;
 CREATE TABLE Issue34228 (id bigint NOT NULL, dt datetime NOT NULL) PARTITION BY RANGE COLUMNS(dt) (PARTITION p202201 VALUES LESS THAN ("2022-02-01"), PARTITION p202202 VALUES LESS THAN ("2022-03-01"));