Merge #118425

118425: changefeedccl: make changefeed random expressions test deterministic r=jayshrivastava a=rharding6373

Before this change, it was difficult to reproduce errors from TestChangefeedRandomExpressions because the test setup involved non-deterministic queries. This PR both improves the test's determinism, logs setup information for easier reproducibility, and reduces the number of initial inserts to populate the test table.

Epic: None
Informs: #118061

Release note: none

Co-authored-by: rharding6373 <[email protected]>

pkg/ccl/backupccl/backuprand/backup_rand_test.go

@@ -86,7 +86,7 @@ database_name = 'rand' AND schema_name = 'public'`)
 		// Note: we do not care how many rows successfully populate
 		// the given table
 		if _, err := randgen.PopulateTableWithRandData(rng, tc.Conns[0], tableName,
-			numInserts); err != nil {
+			numInserts, nil); err != nil {
 			t.Fatal(err)
 		}
 		tableNames = append(tableNames, tableName)

pkg/ccl/changefeedccl/changefeed_test.go

@@ -1039,25 +1039,42 @@ func TestChangefeedRandomExpressions(t *testing.T) {
 		tblName := "seed"
 		defer s.DB.Close()
 
-		sqlDB.ExecMultiple(t, sqlsmith.Setups[tblName](rng)...)
+		setup := sqlsmith.Setups[tblName](rng)
+		sqlDB.ExecMultiple(t, setup...)
 
 		// TODO: PopulateTableWithRandData doesn't work with enums
-		sqlDB.Exec(t, "ALTER TABLE seed DROP COLUMN _enum")
+		dropEnumQry := "ALTER TABLE seed DROP COLUMN _enum;"
+		sqlDB.Exec(t, dropEnumQry)
 
+		// Attempt to insert a few more than our target 100 values, since there's a
+		// small chance we may not succeed that many inserts.
+		numInserts := 110
+		inserts := make([]string, 0, numInserts)
 		for rows := 0; rows < 100; {
 			var err error
 			var newRows int
-			if newRows, err = randgen.PopulateTableWithRandData(rng, s.DB, tblName, 200); err != nil {
+			if newRows, err = randgen.PopulateTableWithRandData(rng, s.DB, tblName, numInserts, &inserts); err != nil {
 				t.Fatal(err)
 			}
 			rows += newRows
 		}
 
-		sqlDB.Exec(t, `DELETE FROM seed WHERE rowid NOT IN (SELECT rowid FROM seed LIMIT 100)`)
+		limitQry := "DELETE FROM seed WHERE rowid NOT IN (SELECT rowid FROM seed ORDER BY rowid LIMIT 100);"
+		sqlDB.Exec(t, limitQry)
 
 		// Put the enums back. enum_range('hi'::greeting)[rowid%7] will give nulls when rowid%7=0 or 6.
-		sqlDB.Exec(t, `ALTER TABLE seed ADD COLUMN _enum greeting`)
-		sqlDB.Exec(t, `UPDATE seed SET _enum = enum_range('hi'::greeting)[rowid%7]`)
+		addEnumQry := "ALTER TABLE seed ADD COLUMN _enum greeting;"
+		sqlDB.Exec(t, addEnumQry)
+		populateEnumQry := "UPDATE seed SET _enum = enum_range('hi'::greeting)[rowid%7];"
+		sqlDB.Exec(t, populateEnumQry)
+		// Get values to log setup.
+		t.Logf("setup:\n%s\n%s\n%s\n%s\n%s\n%s",
+			strings.Join(setup, "\n"),
+			dropEnumQry,
+			strings.Join(inserts, "\n"),
+			limitQry,
+			addEnumQry,
+			populateEnumQry)
 
 		queryGen, err := sqlsmith.NewSmither(s.DB, rng,
 			sqlsmith.DisableWith(),

pkg/sql/importer/exportparquet_test.go

@@ -223,7 +223,7 @@ func TestRandomParquetExports(t *testing.T) {
 
 		for i = 0; i < numTables; i++ {
 			tableName = string(stmts[i].(*tree.CreateTable).Table.ObjectName)
-			numRows, err := randgen.PopulateTableWithRandData(rng, db, tableName, 20)
+			numRows, err := randgen.PopulateTableWithRandData(rng, db, tableName, 20, nil)
 			require.NoError(t, err)
 			if numRows > 5 {
 				// Ensure the table only contains columns supported by EXPORT Parquet. If an

pkg/sql/importer/read_import_avro_logical_test.go

@@ -281,7 +281,7 @@ func TestImportAvroLogicalTypes(t *testing.T) {
 	rng, _ := randutil.NewTestRand()
 	success := false
 	for i := 1; i <= 5; i++ {
-		numRowsInserted, err := randgen.PopulateTableWithRandData(rng, db, origTableName, 30)
+		numRowsInserted, err := randgen.PopulateTableWithRandData(rng, db, origTableName, 30, nil)
 		require.NoError(t, err)
 		if numRowsInserted > 5 {
 			success = true

pkg/sql/randgen/schema.go

@@ -319,11 +319,14 @@ func generateInsertStmtVals(rng *rand.Rand, colTypes []*types.T, nullable []bool
 //   - UNIQUE or CHECK constraint violation. RandDatum is naive to these constraints.
 //   - Out of range error for a computed INT2 or INT4 column.
 //
+// If a non-nil inserts is provided, it will be populated with the successful
+// insert statements.
+//
 // If numRowsInserted == 0, PopulateTableWithRandomData or RandDatum couldn't
 // handle this table's schema. Consider increasing numInserts or filing a bug.
 // TODO(harding): Populate data in partitions.
 func PopulateTableWithRandData(
-	rng *rand.Rand, db *gosql.DB, tableName string, numInserts int,
+	rng *rand.Rand, db *gosql.DB, tableName string, numInserts int, inserts *[]string,
 ) (numRowsInserted int, err error) {
 	var createStmtSQL string
 	res := db.QueryRow(fmt.Sprintf("SELECT create_statement FROM [SHOW CREATE TABLE %s]", tree.NameString(tableName)))
@@ -396,6 +399,9 @@ func PopulateTableWithRandData(
 		_, err := db.Exec(insertStmt)
 		if err == nil {
 			numRowsInserted++
+			if inserts != nil {
+				*inserts = append(*inserts, insertStmt)
+			}
 		}
 	}
 	return numRowsInserted, nil

pkg/sql/randgen/schema_test.go

@@ -68,7 +68,7 @@ func TestPopulateTableWithRandData(t *testing.T) {
 	for i := 0; i < numTables; i++ {
 		tableName := string(stmts[i].(*tree.CreateTable).Table.ObjectName)
 		numRows := 30
-		numRowsInserted, err := randgen.PopulateTableWithRandData(rng, dbConn, tableName, numRows)
+		numRowsInserted, err := randgen.PopulateTableWithRandData(rng, dbConn, tableName, numRows, nil)
 		require.NoError(t, err)
 		res := sqlDB.QueryStr(t, fmt.Sprintf("SELECT count(*) FROM %s", tree.NameString(tableName)))
 		require.Equal(t, fmt.Sprint(numRowsInserted), res[0][0])