sql: do not collect statistics on virtual columns

PR cockroachdb#68312 intended to update the behavior of  `CREATE STATISTICS` to
prevent statistics collection on virtual computed columns. However, it
failed to account for multi-column statistics and for
`CREATE STATISTICS` statements that explicitly reference virtual
columns. This commit accounts for these two cases.

This prevents internal errors from occuring when the system tries to
collect statistics on `NOT NULL` virtual columns. Virtual column values
are not included in the primary index. So when the statistics
job reads the primary index to sample the virtual column, it assumes the
value is null, which violates the column's `NOT NULL` constraint. This
violation causes an error.

Fixes cockroachdb#71080

Release note (bug fix): A bug has been fixed which caused internal
errors when collecting statistics on tables with virtual computed
columns.

pkg/sql/create_stats.go

@@ -247,6 +247,13 @@ func (n *createStatsNode) makeJobRecord(ctx context.Context) (*jobs.Record, erro
 
 		columnIDs := make([]descpb.ColumnID, len(columns))
 		for i := range columns {
+			if columns[i].IsVirtual() {
+				return nil, pgerror.Newf(
+					pgcode.InvalidColumnReference,
+					"cannot create statistics on virtual column %q",
+					columns[i].ColName(),
+				)
+			}
 			columnIDs[i] = columns[i].GetID()
 		}
 		col, err := tableDesc.FindColumnWithID(columnIDs[0])
@@ -441,9 +448,16 @@ func createStatsDefaultColumns(
 				continue
 			}
 
-			colIDs := make([]descpb.ColumnID, j+1)
+			colIDs := make([]descpb.ColumnID, 0, j+1)
 			for k := 0; k <= j; k++ {
-				colIDs[k] = idx.GetKeyColumnID(k)
+				col, err := desc.FindColumnWithID(idx.GetKeyColumnID(k))
+				if err != nil {
+					return nil, err
+				}
+				if col.IsVirtual() {
+					continue
+				}
+				colIDs = append(colIDs, col.GetID())
 			}
 
 			// Check for existing stats and remember the requested stats.

pkg/sql/logictest/testdata/logic_test/distsql_stats

@@ -824,11 +824,18 @@ upper_bound  range_rows  distinct_range_rows  equal_rows
 3            0           0                    1
 
 # Test that stats are not collected for virtual columns.
+statement ok
+SET CLUSTER SETTING sql.stats.multi_column_collection.enabled = true
+
 statement ok
 CREATE TABLE virt (
   a INT,
+  b INT,
   v INT AS (a + 10) VIRTUAL,
-  INDEX (v)
+  INDEX (v),
+  INDEX (a, v),
+  INDEX (a, v, b),
+  INDEX (a) WHERE v > 0
 )
 
 statement ok
@@ -850,7 +857,9 @@ ORDER BY
   column_names::STRING, created
 ----
 statistics_name  column_names  row_count  null_count  has_histogram
+s                {a,b}         3          0           false
 s                {a}           3          0           true
+s                {b}           3          3           true
 s                {rowid}       3          0           true
 
 # Test that stats are not collect for inaccessible virtual columns that
@@ -861,7 +870,9 @@ CREATE TABLE expression (
   b INT,
   j JSON,
   INDEX a_plus_b ((a + b)),
-  INVERTED INDEX j_a ((j->'a'))
+  INDEX a_a_plus_b (a, (a + b)),
+  INVERTED INDEX j_a ((j->'a')),
+  INVERTED INDEX a_j_a (a, (j->'a'))
 );
 INSERT INTO expression VALUES (1, 1, '{"a": "b"}'), (2, 10, '{"c": "d"}'), (3, 1, '{"e": "f"}');
 
@@ -1037,6 +1048,7 @@ ORDER BY
 column_names  has_histogram
 {id}          true
 {j}           true
+{s,j}         false
 {s}           true
 
 statement ok
@@ -1181,3 +1193,28 @@ CREATE TABLE t63387 (
 );
 INSERT INTO t63387 VALUES (1, '{}');
 CREATE STATISTICS s FROM t63387;
+
+# Regression test for #71080. Stats collection should succeed on tables with NOT
+# NULL virtual columns.
+statement ok
+SET CLUSTER SETTING sql.stats.multi_column_collection.enabled = true
+
+statement ok
+CREATE TABLE t71080 (
+  k INT PRIMARY KEY,
+  a INT,
+  b INT NOT NULL AS (a + 10) VIRTUAL,
+  INDEX (a, b)
+);
+
+statement ok
+INSERT INTO t71080 VALUES (1, 2);
+
+statement ok
+CREATE STATISTICS s FROM t71080;
+
+statement error cannot create statistics on virtual column \"b\"
+CREATE STATISTICS s ON b FROM t71080;
+
+statement error cannot create statistics on virtual column \"b\"
+CREATE STATISTICS s ON a, b FROM t71080;