Details
-
Bug
-
Status: Closed (View Workflow)
-
Blocker
-
Resolution: Fixed
-
10.5, 10.6, 10.11, 11.1(EOL), 11.2, 11.4, 10.5.25, 10.6.18, 10.6.19, 11.0(EOL), 11.3(EOL), 11.5(EOL)
Description
For specific Aria table and data (see later private message for more details):
MariaDB [db]> show create table T\G
|
*************************** 1. row ***************************
|
Table: imsmvi_fa_redacted
|
Create Table: CREATE TABLE `T` (
|
`SMFDATETIME` datetime DEFAULT NULL,
|
...
|
`SMFHOUR` int(11) DEFAULT NULL,
|
...
|
) ENGINE=Aria DEFAULT CHARSET=latin1 COLLATE=latin1_swedish_ci PAGE_CHECKSUM=0 ROW_FORMAT=PAGE TRANSACTIONAL=0;
|
|
|
MariaDB [db]> select count(*) from T;
|
+----------+
|
| count(*) |
|
+----------+
|
| 8594605 |
|
+----------+
|
1 row in set (0,050 sec)
|
with no indexes, the index on a column with a small number of distinct values (hours of the day) is create corrupted, on 10.6.19 built from current GitHub source and started with --no-defaults:
MariaDB [db]> create index T_smfhour on T(smfhour); show warnings;
|
ERROR 1034 (HY000): Number of rows changed from 8594605 to 2296732
|
+-------+------+------------------------------------------------+
|
| Level | Code | Message |
|
+-------+------+------------------------------------------------+
|
| Error | 1034 | Number of rows changed from 8594605 to 2296732 |
|
+-------+------+------------------------------------------------+
|
1 row in set (0,000 sec)
|
|
|
MariaDB [db]> select count(*) from T;
|
+----------+
|
| count(*) |
|
+----------+
|
| 2296732 |
|
+----------+
|
1 row in set (0,015 sec)
|
|
|
MariaDB [bad_redacted_db]> check table T;
|
+------------------------------------+-------+----------+--------------------------------------------------------------+
|
| Table | Op | Msg_type | Msg_text |
|
+------------------------------------+-------+----------+--------------------------------------------------------------+
|
| db.T | check | error | Record-count is not ok; found 8594605 Should be: 2296732 |
|
| db.T | check | error | Key pointers and record positions doesn't match |
|
| db.T | check | error | Checksum for key: 1 doesn't match checksum for records |
|
| db.T | check | error | Corrupt |
|
+------------------------------------+-------+----------+--------------------------------------------------------------+
|
4 rows in set (1 min 33,710 sec)
|
This was NOT the case on a pre-release 10.6.18 built from GitHub, so somehow there is a recent regression introduced at the later stage of 10.6.18 and/or 10.6.19 development..
Attachments
Issue Links
- is caused by
-
-
- Closed
-
- is duplicated by
-
-
- Closed
-
-
-
- Needs Feedback
-