[MDEV-8901] InnoDB: Punch hole is incorrecty done also to log files causing assertion and database corruption Created: 2015-10-06  Updated: 2015-10-07  Resolved: 2015-10-07

Status: Closed
Project: MariaDB Server
Component/s: Storage Engine - InnoDB, Storage Engine - XtraDB
Affects Version/s: 10.1.7
Fix Version/s: 10.1.8

Type: Bug Priority: Critical
Reporter: Jan Lindström (Inactive) Assignee: Jan Lindström (Inactive)
Resolution: Fixed Votes: 0
Labels: None


 Description   

8b in abort () from /lib/x86_64-linux-gnu/libc.so.6
---Type <return> to continue, or q <return> to quit---
#2  0x0000555555fed2ff in os_file_trim (slot=0x7ffff5779000)
    at /home/jplindst/10.1/storage/xtradb/os/os0file.cc:6313
#3  0x0000555555feb4ac in os_aio_linux_collect (array=0x7ffff57fe0b0, segment=0, seg_size=256)
    at /home/jplindst/10.1/storage/xtradb/os/os0file.cc:5395
#4  0x0000555555feb73c in os_aio_linux_handle (global_seg=1, message1=0x7fffd7ffdde8, 
    message2=0x7fffd7ffddf0, type=0x7fffd7ffddf8, space_id=0x7fffd7ffde00)
    at /home/jplindst/10.1/storage/xtradb/os/os0file.cc:5515
#5  0x00005555561c229a in fil_aio_wait (segment=1)
    at /home/jplindst/10.1/storage/xtradb/fil/fil0fil.cc:5992
#6  0x00005555560b4eb0 in io_handler_thread (arg=0x55555750a5a8 <n+8>)
    at /home/jplindst/10.1/storage/xtradb/srv/srv0start.cc:542
#7  0x00007ffff68f9e9a in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0
#8  0x00007ffff6026cbd in clone () from /lib/x86_64-linux-gnu/libc.so.6
#9  0x0000000000000000 in ?? ()



 Comments   
Comment by Jan Lindström (Inactive) [ 2015-10-06 ]

After two bug fixes I see page compressed table to use less actual disk space:

jplindst@perro:~/db$ ../10.1/client/mysql -u root -S mysql.sock2
Welcome to the MariaDB monitor.  Commands end with ; or \g.
Your MariaDB connection id is 2
Server version: 10.1.8-MariaDB-debug Source distribution
 
Copyright (c) 2000, 2015, Oracle, MariaDB Corporation Ab and others.
 
Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.
 
MariaDB [(none)]> use test;
Database changed
MariaDB [test]> create table a(c int not null default 0,t char(255) not null default '') engine=innodb;insert into a VALUES(912312,'asdljkashdkhqgerhwvqefasqeldkjasdljadkljasdasdd');
Query OK, 0 rows affected (0.91 sec)
 
Query OK, 1 row affected (0.07 sec)
 
MariaDB [test]>  insert into a select * from a;
Query OK, 1 row affected (0.70 sec)
Records: 1  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 2 rows affected (0.06 sec)
Records: 2  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 4 rows affected (0.05 sec)
Records: 4  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 8 rows affected (0.04 sec)
Records: 8  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 16 rows affected (0.04 sec)
Records: 16  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 32 rows affected (0.06 sec)
Records: 32  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 64 rows affected (0.10 sec)
Records: 64  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 128 rows affected (0.21 sec)
Records: 128  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 256 rows affected (0.32 sec)
Records: 256  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 512 rows affected (0.58 sec)
Records: 512  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 1024 rows affected (1.35 sec)
Records: 1024  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 2048 rows affected (0.34 sec)
Records: 2048  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 4096 rows affected (0.60 sec)
Records: 4096  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 8192 rows affected (1.77 sec)
Records: 8192  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 16384 rows affected (2.24 sec)
Records: 16384  Duplicates: 0  Warnings: 0
 
MariaDB [test]>  insert into a select * from a;
Query OK, 32768 rows affected (3.90 sec)
Records: 32768  Duplicates: 0  Warnings: 0
 
MariaDB [test]> create table b(c int not null default 0,t char(255) not null default '') page_compressed=1 engine=innodb;
Query OK, 0 rows affected (0.31 sec)
 
MariaDB [test]> insert into b select * from a;
Query OK, 65536 rows affected (13.49 sec)
Records: 65536  Duplicates: 0  Warnings: 0
 
MariaDB [test]> show status;
+--------------------------------------------------------------+----------------------------------------+
| Variable_name                                                | Value                                  |
+--------------------------------------------------------------+----------------------------------------+
<cut>
| Innodb_page_compression_saved                                | 14766080                               |
| Innodb_page_compression_trim_sect512                         | 0                                      |
| Innodb_page_compression_trim_sect1024                        | 0                                      |
| Innodb_page_compression_trim_sect2048                        | 0                                      |
| Innodb_page_compression_trim_sect4096                        | 3605                                   |
| Innodb_page_compression_trim_sect8192                        | 0                                      |
| Innodb_page_compression_trim_sect16384                       | 0                                      |
| Innodb_page_compression_trim_sect32768                       | 0                                      |
| Innodb_num_index_pages_written                               | 1280                                   |
| Innodb_num_non_index_pages_written                           | 1483                                   |
| Innodb_num_pages_page_compressed                             | 1203                                   |
| Innodb_num_page_compressed_trim_op                           | 1203                                   |
| Innodb_num_page_compressed_trim_op_saved                     | 0                                      |
| Innodb_num_pages_page_decompressed                           | 1203                                   |
| Innodb_num_pages_page_compression_error                      | 0                                      |
<cut>
+--------------------------------------------------------------+----------------------------------------+
515 rows in set (0.01 sec)
 
MariaDB [test]> exit
Bye
jplindst@perro:~/db$ killall mysqld
jplindst@perro:~/db$ ls
aria_log.00000001  err.log  ib_logfile0  multi-master.info  my.cnf~  performance_schema
aria_log_control   ibdata1  ib_logfile1  my.cnf             mysql    test
[1]+  Done                    10.1/sql/mysqld --defaults-file=/home/jplindst/db/my.cnf  (wd: ~)
(wd now: ~/db)
jplindst@perro:~/db$ cd test
jplindst@perro:~/db/test$ ls -la
total 40428
drwx------ 2 jplindst my     4096 Oct  6 09:53 .
drwxr-xr-x 5 jplindst my     4096 Oct  6 09:55 ..
-rw-rw---- 1 jplindst my      706 Oct  6 09:52 a.frm
-rw-rw---- 1 jplindst my 28311552 Oct  6 09:53 a.ibd
-rw-rw---- 1 jplindst my      730 Oct  6 09:53 b.frm
-rw-rw---- 1 jplindst my 28311552 Oct  6 09:54 b.ibd
jplindst@perro:~/db/test$ du
40424   .
jplindst@perro:~/db/test$ du a.ibd
27652   a.ibd
jplindst@perro:~/db/test$ du b.ibd
12760   b.ibd

Comment by Jan Lindström (Inactive) [ 2015-10-06 ]

Confirmed that innodb-use-trim=1 causes database corruption at least for row compressed tables because actual page size (zip_size) is not known on os_file_trim.

Comment by Jan Lindström (Inactive) [ 2015-10-06 ]

Ok, fix found.

Comment by Jan Lindström (Inactive) [ 2015-10-07 ]

commit 21adad000a00a62357c0da63ff9ae98850475618
Author: Jan Lindström <jan.lindstrom@mariadb.com>
Date: Tue Oct 6 15:40:26 2015 +0300

MDEV-8901: InnoDB: Punch hole is incorrecty done also to log files causing assertion and database corruption

Analysis: Problem is that punch hole does not know the actual page size
of the page and does the page belong to an data file or to a log file.

Fix: Pass down the file type and page size to os layer to be used
when trim is called. Also fix unsafe null pointer access to
actual write_size.

Generated at Thu Feb 08 07:30:38 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.