| Bug #47678 | Changes to n-tables that happen early in a trans. are only flushed upon commit | ||
|---|---|---|---|
| Submitted: | 28 Sep 2009 15:26 | Modified: | 19 Dec 2009 11:39 |
| Reporter: | Alfranio Tavares Correia Junior | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Row Based Replication ( RBR ) | Severity: | S3 (Non-critical) |
| Version: | 5.1 | OS: | Any |
| Assigned to: | Alfranio Tavares Correia Junior | CPU Architecture: | Any |
| Tags: | regression | ||
[28 Sep 2009 15:32]
Alfranio Tavares Correia Junior
The suggested fix is based on BUG#47287.
[29 Sep 2009 9:17]
Bugs System
A patch for this bug has been committed. After review, it may be pushed to the relevant source trees for release in the next version. You can access the patch from: http://lists.mysql.com/commits/84962 3140 Alfranio Correia 2009-09-29 BUG#47678 Changes to n-tables that happen early in a trans. are only flushed upon commit Let - T be a transactional table and N non-transactional table. - B be begin, C commit and R rollback. - N be a statement that accesses and changes only N-tables. - T be a statement that accesses and changes only T-tables. In RBR, changes to N-tables that happen early in a transaction are not immediately flushed upon committing a statement. This behavior may, however, break consistency in the presence of concurrency since changes done to N-tables become immediately visible to other connections. To fix this problem, we do the following: . B N N T C would log - B N C B N C B T C. . B N N T R would log - B N C B N C B T R. Note that, we are not preserving history from the master as we are introducing a commit that never happened. However, this seems to be more acceptable than the possibility of breaking consistency in the presense of concurrency.
[6 Oct 2009 0:54]
Bugs System
A patch for this bug has been committed. After review, it may be pushed to the relevant source trees for release in the next version. You can access the patch from: http://lists.mysql.com/commits/85810 3140 Alfranio Correia 2009-10-06 BUG#47678 Changes to n-tables that happen early in a trans. are only flushed upon commit Let - T be a transactional table and N non-transactional table. - B be begin, C commit and R rollback. - N be a statement that accesses and changes only N-tables. - T be a statement that accesses and changes only T-tables. In RBR, changes to N-tables that happen early in a transaction are not immediately flushed upon committing a statement. This behavior may, however, break consistency in the presence of concurrency since changes done to N-tables become immediately visible to other connections. To fix this problem, we do the following: . B N N T C would log - B N C B N C B T C. . B N N T R would log - B N C B N C B T R. Note that we are not preserving history from the master as we are introducing a commit that never happened. However, this seems to be more acceptable than the possibility of breaking consistency in the presence of concurrency.
[7 Oct 2009 18:00]
Alfranio Tavares Correia Junior
Pushed to mysql-5.1-bugteam and mysql-pe.
[14 Oct 2009 14:39]
Bugs System
Pushed into 5.1.41 (revid:joro@sun.com-20091014143611-cphb0enjlx6lpat1) (version source revid:alfranio.correia@sun.com-20091006102536-3puq0gqxwe5oqid5) (merge vers: 5.1.40) (pib:13)
[15 Oct 2009 7:55]
Jon Stephens
Documented bugfix in the 5.1.41 changelog as follows:
When using row-based replication, changes to non-transactional
tables that occurred early in a transaction were not immediately
flushed upon committing a statement. This behavior could break
consistency since changes made to non-transactional tables
become immediately visible to other connections.
Set status to NDI in case of additional pushes.
(If this won't get pushed further, you can go ahead and close. Thanks!)
[22 Oct 2009 6:35]
Bugs System
Pushed into 6.0.14-alpha (revid:alik@sun.com-20091022063126-l0qzirh9xyhp0bpc) (version source revid:alik@sun.com-20091019135554-s1pvptt6i750lfhv) (merge vers: 6.0.14-alpha) (pib:13)
[22 Oct 2009 7:07]
Bugs System
Pushed into 5.5.0-beta (revid:alik@sun.com-20091022060553-znkmxm0g0gm6ckvw) (version source revid:alik@sun.com-20091015131113-wldlkdtoe9oebb2z) (merge vers: 5.5.0-beta) (pib:13)
[22 Oct 2009 15:13]
Jon Stephens
Also documented in the 5.5.0 and 6.0.14 changelogs. Closed.
[18 Dec 2009 10:33]
Bugs System
Pushed into 5.1.41-ndb-7.1.0 (revid:jonas@mysql.com-20091218102229-64tk47xonu3dv6r6) (version source revid:jonas@mysql.com-20091218095730-26gwjidfsdw45dto) (merge vers: 5.1.41-ndb-7.1.0) (pib:15)
[18 Dec 2009 10:49]
Bugs System
Pushed into 5.1.41-ndb-6.2.19 (revid:jonas@mysql.com-20091218100224-vtzr0fahhsuhjsmt) (version source revid:jonas@mysql.com-20091217101452-qwzyaig50w74xmye) (merge vers: 5.1.41-ndb-6.2.19) (pib:15)
[18 Dec 2009 11:04]
Bugs System
Pushed into 5.1.41-ndb-6.3.31 (revid:jonas@mysql.com-20091218100616-75d9tek96o6ob6k0) (version source revid:jonas@mysql.com-20091217154335-290no45qdins5bwo) (merge vers: 5.1.41-ndb-6.3.31) (pib:15)
[18 Dec 2009 11:19]
Bugs System
Pushed into 5.1.41-ndb-7.0.11 (revid:jonas@mysql.com-20091218101303-ga32mrnr15jsa606) (version source revid:jonas@mysql.com-20091218064304-ezreonykd9f4kelk) (merge vers: 5.1.41-ndb-7.0.11) (pib:15)
[19 Dec 2009 11:39]
Jon Stephens
No additional changelog entries needed. Setting back to Closed state.
Description: In RBR, changes to non-transactional tables that happen early in a transaction are not immediately flushed upon committing a statement. This behavior may, however, break consistency since changes done to non-transactional tables become immediately visible to other connections. How to repeat: Test case: ---------- --source include/have_binlog_format_row.inc --source include/have_innodb.inc CREATE TABLE tt (a int) engine=innodb; CREATE TABLE nt (a int) engine=myisam; BEGIN; INSERT INTO nt VALUES(1); INSERT INTO nt VALUES(2); INSERT INTO nt VALUES(3); INSERT INTO tt VALUES(1); INSERT INTO tt VALUES(2); INSERT INTO nt VALUES(4); INSERT INTO tt VALUES(3); COMMIT; SHOW BINLOG EVENTS; DROP TABLE tt; DROP TABLE nt; exit; We have the wrong ouput: ------------------------ master-bin.000001 306 Query 1 374 BEGIN master-bin.000001 374 Table_map 1 415 table_id: 23 (test.nt) master-bin.000001 415 Write_rows 1 449 table_id: 23 flags: STMT_END_F master-bin.000001 449 Table_map 1 490 table_id: 23 (test.nt) master-bin.000001 490 Write_rows 1 524 table_id: 23 flags: STMT_END_F master-bin.000001 524 Table_map 1 565 table_id: 23 (test.nt) master-bin.000001 565 Write_rows 1 599 table_id: 23 flags: STMT_END_F master-bin.000001 599 Table_map 1 640 table_id: 24 (test.tt) master-bin.000001 640 Write_rows 1 674 table_id: 24 flags: STMT_END_F master-bin.000001 674 Table_map 1 715 table_id: 24 (test.tt) master-bin.000001 715 Write_rows 1 749 table_id: 24 flags: STMT_END_F master-bin.000001 749 Table_map 1 790 table_id: 23 (test.nt) master-bin.000001 790 Write_rows 1 824 table_id: 23 flags: STMT_END_F master-bin.000001 824 Table_map 1 865 table_id: 24 (test.tt) master-bin.000001 865 Write_rows 1 899 table_id: 24 flags: STMT_END_F master-bin.000001 899 Xid 1 926 COMMIT /* xid=15 */ The correct output should be: ----------------------------- master-bin.000001 4 Format_desc 1 106 Server ver: 5.1.40-debug-log, Binlog ver: 4 master-bin.000001 106 Query 1 206 use `test`; CREATE TABLE tt (a int) engine=innodb master-bin.000001 206 Query 1 306 use `test`; CREATE TABLE nt (a int) engine=myisam master-bin.000001 306 Query 1 374 BEGIN master-bin.000001 374 Table_map 1 415 table_id: 23 (test.nt) master-bin.000001 415 Write_rows 1 449 table_id: 23 flags: STMT_END_F master-bin.000001 449 Query 1 518 COMMIT master-bin.000001 518 Query 1 586 BEGIN master-bin.000001 586 Table_map 1 627 table_id: 23 (test.nt) master-bin.000001 627 Write_rows 1 661 table_id: 23 flags: STMT_END_F master-bin.000001 661 Query 1 730 COMMIT master-bin.000001 730 Query 1 798 BEGIN master-bin.000001 798 Table_map 1 839 table_id: 23 (test.nt) master-bin.000001 839 Write_rows 1 873 table_id: 23 flags: STMT_END_F master-bin.000001 873 Query 1 942 COMMIT master-bin.000001 942 Query 1 1010 BEGIN master-bin.000001 1010 Table_map 1 1051 table_id: 24 (test.tt) master-bin.000001 1051 Write_rows 1 1085 table_id: 24 flags: STMT_END_F master-bin.000001 1085 Table_map 1 1126 table_id: 24 (test.tt) master-bin.000001 1126 Write_rows 1 1160 table_id: 24 flags: STMT_END_F master-bin.000001 1160 Table_map 1 1201 table_id: 23 (test.nt) master-bin.000001 1201 Write_rows 1 1235 table_id: 23 flags: STMT_END_F master-bin.000001 1235 Table_map 1 1276 table_id: 24 (test.tt) master-bin.000001 1276 Write_rows 1 1310 table_id: 24 flags: STMT_END_F master-bin.000001 1310 Xid 1 1337 COMMIT /* xid=15 */ Suggested fix: === modified file 'sql/log.cc' --- sql/log.cc 2009-09-28 12:55:54 +0000 +++ sql/log.cc 2009-09-28 14:22:43 +0000 @@ -1538,7 +1538,10 @@ YESNO(in_transaction), YESNO(thd->transaction.all.modified_non_trans_table), YESNO(thd->transaction.stmt.modified_non_trans_table))); - if (!in_transaction || all) + if (!in_transaction || all || + (!all && !trx_data->at_least_one_stmt_committed && + !stmt_has_updated_trans_table(thd) && + thd->transaction.stmt.modified_non_trans_table)) { Query_log_event qev(thd, STRING_WITH_LEN("COMMIT"), TRUE, TRUE, 0); error= binlog_end_trans(thd, trx_data, &qev, all);