2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
drop table if exists ti, tm;
|
|
|
|
|
create table ti (a int) type=innodb;
|
|
|
|
|
create table tm (a int) type=myisam;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(1);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
commit;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(1)
|
|
|
|
|
master-bin.000001 178 Query 1 79 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 244 Query 1 244 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(2);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
rollback;
|
|
|
|
|
Warning: Some non-transactional changed tables couldn't be rolled back
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(2)
|
|
|
|
|
master-bin.000001 178 Query 1 79 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 244 Query 1 244 use ; ROLLBACK
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(3);
|
|
|
|
|
savepoint my_savepoint;
|
|
|
|
|
insert into ti values(4);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
rollback to savepoint my_savepoint;
|
|
|
|
|
Warning: Some non-transactional changed tables couldn't be rolled back
|
|
|
|
|
commit;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(3)
|
|
|
|
|
master-bin.000001 178 Query 1 79 use ; savepoint my_savepoint
|
|
|
|
|
master-bin.000001 235 Query 1 79 use ; insert into ti values(4)
|
|
|
|
|
master-bin.000001 294 Query 1 79 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 360 Query 1 79 use ; rollback to savepoint my_savepoint
|
|
|
|
|
master-bin.000001 429 Query 1 429 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(5);
|
|
|
|
|
savepoint my_savepoint;
|
|
|
|
|
insert into ti values(6);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
rollback to savepoint my_savepoint;
|
|
|
|
|
Warning: Some non-transactional changed tables couldn't be rolled back
|
|
|
|
|
insert into ti values(7);
|
|
|
|
|
commit;
|
|
|
|
|
select a from ti order by a;
|
|
|
|
|
a
|
|
|
|
|
5
|
|
|
|
|
7
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(5)
|
|
|
|
|
master-bin.000001 178 Query 1 79 use ; savepoint my_savepoint
|
|
|
|
|
master-bin.000001 235 Query 1 79 use ; insert into ti values(6)
|
|
|
|
|
master-bin.000001 294 Query 1 79 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 360 Query 1 79 use ; rollback to savepoint my_savepoint
|
|
|
|
|
master-bin.000001 429 Query 1 79 use ; insert into ti values(7)
|
|
|
|
|
master-bin.000001 488 Query 1 488 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
select get_lock("a",10);
|
|
|
|
|
get_lock("a",10)
|
|
|
|
|
1
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(8);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
select get_lock("a",10);
|
|
|
|
|
get_lock("a",10)
|
|
|
|
|
1
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(8)
|
|
|
|
|
master-bin.000001 178 Query 1 79 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 244 Query 1 244 use ; ROLLBACK
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
insert into ti values(9);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; insert into ti values(9)
|
|
|
|
|
master-bin.000001 138 Query 1 138 use ; insert into tm select * from ti
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
insert into ti values(10);
|
|
|
|
|
begin;
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; insert into ti values(10)
|
|
|
|
|
master-bin.000001 139 Query 1 139 use ; insert into tm select * from ti
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
insert into ti values(11);
|
|
|
|
|
commit;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; insert into ti values(10)
|
|
|
|
|
master-bin.000001 139 Query 1 139 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 205 Query 1 205 use ; BEGIN
|
|
|
|
|
master-bin.000001 245 Query 1 205 use ; insert into ti values(11)
|
|
|
|
|
master-bin.000001 305 Query 1 305 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
alter table tm type=INNODB;
|
|
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(12);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
commit;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(12)
|
|
|
|
|
master-bin.000001 179 Query 1 79 use ; insert into tm select * from ti
|
|
|
|
|
master-bin.000001 245 Query 1 245 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(13);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
rollback;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
|
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(14);
|
|
|
|
|
savepoint my_savepoint;
|
|
|
|
|
insert into ti values(15);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
rollback to savepoint my_savepoint;
|
|
|
|
|
commit;
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(14)
|
|
|
|
|
master-bin.000001 179 Query 1 179 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
delete from ti;
|
|
|
|
|
delete from tm;
|
|
|
|
|
reset master;
|
|
|
|
|
begin;
|
|
|
|
|
insert into ti values(16);
|
|
|
|
|
savepoint my_savepoint;
|
|
|
|
|
insert into ti values(17);
|
|
|
|
|
insert into tm select * from ti;
|
|
|
|
|
rollback to savepoint my_savepoint;
|
|
|
|
|
insert into ti values(18);
|
|
|
|
|
commit;
|
|
|
|
|
select a from ti order by a;
|
|
|
|
|
a
|
|
|
|
|
16
|
|
|
|
|
18
|
|
|
|
|
show binlog events from 79;
|
|
|
|
|
Log_name Pos Event_type Server_id Orig_log_pos Info
|
2003-09-11 18:06:23 +02:00
|
|
|
master-bin.000001 79 Query 1 79 use ; BEGIN
|
|
|
|
|
master-bin.000001 119 Query 1 79 use ; insert into ti values(16)
|
|
|
|
|
master-bin.000001 179 Query 1 79 use ; insert into ti values(18)
|
|
|
|
|
master-bin.000001 239 Query 1 239 use ; COMMIT
|
2 minor edits, plus
fix for BUG#1113 "INSERT into non-trans table SELECT ; ROLLBACK" does not send warning"
and
fix for BUG#873 "In transaction, INSERT to non-trans table is written too early to binlog".
Now we don't always write the non-trans update immediately to the binlog;
if there is something in the binlog cache we write it to the binlog cache
(because the non-trans update could depend on a trans table which was modified
earlier in the transaction); then in case of ROLLBACK, we write the binlog
cache to the binlog, wrapped with BEGIN/ROLLBACK.
This guarantees that the slave does the same updates.
For ROLLBACK TO SAVEPOINT: when we execute a SAVEPOINT command we write it
to the binlog cache. At ROLLBACK TO SAVEPOINT, if some non-trans table was updated,
we write ROLLBACK TO SAVEPOINT to the binlog cache; when the transaction
terminates (COMMIT/ROLLBACK), the binlog cache will be flushed to the binlog
(because of the non-trans update) so we'll have SAVEPOINT and ROLLBACK TO
SAVEPOINT in the binlog.
Apart from this rare case of updates of mixed table types in transaction, the
usual way is still clear the binlog cache at ROLLBACK, or chop it at
ROLLBACK TO SAVEPOINT (meaning the SAVEPOINT command is also chopped, which
is fine).
Note that BUG#873 encompasses subbugs 1) and 2) of BUG#333 "3 binlogging bugs when doing INSERT with mixed InnoDB/MyISAM".
client/mysqldump.c:
Minor edit: one CHANGE MASTER with 2 arguments instead of 2 CHANGE MASTER with one argument each.
mysql-test/r/rpl_loaddata.result:
result update
mysql-test/t/rpl_loaddata.test:
minor edit: simplifying the test.
sql/handler.cc:
Fix for BUG#873. See comments in code, and the description of the changeset.
sql/log.cc:
* Previously, if a query updated a non-transactional table we wrote it immediately
to the real binlog. This causes a bug when the update is done inside a transaction
and uses the content of an updated transactional table (because this makes
a wrong order of queries in the binlog). So if the binlog cache is not empty,
we write the query to the binlog cache; otherwise we can write it to the binlog.
* Previously, when we flushed the binlog cache to the binlog, we wrapped it
with BEGIN/COMMIT. Now it's also possible to wrap it with BEGIN/ROLLBACK, to handle
transactions which update both transactional and non-transactional tables.
sql/log_event.cc:
The slave thread can leave a transaction if COMMIT or if ROLLBACK.
sql/sql_class.h:
prototype
sql/sql_insert.cc:
Fix for BUG#1113:
this was because the INSERT SELECT code did not set OPTION_STATUS_NO_TRANS_UPDATE.
sql/sql_parse.cc:
Don't send ER_WARNING_NOT_COMPLETE_ROLLBACK if this is the SQL slave thread (see comments).
2003-08-22 15:39:24 +02:00
|
|
|
drop table ti,tm;
|