List:Commits« Previous MessageNext Message »
From:Libing Song Date:May 13 2010 2:40pm
Subject:Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)
Bug#49741
View as plain text  
On Thu, 2010-05-13 at 11:18 +0100, Luís Soares wrote:
> Hi Li-Bing,
> 
> On Tue, 2010-05-11 at 12:51 +0800, Libing Song wrote:
> > On Fri, 2010-05-07 at 17:32 +0100, Luís Soares wrote: 
> > > > 
> > > > === modified file
> 'mysql-test/extra/rpl_tests/rpl_insert_delayed.test'
> > > > --- a/mysql-test/extra/rpl_tests/rpl_insert_delayed.test	2007-07-27
> 14:29:48 +0000
> > > > +++ b/mysql-test/extra/rpl_tests/rpl_insert_delayed.test	2010-04-28
> 08:26:07 +0000
> > > > @@ -107,8 +107,9 @@ flush table t1; # to wait for INSERT DEL
> > > >  if  (`SELECT @@global.binlog_format != 'ROW'`)
> > I will use 'if  ($binlog_format_statement)' instead of it.
> 
> Hmm... Won't that cause it to avoid the MIXED and ROW modes, 
> whereas, today, it only skips ROW?
> 
> In suite/rpl/t/rpl_stm_insert_delayed.test:
> 
> let $binlog_format_statement=1;
> set @@global.binlog_format = statement;
> --source extra/rpl_tests/rpl_insert_delayed.test
> 
> let $binlog_format_statement=0;
> set @@global.binlog_format = mixed;
> --source extra/rpl_tests/rpl_insert_delayed.test
> 
In MIXED mode, it will be binlogged as row events. 
This is not what the test cares.
> 

> > [snip]
> > > > === modified file 'mysql-test/suite/rpl/t/rpl_dual_pos_advance.test'
> > > > --- a/mysql-test/suite/rpl/t/rpl_dual_pos_advance.test	2009-01-09
> 14:12:31 +0000
> > > > +++ b/mysql-test/suite/rpl/t/rpl_dual_pos_advance.test	2010-04-28
> 08:26:07 +0000
> > > > @@ -40,6 +40,8 @@ sync_with_master;
> > > >  stop slave;
> > > >  
> > > >  create table t2 (n int); # create one ignored event
> > > > +let $master_log_pos_1= query_get_value(SHOW MASTER STATUS, Position,
> 1);
> > > > +let $master_log_pos_1= `SELECT $master_log_pos_1 + 5`;
> > > >  
> > > >  save_master_pos;
> > > >  connection slave;
> > > > @@ -52,8 +54,12 @@ show tables;
> > > >  save_master_pos;
> > > >  
> > > >  create table t3 (n int) engine=innodb;
> > > > +let $master_log_pos_2= query_get_value(SHOW MASTER STATUS, Position,
> 1);
> > > > +let $master_log_pos_2= `SELECT $master_log_pos_2 + 5`;
> > > >  set @a=1;
> > > >  insert into t3 values(@a);
> > > > +let $master_log_pos_3= query_get_value(SHOW MASTER STATUS, Position,
> 1);
> > > > +let $master_log_pos_3= `SELECT $master_log_pos_3 + 5`;
> > > >  begin;
> > > >  insert into t3 values(2);
> > > >  insert into t3 values(3);
> > > > @@ -66,7 +72,8 @@ connection master;
> > > >  # bug is that START SLAVE UNTIL may stop too late, we test that by
> > > >  # asking it to stop before creation of t3.
> > > >  
> > > > -start slave until
> master_log_file="slave-bin.000001",master_log_pos=195;
> > > > +--replace_result $master_log_pos_1 MASTER_LOG_POS
> > > > +eval start slave until
> master_log_file="slave-bin.000001",master_log_pos=$master_log_pos_1;
> > > 
> > > REQUEST: Please check if one can avoid explicitly stating the
> > >          binary log file.
> > agree and I think this test should run only on statement mode, do you think so?
> 
> Seems it can. However, I would suggest to run it on MIXED (if there is
> guarantee that no coverage is lost). Why? Because PB2 covers MIXED 
> in more platforms than it actually does for STATEMENT.
OK.
> 
[snip]
> 
> > > > === modified file 'mysql-test/t/sp_trans_log.test'
> > > > --- a/mysql-test/t/sp_trans_log.test	2009-04-28 18:42:17 +0000
> > > > +++ b/mysql-test/t/sp_trans_log.test	2010-04-28 08:26:07 +0000
> > > > @@ -34,8 +34,8 @@ end|
> > > >  reset master|
> > > >  --error ER_DUP_ENTRY
> > > >  insert into t2 values (bug23333(),1)| 
> > > > ---replace_column 2 # 5 # 6 #
> > > > -show binlog events from 106 /* with fixes for #23333 will show there
> is the query */|
> > > > +#23333 will show there is the query */|
> > > > +--source include/show_binlog_events.inc
> > > 
> > > REQUEST: can we avoid the result file here? Resorting to 
> > >          wait_for_binlog_event (in this case a Write_rows event?
> > I'm afraid we can't.
> > In fact, wait_for_binlog_event only check the last event's event_info. it
> doesn't check
> > the type of event.
> > I wonder bug#23333 can be test on RBR!
> > It's related to statement mode, so I think it should be test on statement mode.
> 
> Seems it might. Please check if it can be set to mixed instead of
> statement mode without loosing coverage.
Yes, it is set in mixed or row mode.
but it is binlogged as row events in mixed mode.
> 
> > Could you please check it? I will report a new bug, if you agree with me.
> > > 
> > > >  select count(*),@a from t1 /* must be 1,1 */|
> > > >  
> > > >  delimiter ;|
> > > 
> > > [snip]
> > > 
> > > 
> > > Regards,
> > > Luís
> > > 
> > > 
> 
> Regards,
> Luís
> 

-- 
Your Sincerely,
Libing Song
==================================
MySQL Replication Team
Software Engineer


Email : Li-Bing.Song@stripped
Skype : libing.song
MSN   : slb_database@stripped
Phone : +86 010-6505-4020 ext. 319
Mobile: +86 138-1144-2038
==================================

Thread
bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360) Bug#49741Li-Bing.Song28 Apr
  • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Sven Sandberg4 May
    • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Libing Song7 May
      • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Sven Sandberg10 May
  • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Luís Soares7 May
    • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Libing Song11 May
    • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Libing Song11 May
      • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Luís Soares13 May
        • Re: bzr commit into mysql-5.1-bugteam branch (Li-Bing.Song:3360)Bug#49741Libing Song13 May