• guilhem@gbichot2's avatar
    This is the final commit for Worklog tasks: · df3b1a54
    guilhem@gbichot2 authored
       * A more dynamic binlog format which allows small changes (1064)
       * Log session variables in Query_log_event (1063)
    It contains a few bugfixes (which I made when running the testsuite).
    I carefully updated the results of the testsuite (i.e. I checked for every one,
    if the difference between .reject and .result could be explained).
    Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently,
    so I could neither run the few tests which require --manager, nor check
    that they pass nor modify their .result. But for builds, we don't run
    with --manager.
    Apart from --manager, the full testsuite passes, with Valgrind too (no errors).
    I'm going to push in the next minutes. Remains: update the manual.
    Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone;
    this is normal at it makes assumptions on thread ids. I will fix this one day
    in 4.1.
    df3b1a54
rpl_loaddata.test 4.19 KB
# See if replication of a "LOAD DATA in an autoincrement column"
# Honours autoincrement values
# i.e. if the master and slave have the same sequence
#
# check replication of load data for temporary tables with additional
# parameters
#
# check if duplicate entries trigger an error (they should unless IGNORE or
# REPLACE was used on the master) (bug 571).
#
# check if START SLAVE, RESET SLAVE, CHANGE MASTER reset Last_slave_error and
# Last_slave_errno in SHOW SLAVE STATUS (1st and 3rd commands did not: bug 986)

source include/master-slave.inc;

connection slave;
reset master;
connection master;

create table t1(a int not null auto_increment, b int, primary key(a) );
load data infile '../../std_data/rpl_loaddata.dat' into table t1;

create temporary table t2 (day date,id int(9),category enum('a','b','c'),name varchar(60));
load data infile '../../std_data/rpl_loaddata2.dat' into table t2 fields terminated by ',' optionally enclosed by '%' escaped by '@' lines terminated by '\n##\n' starting by '>' ignore 1 lines;

create table t3 (day date,id int(9),category enum('a','b','c'),name varchar(60));
insert into t3 select * from t2;

save_master_pos;
connection slave;
sync_with_master;

select * from t1;
select * from t3;
# We want to be sure that LOAD DATA is in the slave's binlog.
# But we can't simply read this binlog, because as the slave has not been
# restarted for this test, the file_id is uncertain (would cause test
# failures). So instead, we test if the binlog looks long enough to
# contain LOAD DATA. That is, I (Guilhem) have done SHOW BINLOG EVENTS on my
# machine, saw that the binlog is of size 1068 (in 5.0.0) when things go fine.
# If LOAD DATA was not logged, the binlog would be shorter.
show master status;

connection master;

drop table t1;
drop table t2;
drop table t3;
create table t1(a int, b int, unique(b));

save_master_pos;
connection slave;
sync_with_master;

# See if slave stops when there's a duplicate entry for key error in LOAD DATA

insert into t1 values(1,10);

connection master;
load data infile '../../std_data/rpl_loaddata.dat' into table t1;

save_master_pos;
connection slave;
# The SQL slave thread should be stopped now.
wait_for_slave_to_stop;

# Skip the bad event and see if error is cleared in SHOW SLAVE STATUS by START
# SLAVE, even though we are not executing any event (as sql_slave_skip_counter
# takes us directly to the end of the relay log).

set global sql_slave_skip_counter=1;
start slave;
sync_with_master;
--replace_result $MASTER_MYPORT MASTER_PORT
--replace_column 1 # 8 # 9 # 23 # 33 #
show slave status;

# Trigger error again to test CHANGE MASTER

connection master;
set sql_log_bin=0;
delete from t1;
set sql_log_bin=1;
load data infile '../../std_data/rpl_loaddata.dat' into table t1;
save_master_pos;
connection slave;
# The SQL slave thread should be stopped now.
wait_for_slave_to_stop;

# CHANGE MASTER and see if error is cleared in SHOW SLAVE STATUS.
stop slave;
change master to master_user='test';
change master to master_user='root';
--replace_result $MASTER_MYPORT MASTER_PORT
--replace_column 1 # 8 # 9 # 23 # 33 #
show slave status;

# Trigger error again to test RESET SLAVE

set global sql_slave_skip_counter=1;
start slave;
sync_with_master;
connection master;
set sql_log_bin=0;
delete from t1;
set sql_log_bin=1;
load data infile '../../std_data/rpl_loaddata.dat' into table t1;
save_master_pos;
connection slave;
# The SQL slave thread should be stopped now.
wait_for_slave_to_stop;

# RESET SLAVE and see if error is cleared in SHOW SLAVE STATUS.
stop slave;
reset slave;
--replace_result $MASTER_MYPORT MASTER_PORT
--replace_column 1 # 8 # 9 # 23 # 33 #
show slave status;

# Finally, see if logging is done ok on master for a failing LOAD DATA INFILE

connection master;
reset master;
create table t2 (day date,id int(9),category enum('a','b','c'),name varchar(60),
unique(day));
--error 1062;
load data infile '../../std_data/rpl_loaddata2.dat' into table t2 fields
terminated by ',' optionally enclosed by '%' escaped by '@' lines terminated by
'\n##\n' starting by '>' ignore 1 lines;
# To test that there is Create_file & Delete_file, we test if the binlog is as
# long as expected (can't do SHOW BINLOG EVENTS because of varying file_id).
show master status;
drop table t2;