mariadb/mysql-test/extra/rpl_tests/rpl_conflicts.test
Sven Sandberg d20aba9409 BUG#37718: rpl.rpl_stm_mystery22 fails sporadically on pushbuild
Problem 1: The test waits for an error in the slave sql thread,
then resolves the error and issues 'start slave'. However, there
is a gap between when the error is reported and the slave sql
thread stops. If this gap was long, the slave would still be
running when 'start slave' happened, so 'start slave' would fail
and cause a test failure.
Fix 1: Made wait_for_slave_sql_error wait for the slave to stop
instead of wait for error in the IO thread. After stopping, the
error code is verified. If the error code is wrong, debug info
is printed. To print debug info, the debug printing code in
wait_for_slave_param.inc was moved out to a new file,
show_rpl_debug_info.inc.
Problem 2: rpl_stm_mystery22 is a horrible name, the comments in
the file didn't explain anything useful, the test was generally
hard to follow, and the test was essentially duplicated between
rpl_stm_mystery22 and rpl_row_mystery22.
Fix 2: The test is about conflicts in the slave SQL thread,
hence I renamed the tests to rpl_{stm,row}_conflicts. Refactored
the test so that the work is done in
extra/rpl_tests/rpl_conflicts.inc, and
rpl.rpl_{row,stm}_conflicts merely sets some variables and then
sourced extra/rpl_tests/rpl_conflicts.inc.
The tests have been rewritten and comments added.
Problem 3: When calling wait_for_slave_sql_error.inc, you always
want to verify that the sql thread stops because of the expected
error and not because of some other error. Currently,
wait_for_slave_sql_error.inc allows the caller to omit the error
code, in which case all error codes are accepted.
Fix 3: Made wait_for_slave_sql_error.inc fail if no error code
is given. Updated rpl_filter_tables_not_exist accordingly.
Problem 4: rpl_filter_tables_not_exist had a typo, the dollar
sign was missing in a 'let' statement.
Fix 4: Added dollar sign.
Problem 5: When replicating from other servers than the one named
'master', the wait_for_slave_* macros were unable to print debug
info on the master.
Fix 5: Replace parameter $slave_keep_connection by
$master_connection.


mysql-test/extra/rpl_tests/rpl_conflicts.test:
  rpl_stm_mystery22 and rpl_row_mystery22 have now been refactored and renamed:
  The two test cases rpl.rpl_stm_conflicts.test and rpl.rpl_row_conflicts.test
  just set some parameters, and then source include/rpl_tests/rpl_conflicts.test.
  Also, cleaned up the test case a bit, and fixed BUG#37718.
mysql-test/include/show_rpl_debug_info.inc:
  Factored out the debug printing code from wait_for_slave_param.inc to
  a new file, show_rpl_debug_info.inc.
  Also removed the $slave_keep_connection parameter, and replaced it by
  $master_connection. This allows printing debug info on the master, no
  matter what the name of the master connection is.
mysql-test/include/start_slave.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/stop_slave.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/sync_slave_io_with_master.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_io_to_start.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_io_to_stop.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_param.inc:
  Factored out the debug printing code from wait_for_slave_param.inc to
  a new file, show_rpl_debug_info.inc.
  Also removed the $slave_keep_connection parameter, and replaced it by
  $master_connection. This allows printing debug info on the master, no
  matter what the name of the master connection is.
  Had to move the printing of debug info out of the while loop because
  of BUG number 41913.
mysql-test/include/wait_for_slave_sql_error.inc:
  Made it wait until the slave sql thread has stopped. This
  takes very short time and avoids race condition bugs in
  test cases (e.g., fixes BUG#37718).
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_sql_error_and_skip.inc:
  Since wait_for_slave_sql_error now waits for the slave sql
  thread to stop too, wait_for_slave_sql_error_and_skip does not
  have to wait for the slave sql thread to stop.
  Also, since wait_for_slave_sql_error now requires the parameter
  $slave_sql_errno to be set, wait_for_slave_sql_error_and_skip
  requires that as well: updated the usage instructions.
mysql-test/include/wait_for_slave_sql_to_start.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_sql_to_stop.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_to_start.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/include/wait_for_slave_to_stop.inc:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/suite/rpl/r/rpl_row_conflicts.result:
  update result file
mysql-test/suite/rpl/r/rpl_stm_conflicts.result:
  update result file
mysql-test/suite/rpl/t/rpl_dual_pos_advance.test:
  Replaced $slave_keep_connection by $master_connection.
mysql-test/suite/rpl/t/rpl_filter_tables_not_exist.test:
  Set $slave_sql_errno, since it is now required.
  Add dollar sign to $show_sql_error (without dollar sign,
  mtr makes it an environment variable).
mysql-test/suite/rpl/t/rpl_row_conflicts.test:
  rpl_stm_mystery22 and rpl_row_mystery22 have now been refactored and renamed:
  The two test cases rpl.rpl_stm_conflicts.test and rpl.rpl_row_conflicts.test
  just set some parameters, and then source include/rpl_tests/rpl_conflicts.test.
  Also, cleaned up the test case a bit, and fixed BUG#37718.
mysql-test/suite/rpl/t/rpl_row_mystery22.test:
  rpl_stm_mystery22 and rpl_row_mystery22 have now been refactored and renamed:
  The two test cases rpl.rpl_stm_conflicts.test and rpl.rpl_row_conflicts.test
  just set some parameters, and then source include/rpl_tests/rpl_conflicts.test.
  Also, cleaned up the test case a bit, and fixed BUG#37718.
mysql-test/suite/rpl/t/rpl_stm_conflicts.test:
  rpl_stm_mystery22 and rpl_row_mystery22 have now been refactored and renamed:
  The two test cases rpl.rpl_stm_conflicts.test and rpl.rpl_row_conflicts.test
  just set some parameters, and then source include/rpl_tests/rpl_conflicts.test.
  Also, cleaned up the test case a bit, and fixed BUG#37718.
2009-01-09 15:12:31 +01:00

168 lines
5.2 KiB
Text

# ==== Purpose ====
#
# Test that slave behaves well in some conflict situations. The
# following are tested:
#
# - The slave SQL thread sees an 'INSERT' of a row with a key that
# already exists in the table;
#
# - The slave SQL thread sees a 'DELETE' of a row that does not
# exist in the table.
#
# In statement-logging mode, the first conflict type causes the slave
# to stop with an error and the second conflict is ignored.
#
# In row-logging mode, the slave behavior depends the value of
# @@slave_exec_mode on the slave: if @@slave_exec_mode is IDEMPOTENT,
# the slave should ignore the conflicting statement and continue
# normally. If @@slave_exec_mode is STRICT, the slave should stop
# with an error.
#
# This test was previously named rpl_stm_mystery22/rpl_row_mystery22.
#
#
# ==== Method ====
#
# Create a table on master and slave, insert a row on slave, and
# insert the same row on master.
#
# Create a table on master and slave, insert a row on master with
# binlogging turned off, and remove the row on master with binlogging
# turned on.
#
#
# ==== Related bugs ====
#
# BUG#31552: Replication breaks when deleting rows from out-of-sync table without PK
# BUG#31609: Not all RBR slave errors reported as errors
#
# Bug in this test case:
# BUG#37718: rpl.rpl_stm_mystery22 fails sporadically on pushbuild
#
#
# ==== Usage ====
#
# This file assumes the following:
#
# - The test language variable $slave_is_idempotent is set to 1 if the
# slave is expected to stop on duplicate key errors (i.e., if the
# binlog is in statement mode or
# @@global.slave_exec_mode=STRICT). It is set to 0 otherwise.
#
# - Replication has been initialized by include/master-slave.inc
#
# - The test adds a suppression for the following warning:
# Slave: Can't find record in 't1' Error_code: 1032
--echo ==== Initialize ====
--echo [on master]
connection master;
CREATE TABLE t1(a INT PRIMARY KEY);
--echo [on slave]
sync_slave_with_master;
--echo ==== Test: SQL thread sees 'INSERT' of existing key ====
--echo ---- Prepare slave so that it will get duplicate key error ----
# This row will be in the way of the row inserted by master.
INSERT INTO t1 VALUES (1);
--echo ---- Insert rows on master ----
--echo [on master]
connection master;
# Insert the same row on master
INSERT INTO t1 VALUES (1);
save_master_pos;
SELECT * FROM t1;
--echo [on slave]
connection slave;
# If we are statement-logging or if slave_exec_mode=STRICT, we now
# expect to see an error on the slave. Otherwise (i.e., we are
# row-logging and slave_exec_mode=IDEMPOTENT), we expect that the
# duplicate row is ignored by the slave and replication continues.
if (`SELECT @@global.binlog_format != 'ROW' OR @@global.slave_exec_mode = 'STRICT'`) {
--echo ---- Wait until slave stops with an error ----
# Wait until the slave tries to run the query, fails with duplicate
# key error, and stops the SQL thread.
let $slave_sql_errno= 1062; # ER_DUP_ENTRY
source include/wait_for_slave_sql_error.inc;
let $err= query_get_value("SHOW SLAVE STATUS", Last_SQL_Error, 1);
--echo Last_SQL_Error = $err (expected "duplicate key" error)
SELECT * FROM t1;
--echo ---- Resolve the conflict on the slave and restart SQL thread ----
DELETE FROM t1 WHERE a = 1;
START SLAVE SQL_THREAD;
source include/wait_for_slave_sql_to_start.inc;
}
--echo ---- Sync slave and verify that there is no error ----
sync_with_master;
let $err= query_get_value("SHOW SLAVE STATUS", Last_SQL_Error, 1);
--echo Last_SQL_Error = '$err' (expected no error)
SELECT * FROM t1;
--echo ==== Test: SQL thread sees 'DELETE' of non-existing row ====
--echo ---- On master, insert two rows, the second with binlogging off ----
--echo [on master]
connection master;
DELETE FROM t1;
INSERT INTO t1 VALUES (1);
--echo [on slave]
sync_slave_with_master;
DELETE FROM t1 WHERE a = 1;
--echo ---- On master, remove the row that does not exist on slave ----
--echo [on master]
connection master;
DELETE FROM t1 WHERE a = 1;
SELECT * FROM t1;
save_master_pos;
--echo [on slave]
connection slave;
# If we are row-logging and slave_exec_mode is STRICT, we now expect
# an error since the row to delete does not exist on slave. Otherwise
# (i.e., either we are statement-logging or slave_exec_mode is
# IDEMPOTENT), the absence of the row to delete is ignored and
# replication continues.
if (`SELECT @@global.binlog_format = 'ROW' AND @@global.slave_exec_mode = 'STRICT'`) {
--echo ---- Wait until slave stops with an error ----
let $slave_sql_errno= 1032; # ER_KEY_NOT_FOUND
source include/wait_for_slave_sql_error.inc;
let $err= query_get_value("SHOW SLAVE STATUS", Last_SQL_Error, 1);
--echo Last_SQL_Error = $err (expected "can't find record" error)
SELECT * FROM t1;
--echo ---- Resolve the conflict on the slave and restart SQL thread ----
INSERT INTO t1 VALUES (1);
START SLAVE SQL_THREAD;
source include/wait_for_slave_sql_to_start.inc;
}
--echo ---- Sync slave and verify that there is no error ----
# The slave should sync ok, and SHOW SLAVE STATUS should give no
# error.
sync_with_master;
let $err= query_get_value("SHOW SLAVE STATUS", Last_SQL_Error, 1);
--echo Last_SQL_Error = $err (expected no error)
SELECT * FROM t1;
--echo ==== Clean up ====
--echo [on master]
connection master;
DROP TABLE t1;
--echo [on slave]
sync_slave_with_master;