Commit e8de82d3 authored by Andrei Elkin's avatar Andrei Elkin

Bug #50296 Slave reconnects earlier than the prescribed slave_net_timeout value

There was auto-reconnecting by slave earlier than a prescribed by slave_net_timeout value.
The issue happened on 64bit solaris that spotted rather incorrect casting of 
the ulong slave_net_timeout into the uint of mysql.options.read_timeout.

Notice, that there is no reason for slave_net_timeout to be of type of ulong.
Since it's primarily passed as arg to mysql_options the type can be made
as uint to avoid all conversion hassles.
That's what the fixes are made.
A "side" effect of the patch is a new value for the max of slave_net_timeout
to be the max of the unsigned int type (therefore to vary across platforms).

Note, a regression test can't be made to run reliably without making it to last over some 
20 secs. That's why it is placed in suite/large_tests.
parent 0c361758
stop slave;
drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t9;
reset master;
reset slave;
drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t9;
start slave;
set @save_general_log = @@global.general_log;
set @save_log_output = @@global.log_output;
set @save_slave_net_timeout = @@global.slave_net_timeout;
set @@global.general_log = ON;
set @@global.log_output = 'table,file';
include/stop_slave.inc
set @@global.slave_net_timeout = @@global.net_read_timeout * 2;
change master to master_host = '127.0.0.1',master_port = 13020,
master_user = 'root', master_heartbeat_period = 0;
include/start_slave.inc
include/stop_slave.inc
select event_time from (select event_time from mysql.general_log as t_1 where command_type like 'Connect' order by event_time desc limit 2) as t_2 order by event_time desc limit 1 into @ts_last;
select event_time from (select event_time from mysql.general_log as t_1 where command_type like 'Connect' order by event_time desc limit 2) as t_2 order by event_time asc limit 1 into @ts_prev;
select @result as 'Must be 1';
Must be 1
1
set @@global.general_log = @save_general_log;
set @@global.log_output = @save_log_output;
set @@global.slave_net_timeout = @save_slave_net_timeout;
#
# Testing reconnecting by slave as specified by `slave_net_timeout'
#
# Bug #50296 Slave reconnects earlier than the prescribed slave_net_timeout value
#
--source include/have_csv.inc
--source include/master-slave.inc
# save global env
connection master;
set @save_general_log = @@global.general_log;
set @save_log_output = @@global.log_output;
connection slave;
set @save_slave_net_timeout = @@global.slave_net_timeout;
connection master;
set @@global.general_log = ON;
set @@global.log_output = 'table,file';
connection slave;
--source include/stop_slave.inc
--disable_warnings
set @@global.slave_net_timeout = @@global.net_read_timeout * 2;
--enable_warnings
let $idle_time=`select @@global.slave_net_timeout * 2`;
#
# if heartbeat is disabled then reconnecting to the idle master
# should happen with `slave_net_timeout' period.
# Since it's the real time that is measured, `slave_net_timeout'
# merely guarantees that reconnecting can *not* happen earlier of a value specified.
# That is there can't an exact estimate for how many time it will happen.
#
# The following lines verify that having idle master
# for more than 2 * slave_net_timeout seconds and
# slave.net_read_timeout < slave_net_timeout
# won't cause reconnecting by the slave within at least
# slave_net_timeout interval.
eval change master to master_host = '127.0.0.1',master_port = $MASTER_MYPORT,
master_user = 'root', master_heartbeat_period = 0;
let $slave_net_timeout = `select @@global.slave_net_timeout`;
--source include/start_slave.inc
--disable_query_log
--disable_result_log
eval select 'master is idle for ', sleep($idle_time);
--enable_result_log
--enable_query_log
--source include/stop_slave.inc
# querying general-log
connection master;
# In particular the last reconnection timestamp must be greater or equal to
# the previous one + slave_net_timeout
select event_time from (select event_time from mysql.general_log as t_1 where command_type like 'Connect' order by event_time desc limit 2) as t_2 order by event_time desc limit 1 into @ts_last;
select event_time from (select event_time from mysql.general_log as t_1 where command_type like 'Connect' order by event_time desc limit 2) as t_2 order by event_time asc limit 1 into @ts_prev;
--disable_query_log
eval select time_to_sec(@ts_last) - $slave_net_timeout >= time_to_sec(@ts_prev) into @result;
--enable_query_log
select @result as 'Must be 1';
# cleanup
# restore global env
connection master;
set @@global.general_log = @save_general_log;
set @@global.log_output = @save_log_output;
connection slave;
set @@global.slave_net_timeout = @save_slave_net_timeout;
......@@ -1970,7 +1970,8 @@ extern ulong slow_launch_threads, slow_launch_time;
extern ulong table_cache_size, table_def_size;
extern MYSQL_PLUGIN_IMPORT ulong max_connections;
extern ulong max_connect_errors, connect_timeout;
extern ulong slave_net_timeout, slave_trans_retries;
extern ulong slave_trans_retries;
extern uint slave_net_timeout;
extern ulong what_to_log,flush_time;
extern ulong query_buff_size;
extern ulong max_prepared_stmt_count, prepared_stmt_count;
......
......@@ -453,7 +453,8 @@ ulong table_cache_size, table_def_size;
ulong what_to_log;
ulong query_buff_size, slow_launch_time, slave_open_temp_tables;
ulong open_files_limit, max_binlog_size, max_relay_log_size;
ulong slave_net_timeout, slave_trans_retries;
ulong slave_trans_retries;
uint slave_net_timeout;
uint slave_exec_mode_options;
ulonglong slave_type_conversions_options;
ulong thread_cache_size=0, thread_pool_size= 0;
......
......@@ -2811,7 +2811,7 @@ static Sys_var_ulong Sys_slave_net_timeout(
"slave_net_timeout", "Number of seconds to wait for more data "
"from a master/slave connection before aborting the read",
GLOBAL_VAR(slave_net_timeout), CMD_LINE(REQUIRED_ARG),
VALID_RANGE(1, LONG_TIMEOUT), DEFAULT(SLAVE_NET_TIMEOUT), BLOCK_SIZE(1),
VALID_RANGE(1, UINT_MAX), DEFAULT(SLAVE_NET_TIMEOUT), BLOCK_SIZE(1),
NO_MUTEX_GUARD, NOT_IN_BINLOG, ON_CHECK(0),
ON_UPDATE(fix_slave_net_timeout));
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment