mysql slave 备库延迟是怎么得到的
? ??在mysql的备库的监控中有一项很重要的指标:Seconds_Behind_Master,这个值是怎么得到的呢？下面从5.1.58的代码中分析一下：
? ? mysql的replication中有2个比较重要的class：Master_info(rpl_mi.h), Relay_log_info(rpl_rli.h)，他们分别对应于master,info文件和slave.info文件；很显然，Master_info是io_thread需要的，Relay_log_info是sql_thread需要的。Master_info中有一个变量 clock_diff_with_master，这个值记录着mysql的主库和备库的时间差，可以理解为主备的主机时间差。clock_diff_with_master变量的定义如下：
/*
The difference in seconds between the clock of the master and the clock of
the slave (second - first). It must be signed as it may be <0 or >0.
clock_diff_with_master is computed when the I/O thread starts; for this the
I/O thread does a SELECT UNIX_TIMESTAMP() on the master.
"how late the slave is compared to the master" is computed like this:
clock_of_slave - last_timestamp_executed_by_SQL_thread - clock_diff_with_master
*/
long clock_diff_with_master;
? ? ?这个变量的注释直接提到了Seconds_Behind_Master的计算方法：clock_of_slave – last_timestamp_executed_by_SQL_thread – clock_diff_with_master。clock_of_slave是slave的当前时间–执行show slave status的当前时间。
? ? ?先看一下clock_diff_with_master的计算：(slave.cc)。执行”start slave;“/“start slave io_thread;”后，会执行start_slave_threads来启动io thread，io thread启动后首先做的就是获取主库的mysql版本和主库的当前时间(mysql_real_query(mysql, STRING_WITH_LEN(“SELECT UNIX_TIMESTAMP()”)))，获取到主库的当前时间后，用备库的当前时间减去主库的时间，得到clock_diff_with_master。
int start_slave_threads(bool need_slave_mutex, bool wait_for_start,
Master_info* mi, const char* master_info_fname,
const char* slave_info_fname, int thread_mask)
{
...
if (thread_mask & SLAVE_IO)
error=start_slave_thread(handle_slave_io,lock_io,lock_cond_io,
cond_io,
&mi->slave_running, &mi->slave_run_id,
mi, 1); //high priority, to read the most possible
...
}
...
/**
Slave IO thread entry point.
@param arg Pointer to Master_info struct that holds information for
the IO thread.
@return Always 0.
*/
pthread_handler_t handle_slave_io(void *arg)
{
...
thd_proc_info(thd, "Checking master version");
ret= get_master_version_and_clock(mysql, mi);
...
}
...
static int get_master_version_and_clock(MYSQL* mysql, Master_info* mi)
{
...
master_res= NULL;
if (!mysql_real_query(mysql, STRING_WITH_LEN("SELECT UNIX_TIMESTAMP()")) &&
(master_res= mysql_store_result(mysql)) &&
(master_row= mysql_fetch_row(master_res)))
{
mi->clock_diff_with_master=
(long) (time((time_t*) 0) - strtoul(master_row[0], 0, 10));
}
else if (is_network_error(mysql_errno(mysql)))
{
mi->report(WARNING_LEVEL, mysql_errno(mysql),
"Get master clock failed with error: %s", mysql_error(mysql));
goto network_err;
}
else
{
mi->clock_diff_with_master= 0; /* The "most sensible" value */
sql_print_warning(""SELECT UNIX_TIMESTAMP()" failed on master, "
"do not trust column Seconds_Behind_Master of SHOW "
"SLAVE STATUS. Error: %s (%d)",
mysql_error(mysql), mysql_errno(mysql));
}
...
}
? ? ?last_timestamp_executed_by_SQL_thread则是sql slave thread最近一次执行的event的时间戳(rpl.rli.cc中的last_event_start_time变量)。而这个变量的更新则是在stmt_done()函数中完成的(rpl.rli.cc)，相关定义如下:
/*
Used by row-based replication to detect that it should not stop at
this event, but give it a chance to send more events. The time
where the last event inside a group started is stored here. If the
variable is zero, we are not in a group (but may be in a
transaction).
*/
time_t last_event_start_time;
/**
Helper function to do after statement completion.
This function is called from an event to complete the group by
either stepp
