MySQL 5.6主从复制第四部分[一些被忽视的操作细节]
1. STOP SLAVE
从服务器上负责同步的有二类线程:
1) IO thread
2) SQL thread
IO thread负责获取master上的binary log, 然后多个sql threads负责执行。
IO thread 决定了Retrieved_Gtid_Set
SQL thread 决定了Executed_Gtid_Set
由于IO thread先于SQL thread，Retrieved_Gtid_Set可能会略多于Executed_Gtid_Set。
比如:
mysql [localhost] {msandbox} (test) > SHOW slave STATUS \G
.......
.......
Retrieved_Gtid_Set: 67cd9435-7cae-11e2-aa8d-00241db92e69:1-9
Executed_Gtid_Set: 67cd9435-7cae-11e2-aa8d-00241db92e69:1-7
Auto_Position: 1
所以，在stop slave的时候，正确的操作是：
1) stop slave io_thread;
2) show slave status 确定Executed_Gtid_Set赶上了Retrieved_Gtid_Set
3) stop slave sql_thread.
2.flush tables with read lock 与 show slave status
在一台完全正常的从服务器上开一个session 1:
mysql> flush tables with read lock;
如果主服务器有更新,
在此从服务器上再开一个session2:
mysql> show slave status，将会卡住， 直到在session1中执行unlock tables。
如果show slave status也是在session 1中执行的， 那么就没办法恢复了。。。。
mysql [localhost] {msandbox} (test) > flush tables with read lock;
Query OK, 0 rows affected (0.01 sec)
//这时主服务器发生了更新操作。
mysql [localhost] {msandbox} (test) > show slave status;
卡在这里…
当然ctrl+c可以取消，
Ctrl-C — sending “KILL QUERY 1″ to server …
Ctrl-C — query aborted.
Ctrl-C — sending “KILL 1″ to server …
Ctrl-C — query aborted.
ERROR 2013 (HY000): Lost connection to MySQL server during query
mysql [localhost] {msandbox} (test) >
即使这时再unlock tables也没有用。。早已经断开连接了。。
mysql [localhost] {msandbox} (test) > unlock tables;
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect…
Connection id: 14
Current database: test
Query OK, 0 rows affected (0.01 sec)
而且现在mysqld都无法stop了…
[modify@H209 msb_5_6_10_b]$ ./stop
Warning; Aborted waiting on pid file: ‘/home/modify/sandboxes/msb_5_6_10_b/data/mysql_sandbox5612.pid’ after 190 seconds
Attempting normal termination — kill -15 10858
所以在 flush tables with read lock 之前，要先stop slave…
http://bugs.mysql.com/ id=68460
相关阅读：