问题描述

如下图，mysql_upgrade 过程中，执行 DROP DATABASE IF EXISTS performance_schema 一直在等待 metadata lock

gdb_5e

问题排查

简单粗暴的方法

有一种简单的解决方法，把其他连接kill掉，释放 metadata lock

对于这个案例，占用元数据锁的是 Id = 107768，User = xx1 的连接

但是这种方法指标不治本，案例中占用元数据锁的连接，是一个agent服务建立的

mysql_upgrade也是程序执行，不能每次都手工kill连接，需要查明为什么占用锁

详细查明问题原因

据业务方反馈，agent服务和调用mysql_upgrade的代码和5.6也在用，没有出现问题。

怀疑是5.7引入的bug

根据上述现象，显然是agent占了metadata lock，大概率不是mysql的bug

为了说服业务方，我们继续排查是在等待什么锁

查询 performance_schema.metadata_locks

首先想到5.7的 performance_schema.metadata_locks ，很遗憾这张表里并没有记录

gdb_mysql_02

gdb 获取元数据锁信息

我们尝试使用 gdb 获取锁等待信息

​​ps aux | grep 端口号​​​，找出mysqld进程号 pid，​​pstack pid > stack.log​​

在stack.log中搜索 acquire_lock（请求mdl锁的函数），可以看出是 thread 3 在请求元数据锁

gdb_5e_03

gdb -p pid
thread 3
切换到目标线程

#0  0x0000003fe940ba5e in pthread_cond_timedwait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0
#1  0x0000000000bd3fb2 in native_cond_timedwait (this=0x7eff640e05d8, owner=0x7eff640e0540, abs_timeout=0x7effa83b2ce0, set_status_on_timeout=Unhandled dwarf expression opcode 0xf3
) 
#2  my_cond_timedwait (this=0x7eff640e05d8, owner=0x7eff640e0540, abs_timeout=0x7effa83b2ce0, set_status_on_timeout=Unhandled dwarf expression opcode 0xf3
) 
#3  inline_mysql_cond_timedwait (this=0x7eff640e05d8, owner=0x7eff640e0540, abs_timeout=0x7effa83b2ce0, set_status_on_timeout=Unhandled dwarf expression opcode 0xf3
) 
#4  MDL_wait::timed_wait (this=0x7eff640e05d8, owner=0x7eff640e0540, abs_timeout=0x7effa83b2ce0, set_status_on_timeout=Unhandled dwarf expression opcode 0xf3
) 
#5  0x0000000000bd6048 in MDL_context::acquire_lock (this=0x7eff640e05d8, mdl_request=0x7eff640aa870, lock_wait_timeout=Unhandled dwarf expression opcode 0xf3
) 

f 5
跳转到  MDL_context::acquire_lock
acquire_lock 函数参数中有 MDL_request
MDL_request::MDL_key 中有详细的锁信息

p mdl_request->key

{m_length = 34, m_db_name_length = 18, m_ptr = "\003performance_schema\000global_status", '\000' <repeats 353 times>, static m_namespace_to_wait_state_name = \{ \{m_key = 0, 
  m_name = 0x130aa9b "Waiting for global read lock", m_flags = 0}, {m_key = 0, m_name = 0x130abb0 "Waiting for tablespace metadata lock", m_flags = 0}, {m_key = 0, 
  m_name = 0x130abd8 "Waiting for schema metadata lock", m_flags = 0}, {m_key = 0, m_name = 0x130ac00 "Waiting for table metadata lock", m_flags = 0}, {m_key = 0, 
  m_name = 0x130ac20 "Waiting for stored function metadata lock", m_flags = 0}, {m_key = 0, m_name = 0x130ac50 "Waiting for stored procedure metadata lock", m_flags = 0}, {m_key = 0, 
  m_name = 0x130ac80 "Waiting for trigger metadata lock", m_flags = 0}, {m_key = 0, m_name = 0x130aca8 "Waiting for event metadata lock", m_flags = 0}, {m_key = 0, 
  m_name = 0x130aab8 "Waiting for commit lock", m_flags = 0}, {m_key = 0, m_name = 0x130aad0 "User lock", m_flags = 0}, {m_key = 0, m_name = 0x130acc8 "Waiting for locking service lock", 
  m_flags = 0} } }

上述信息可以看出，正在请求performance_schema.global_status这张表的锁

排查业务代码

和业务方确认，agent中确实执行了 “show global status” , 但是已经设置了autocommit

简化逻辑后，agent代码如下

import MySQLdb
from time import sleep
conn = MySQLdb.connect(host='47.93.49.119', port=3001, user='xx1')
conn.autocommit = True
cur=conn.cursor()
cur.execute("show global status")
while 1:
    sleep(1)

代码中确实设置了autocommit，但是并没有生效（如果执行了commit，不可能不释放元数据锁）

MySQLdb.connect 返回 Connection 类，根据上述代码，autocommit是 Connection的成员属性

class Connection(_mysql.connection):

Connection 继承自_mysql.connection，_mysql 是c语言实现的python库，查看_mysql.c

static PyMethodDef _mysql_ConnectionObject_methods[] = {
    {
        "affected_rows",
        (PyCFunction)_mysql_ConnectionObject_affected_rows,
        METH_VARARGS,
        _mysql_ConnectionObject_affected_rows__doc__
    },
    {
        "autocommit",
        (PyCFunction)_mysql_ConnectionObject_autocommit,
        METH_VARARGS,
        _mysql_ConnectionObject_autocommit__doc__
    },
    {
        "commit",
        (PyCFunction)_mysql_ConnectionObject_commit,
        METH_VARARGS,
        _mysql_ConnectionObject_commit__doc__
    },

autommit 并不是成员属性，而是一个成员方法

结论

conn.autocommit = True 强行将 autocommit 的函数指针赋值为 True，并没有真正设置autocommit

5.6中没有发现这个问题

一是 agent 中只有查询语句，不设autocommit也能返回查询结果

二是 5.6中 “show global status” 查询的是 information_shcema，5.7中是performance_schema，5.6中不会影响 drop database performance_schema

###########################################################

MDL(Meta Data LocK)的作用

在MySQL5.1及之前的版本中，如果有未提交的事务trx，当执行DROP/RENAME/ALTER TABLE RENAME操作时，不会被其他事务阻塞住。这会导致如下问题(MySQL bug#989)

master： 未提交的事务，但SQL已经完成(binlog也准备好了)，表schema发生更改，在commit的时候不会被察觉到.

slave： 在binlog里是以事务提交顺序记录的，DDL隐式提交，因此在备库先执行DDL，后执行事务trx，由于trx作用的表已经发生了改变，因此trx会执行失败。 在DDL时的主库DML压力越大，这个问题触发的可能性就越高

在5.5引入了MDL（meta data lock）锁来解决在这个问题

MDL锁的类型

metadata lock也是一种锁。每个metadata lock都会定义锁住的对象，锁的持有时间和锁的类型

属性

范围

作用

GLOBAL

全局锁

主要作用是防止DDL和写操作的过程中执行 set golbal_read_only =on 或flush tables with read lock;

commit

提交保护锁

主要作用是执行flush tables with read lock后，防止已经开始在执行的写事务提交

SCHEMA

库锁

对象

TABLE

表锁

对象

FUNCTION

函数锁

对象

PROCEDURE

存储过程锁

对象

TRIGGER

触发器锁

对象

EVENT

事件锁

对象

这些锁具有以下层级关系

gdb_元数据_04

MDL锁的简单示例

在实际工作中，最常见的MDL冲突就DDL的操作被没用提交的事务所阻塞。 我们下面通过一个具体的实例来演示DDL加MDL锁的过程。在这个实例中，利用gdb来跟踪DDL申请MDL锁的过程。

会话1:

mysql> create table ti(id int primary key, c1 int, key(c1)) engine=InnoDB  
stats_auto_recalc=default;
Query OK, 0 rows affected (0.03 sec)

mysql> insert into ti values (1,1), (2,2);
Query OK, 2 rows affected (0.03 sec)
Records: 2  Duplicates: 0  Warnings: 0

mysql> start transaction;
Query OK, 0 rows affected (0.00 sec)

mysql> select * from ti;
+----+------+
| id | c1   |
+----+------+
|  1 |    1 |
|  2 |    2 |
+----+------+
2 rows in set (0.00 sec)

再开启第二个会话,利用gdb来跟踪mysql加MDL的过程 会话2：

[root@localhost mysql]# ps -ef|grep mysql
root      3336  2390  0 06:33 pts/2    00:00:01 /u02/mysql/bin/mysqld --basedir=/u02/mysql/ --datadir=/u02/mysql/data 
--plugin-dir=/u02/mysql//lib/plugin --user=root 
--log-error=/u02/mysql/tmp/error1.log --open-files-limit=10240 
--pid-file=/u02/mysql/tmp/mysql.pid 
--socket=/u02/mysql/tmp/mysql.sock --port=3306

[root@localhost mysql]# gdb -p 3336
----在GDB设置以下断点
(gdb) b MDL_context::acquire_lock
Breakpoint 1 at 0x730cab: file /u02/mysql-server-5.6/sql/mdl.cc, line 2187.
(gdb) b lock_rec_lock
Breakpoint 2 at 0xb5ef50: file /u02/mysql-server-5.6/storage/innobase/lock/lock0lock.cc, line 2296.

(gdb) c
Continuing.....

开启第三个会话

mysql> alter table ti stats_auto_recalc=1;
这个操作被hang住

在会话2中执行下面的操作

(gdb) p mdl_request
$1 = (MDL_request *) 0x7f697d1c3bd0
(gdb) p *mdl_request
$2 = {
type = MDL_INTENTION_EXCLUSIVE, duration = MDL_STATEMENT, next_in_list = 0x7f697002a560, prev_in_list = 0x7f697d1c3df8, ticket = 0x0, key = {m_length = 3, m_db_name_length = 0,
    m_ptr = '\000' <repeats 20 times>, "0|\002p\000\000\001\000\060<\034}i\177\000\000>\240\344\000\000\000\000\000\000\t\000pi\177\000\000\000\t\000pi\177\000\000`>\034}i\177\000\000V\312\344\000\000\000\000\000\240>\034}i\177\000\000\333\361\254\000b\001\000\000\a?\000\001", '\000' <repeats 20 times>, "0|\002p\000\000\001\000\220<\034}i\177\000\000>\240\344\000\000\000\000\000\340\236\002pi\177\000\000\333\361\254\000\000\000\000\000\a?\000\001", '\000' <repeats 12 times>"\340, >\034}i\177\000\000\060|\002p\000\000\001\000\350\062\220\003\000\000\000\000\333\361\254\000\000\000\000\000$\226\363", '\000' <repeats 14 times>,
"?\034}i\177\000\000\060|\002p\000\000\001\000\000=\034}i\177\000\000>\240\344\000\000\000\000\000\000"...,
static m_namespace_to_wait_state_name = {
{m_key = 101,
        m_name = 0xf125a2 "Waiting for global read lock", m_flags = 0}, 
{m_key = 102, 
    m_name = 0xf125c0 "Waiting for schema metadata lock", m_flags = 0}, 
{m_key = 103,
        m_name = 0xf125e8 "Waiting for table metadata lock", m_flags = 0}, 
{m_key = 104, 
    m_name = 0xf12608 "Waiting for stored function metadata lock", m_flags = 0}, 
{m_key = 105,
        m_name = 0xf12638 "Waiting for stored procedure metadata lock", m_flags = 0}, 
{m_key = 106, 
    m_name = 0xf12668 "Waiting for trigger metadata lock", m_flags = 0}, 
{m_key = 107,
        m_name = 0xf12690 "Waiting for event metadata lock", m_flags = 0}, 
{m_key = 108, 
    m_name = 0xf126b0 "Waiting for commit lock", m_flags = 0}}}}
(gdb)

从上面的输出中，我只能看到申请了一个语句级别的MDL_INTENTION_EXCLUSIVE。并没有看到什么其他有意义的信息。我们继续gdb跟踪

(gdb) p *(mdl_request->next_in_list)
$3 = {type = MDL_INTENTION_EXCLUSIVE, duration = MDL_TRANSACTION, next_in_list = 0x7f697002a388, prev_in_list = 0x7f697d1c3bd8, ticket = 0x0, key = {m_length = 7, m_db_name_length = 4,
    m_ptr = "\001test\000\000\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217\217", 
static m_namespace_to_wait_state_name = {
{m_key = 101,
        m_name = 0xf125a2 "Waiting for global read lock", m_flags = 0}, 
{m_key = 102, 
    m_name = 0xf125c0 "Waiting for schema metadata lock", m_flags = 0}, 
{m_key = 103,
        m_name = 0xf125e8 "Waiting for table metadata lock", m_flags = 0}, 
{m_key = 104, 
    m_name = 0xf12608 "Waiting for stored function metadata lock", m_flags = 0}, 
{m_key = 105,
        m_name = 0xf12638 "Waiting for stored procedure metadata lock", m_flags = 0}, 
{m_key = 106, 
    m_name = 0xf12668 "Waiting for trigger metadata lock", m_flags = 0}, 
{m_key = 107,
        m_name = 0xf12690 "Waiting for event metadata lock", m_flags = 0}, 
{m_key = 108, 
    m_name = 0xf126b0 "Waiting for commit lock", m_flags = 0}}}}

从上面的输出中，我们看到了需要在test（见输出中的 m_ptr = “\001test）数据库上加一把事务级的MDL_INTENTION_EXCLUSIVE锁。它并没有告诉我们最终的MDL会落在哪个对象上。我们继续跟踪

$4 = {type = MDL_SHARED_UPGRADABLE, duration = MDL_TRANSACTION, next_in_list = 0x0, prev_in_list = 0x7f697002a568, ticket = 0x0, key = {m_length = 9, m_db_name_length = 4,
    m_ptr = "\002test\000ti", '\000' <repeats 378 times>, 
static m_namespace_to_wait_state_name = {
{m_key = 101, 
    m_name = 0xf125a2 "Waiting for global read lock", m_flags = 0}, 
{m_key = 102,
    m_name = 0xf125c0 "Waiting for schema metadata lock", m_flags = 0}, 
{m_key = 103, 
    m_name = 0xf125e8 "Waiting for table metadata lock", m_flags = 0}, 
{m_key = 104,
        m_name = 0xf12608 "Waiting for stored function metadata lock", m_flags = 0}, 
{m_key = 105, 
    m_name = 0xf12638 "Waiting for stored procedure metadata lock", m_flags = 0}, 
{m_key = 106,
        m_name = 0xf12668 "Waiting for trigger metadata lock", m_flags = 0}, 
{m_key = 107, 
    m_name = 0xf12690 "Waiting for event metadata lock", m_flags = 0}, 
{m_key = 108,
        m_name = 0xf126b0 "Waiting for commit lock", m_flags = 0}}}}

从上面的输出中，我们可以看出最终是要在test数据库的ti对象上加一把MDL_SHARED_UPGRADABLE锁。在做DDL时会先加MDL_SHARED_UPGRADABLE锁，然后升级到MDL_EXCLUSIVE锁

我来执行下面的过程 会话1

mysql> commit;
Query OK, 0 rows affected (5.51 sec)

会话2

(gdb) p *mdl_request
$5 = {type = MDL_EXCLUSIVE, duration = MDL_TRANSACTION, next_in_list = 0x20302000000, prev_in_list = 0x200000001, ticket = 0x0, key = {m_length = 9, m_db_name_length = 4,
    m_ptr = "\002test\000ti\000\000\000\000@\031\220\003\000\000\000\000\333\361\254\000\000\000\000\000\260<\034}i\177\000\000\302\362\254\000\000\000\000\000\300<\034}i\177\000\000\060|\002pi\177\000\000\320<\034}i\177\000\000\360\236\344\000\000\000\000\000\000\t\000pi\177\000\000(}\002pi\177\000\000\360<\034}i\177\000\000\234\312\344\000\000\000\000\000H\245\002pi\177\000\000\333\361\254\000\000\000\000\000\023\360\000\001", '\000' <repeats 12 times>, "`S\005pi\177\000\000\060|\002p\000\000\001\000\060=\034}i\177\000\000>\240\344\000\000\000\000\000\000\t\000pi\177\000\000\000\t\000pi\177\000\000\200=\034}i\177\000\000\231\310\344\000\000\000\000\000\240=\034}i\177\000\000l-d0t\b\000\000H\344\000\001\000\000\000\000\023\360\000\001\000\000\000\000\226"..., 
static m_namespace_to_wait_state_name = {
{m_key = 101, 
    m_name = 0xf125a2 "Waiting for global read lock", m_flags = 0}, 
{m_key = 102, 
    m_name = 0xf125c0 "Waiting for schema metadata lock", m_flags = 0}, 
{m_key = 103,
        m_name = 0xf125e8 "Waiting for table metadata lock", m_flags = 0}, 
{m_key = 104, 
    m_name = 0xf12608 "Waiting for stored function metadata lock", m_flags = 0}, 
{m_key = 105,
        m_name = 0xf12638 "Waiting for stored procedure metadata lock", m_flags = 0}, 
{m_key = 106, 
    m_name = 0xf12668 "Waiting for trigger metadata lock", m_flags = 0}, 
{m_key = 107,
        m_name = 0xf12690 "Waiting for event metadata lock", m_flags = 0}, 
{m_key = 108, 
    m_name = 0xf126b0 "Waiting for commit lock", m_flags = 0}}}}

从上面的输出中，我们看到了最终是在test.ti上申请了事务级别的MDL_EXCLUSIVE锁。

会话3

mysql> alter table ti stats_auto_recalc=1;
Query OK, 0 rows affected (22 min 58.99 sec)
Records: 0  Duplicates: 0  Warnings: 0

小结

本例只是简单的演示了，在同一个事务的不同时期加的不同的MDL的锁。MYSQL中DDL的操作不属于事务操作的范围。这就给mysql主备基于语句级别同步带来了困难。mysql主备在同步的过程中，为了保证主备结构一致性，而引入了MDL机制。为了尽可能的降低MDL带来的影响。请在业务低谷的时候，执行DDL操作。