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
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 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 -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
