首页 > 其他分享 >故障解析丨一次死锁问题的解决

故障解析丨一次死锁问题的解决

时间:2024-03-18 09:55:32浏览次数:14  
标签:info insert NULL apple server 故障 死锁 解析 id

背景

业务端遇到报错为"Deadlock found when trying to get lock; try restarting transaction"则表明有死锁发生

名称 配置
数据库版本 GreatSQL 8.0.26
隔离级别 Read-Commited

innodb status 日志

greatsql> show engine innodb status\G
*************************** 1. row ***************************
  Type: InnoDB
  Name: 
Status: 
=====================================
2024-01-28 16:55:38 140737023727360 INNODB MONITOR OUTPUT
=====================================
Per second averages calculated from the last 14 seconds
-----------------
BACKGROUND THREAD
-----------------
srv_master_thread loops: 41 srv_active, 0 srv_shutdown, 17830 srv_idle
srv_master_thread log flush and writes: 0
----------
SEMAPHORES
----------
-------------
RW-LATCH INFO
-------------
Total number of rw-locks 132361
OS WAIT ARRAY INFO: reservation count 11180
OS WAIT ARRAY INFO: signal count 11177
RW-shared spins 0, rounds 0, OS waits 0
RW-excl spins 0, rounds 0, OS waits 0
RW-sx spins 0, rounds 0, OS waits 0
Spin rounds per wait: 0.00 RW-shared, 0.00 RW-excl, 0.00 RW-sx
------------------------
LATEST DETECTED DEADLOCK
------------------------
2024-01-28 16:53:40 140735053358848
*** (1) TRANSACTION:
TRANSACTION 37616, ACTIVE 8 sec inserting
mysql tables in use 1, locked 1
LOCK WAIT 2 lock struct(s), heap size 1192, 1 row lock(s), undo log entries 1
MySQL thread id 16, OS thread handle 140737023432448, query id 652 127.0.0.1 root update
insert into info values (50,11)

*** (1) HOLDS THE LOCK(S):
RECORD LOCKS space id 26 page no 5 n bits 80 index uk_name of table `apple`.`info` trx id 37616 lock mode S waiting
Record lock, heap no 7 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
 0: len 4; hex 8000000b; asc     ;;
 1: len 4; hex 80000028; asc    (;;


*** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 26 page no 5 n bits 80 index uk_name of table `apple`.`info` trx id 37616 lock mode S waiting
Record lock, heap no 7 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
 0: len 4; hex 8000000b; asc     ;;
 1: len 4; hex 80000028; asc    (;;


*** (2) TRANSACTION:
TRANSACTION 37615, ACTIVE 24 sec inserting
mysql tables in use 1, locked 1
LOCK WAIT 3 lock struct(s), heap size 1192, 2 row lock(s), undo log entries 2
MySQL thread id 15, OS thread handle 140737024022272, query id 653 127.0.0.1 root update
insert into info values (60,8)

*** (2) HOLDS THE LOCK(S):
RECORD LOCKS space id 26 page no 5 n bits 80 index uk_name of table `apple`.`info` trx id 37615 lock_mode X locks rec but not gap
Record lock, heap no 7 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
 0: len 4; hex 8000000b; asc     ;;
 1: len 4; hex 80000028; asc    (;;


*** (2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 26 page no 5 n bits 80 index uk_name of table `apple`.`info` trx id 37615 lock_mode X locks gap before rec insert intention waiting
Record lock, heap no 7 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
 0: len 4; hex 8000000b; asc     ;;
 1: len 4; hex 80000028; asc    (;;

*** WE ROLL BACK TRANSACTION (1)
------------
TRANSACTIONS
------------

查看表结构

greatsql> show create table info \G
*************************** 1. row ***************************
       Table: info
Create Table: CREATE TABLE `info` (
  `id` int NOT NULL AUTO_INCREMENT,
  `name` int NOT NULL,
  PRIMARY KEY (`id`),
  UNIQUE KEY `uk_name` (`name`)
) ENGINE=InnoDB AUTO_INCREMENT=61 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci
1 row in set (0.01 sec)

梳理 innodb status 日志

  • 整理如下:
事务 T1 T2
操作 insert into info values (50,11) insert into info values (60,8)
关联的对象 表apple.info的唯一索引 uk_name 表apple.info的唯一索引 uk_name
持有的锁 lock mode S waitingheap no 7 11,40(十六进制为8,28) lock_mode X locks rec but not gapheap no 7 11,40(十六进制为8,28)
等待的锁 lock mode S waitingheap no 7 11,40(十六进制为8,28) lock_mode X locks gap before rec insert intention waitingheap no 7 11,40(十六进制为8,28)
  • 首先事务T2获取到了uk_name中记录11的 lock x,rec not not gap 锁

  • 事务T1尝试获取uk_name中记录11的lock s, next key lock,由于T2持有了记录的独占锁,因此被T1堵塞

  • 事务T2尝试获取uk_name中记录11的lock x, gap before rec,insert intention,但被堵塞

获取业务历史SQL语句

通过系统表方式

通过performance_schema.threads、performance_schema.events_statements_history、performance_schema.events_statements_history_long等系统表获取历史SQL

  • 根据GreatSQL thread id获得线程id
greatsql> select PROCESSLIST_ID,THREAD_ID,THREAD_OS_ID from  performance_schema.threads where processlist_id in (15,16);
+----------------+-----------+--------------+
| PROCESSLIST_ID | THREAD_ID | THREAD_OS_ID |
+----------------+-----------+--------------+
|             15 |        61 |         5714 |
|             16 |        62 |         5719 |
+----------------+-----------+--------------+
2 rows in set (0.00 sec)
  • 根据线程id获得线程历史SQL
greatsql> select THREAD_ID,EVENT_ID,CURRENT_SCHEMA,SQL_TEXT,MESSAGE_TEXT,EVENT_NAME,SOURCE from performance_schema.events_statements_history where thread_id in (61,62) order by THREAD_ID,EVENT_ID;
+-----------+----------+----------------+---------------------------------+--------------------------------------------------------------------+--------------------------+---------------------------------+
| THREAD_ID | EVENT_ID | CURRENT_SCHEMA | SQL_TEXT                        | MESSAGE_TEXT                                                       | EVENT_NAME               | SOURCE                          |
+-----------+----------+----------------+---------------------------------+--------------------------------------------------------------------+--------------------------+---------------------------------+
|        61 |     3762 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     3807 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     3852 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     3897 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     3942 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     3987 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     4032 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        61 |     4077 | apple          | begin                           | NULL                                                               | statement/sql/begin      | init_net_server_extension.cc:94 |
|        61 |     4100 | apple          | insert into info values (40,11) | NULL                                                               | statement/sql/insert     | init_net_server_extension.cc:94 |
|        61 |     4569 | apple          | insert into info values (60,8)  | NULL                                                               | statement/sql/insert     | init_net_server_extension.cc:94 |
|        62 |     3215 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3260 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3305 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3350 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3395 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3440 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3485 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3530 | apple          | NULL                            | NULL                                                               | statement/com/Field List | init_net_server_extension.cc:94 |
|        62 |     3575 | apple          | begin                           | NULL                                                               | statement/sql/begin      | init_net_server_extension.cc:94 |
|        62 |     3598 | apple          | insert into info values (50,11) | Deadlock found when trying to get lock; try restarting transaction | statement/sql/insert     | init_net_server_extension.cc:94 |
+-----------+----------+----------------+---------------------------------+--------------------------------------------------------------------+--------------------------+---------------------------------+
20 rows in set (0.00 sec)
  • 观察show engine innodb status中的GreatSQL thread id 16和GreatSQL thread id 15

  • 通过performance_schema.threads获取THREAD_ID

  • 通过performance_schema.events_statements_history获取THREAD_ID执行的历史SQL以及执行时间

最终可复现出如下业务SQL:

事务 T1 T2
语句 begin; begin;
语句 insert into info values (40,11);
语句 insert into info values (50,11);
语句 insert into info values (60,8);

通过解析binlog

$ mysqlbinlog -vv --base64-output=decode-rows bin.000030

SET @@SESSION.GTID_NEXT= 'e319a624-b2ce-11ee-9aac-00163e62ca8a:8696'/*!*/;
# at 10314
#240128 16:52:35 server id 1024  end_log_pos 10390 CRC32 0x59edb313         Query        thread_id=18        exec_time=0        error_code=0
SET TIMESTAMP=1706431955/*!*/;
BEGIN
/*!*/;
# at 10390
#240128 16:52:35 server id 1024  end_log_pos 10442 CRC32 0xc03dea61         Table_map: `apple`.`info` mapped to number 370
# at 10442
#240128 16:52:35 server id 1024  end_log_pos 10486 CRC32 0x670e0c66         Write_rows: table id 370 flags: STMT_END_F
### INSERT INTO `apple`.`info`
### SET
###   @1=30 /* INT meta=0 nullable=0 is_null=0 */
###   @2=30 /* INT meta=0 nullable=0 is_null=0 */
# at 10486
#240128 16:52:35 server id 1024  end_log_pos 10517 CRC32 0xab4e0d89         Xid = 598
COMMIT/*!*/;
# at 10517
#240128 19:22:12 server id 1024  end_log_pos 10596 CRC32 0x4f4cf08e         GTID        last_committed=30        sequence_number=36        rbr_only=yes        original_committed_timestamp=1706440932450590        immediate_commit_timestamp=1706440932450590 transaction_length=378
/*!50718 SET TRANSACTION ISOLATION LEVEL READ COMMITTED*//*!*/;
# original_commit_timestamp=1706440932450590 (2024-01-28 19:22:12.450590 CST)
# immediate_commit_timestamp=1706440932450590 (2024-01-28 19:22:12.450590 CST)
/*!80001 SET @@session.original_commit_timestamp=1706440932450590*//*!*/;
/*!80014 SET @@session.original_server_version=80026*//*!*/;
/*!80014 SET @@session.immediate_server_version=80026*//*!*/;
SET @@SESSION.GTID_NEXT= 'e319a624-b2ce-11ee-9aac-00163e62ca8a:8697'/*!*/;
# at 10596
#240128 16:53:16 server id 1024  end_log_pos 10672 CRC32 0xf222c003         Query        thread_id=15        exec_time=0        error_code=0
SET TIMESTAMP=1706431996/*!*/;
BEGIN
/*!*/;
# at 10672
#240128 16:53:16 server id 1024  end_log_pos 10724 CRC32 0x20cb8c86         Table_map: `apple`.`info` mapped to number 370
# at 10724
#240128 16:53:16 server id 1024  end_log_pos 10768 CRC32 0xd8f53958         Write_rows: table id 370 flags: STMT_END_F
### INSERT INTO `apple`.`info`
### SET
###   @1=40 /* INT meta=0 nullable=0 is_null=0 */
###   @2=11 /* INT meta=0 nullable=0 is_null=0 */
# at 10768
#240128 16:53:40 server id 1024  end_log_pos 10820 CRC32 0x23f22580         Table_map: `apple`.`info` mapped to number 370
# at 10820
#240128 16:53:40 server id 1024  end_log_pos 10864 CRC32 0x182ecdef         Write_rows: table id 370 flags: STMT_END_F
### INSERT INTO `apple`.`info`
### SET
###   @1=60 /* INT meta=0 nullable=0 is_null=0 */
###   @2=8 /* INT meta=0 nullable=0 is_null=0 */
# at 10864
#240128 19:22:12 server id 1024  end_log_pos 10895 CRC32 0x57fd1d3c         Xid = 650
COMMIT/*!*/;
SET @@SESSION.GTID_NEXT= 'AUTOMATIC' /* added by mysqlbinlog */ /*!*/;
DELIMITER ;
# End of log file
/*!50003 SET COMPLETION_TYPE=@OLD_COMPLETION_TYPE*/;
/*!50530 SET @@SESSION.PSEUDO_SLAVE_MODE=0*/;

根据binlog中部分SET @@SESSION.GTID_NEXT= 'e319a624-b2ce-11ee-9aac-00163e62ca8a:8697'该GTID的事务信息,可恢复T2,但T1执行的语句由于被回滚了,则不会记录到binlog,可开启general log日志获取排查

事务 T1 T2
语句 begin; begin;
语句 insert into info values (40,11);
语句 insert into info values (50,11);
语句 insert into info values (60,8);

分析死锁

  • T1、T2开启了一个事务

  • 随后T2执行了插入(40,11)的insert语句:insert into info values (40,11)

  • T1执行了插入(50,11)的insert语句:insert into info values (50,11) 进行唯一性冲突检查,尝试获取LOCK_S

  • 然后T1所在的连接会将T2中的隐式锁转换为显示锁,此时T2将获取Lock X, Rec_not_gap。由于T2的Lock X, Rec_not_gap与T1的LOCK S不兼容,因此T1被堵塞

  • 随后,T2又执行了(60,8)的insert语句:insert into info values (60,8) 由于其插入的唯一索引值是8,因此不存在主键冲突,直接执行乐观插入操作。执行乐观插入时,需要检查其它事务是否堵塞insert操作。其核心是获取待插入记录的下一个值(这里刚好是10),并获取该记录上的所有锁,与需要添加的锁判断是否存在冲突。

  • T1持有了记录11的LOCK_S锁与T2的LOCK_X、LOCK_INSERT_INTENTION不兼容,因此T2被T1堵塞

  • 死锁形成。

解决

• 适当的减少Unique索引

• 避免插入重复的值(唯一索引所在列)


Enjoy GreatSQL

标签:info,insert,NULL,apple,server,故障,死锁,解析,id
From: https://www.cnblogs.com/greatsql/p/18079727

相关文章

  • 数据解析正则re
    正则re1.为什么需要正则需求 判断一个字符串是否是手机号解决 编写一个函数,给函数一个字符串,如果是手机号则返回True,否则返回FalsedefisPhone(phone):#长度为11#全部都是数字字符#以1开头passifisPhone("15125324528"):......
  • go语言请求http接口示例 并解析json
    本例请求了天气api接口对接流程注册一个账号,对接免费实况天气接口阅读接口文档http://tianqiapi.com/index/doc?version=day请求接口解析json开发流程创建一个json.go文件需要引入的包import( "encoding/json" "fmt" "io/ioutil" "net/http")定义Wea......
  • LinkedList源码解析和设计思路
    一、继承体系LinkedList类位于java.util包中,它实现了List接口和Deque接口,LinkedList可以被当做链表、双端队列使用,并且继承自AbstractSequentialList类。在继承关系中,它的父类是AbstractSequentialList,而AbstractSequentialList又继承自AbstractList,AbstractList继承自Abs......
  • “城市绿肺诊断:集成GIS、RS、VORS模型、CCDM模型、geodetecto、GWR模型技术深入解析生
    基于GIS、RS、VORS模型、CCDM模型、geodetecto、GWR模型集成的生态系统健康的耦合协调分析城市群是一国经济发展水平的象征,也是一国经济发展到一定阶段的标志,我国城市群建设体量不断增加,将成为全球经济的核心,中国城市群的建设逐步引领全球进入到了21世纪的中国新时代。然而,高......
  • 【网络原理】TCP协议详细解析
    文章目录......
  • 【2024年5月备考新增】《软考真题分章练习(答案解析) - 5 项目进度管理(高项)》
    1、()isatechniqueforestimatingthedurationorcostofanactivityoraprojectusinghistoricaldatafromasimilaractivityorproject.A.AnalogousestimatingB.parametricestimatingC.Three-PointestimatingD.Bottomestimating【答案】A【解......
  • 【算法与数据结构】堆排序&&TOP-K问题之深入解析二叉树(三)
    文章目录......
  • 无感扩声解决方案特色解析
    需求分析 在教育,金融,安防领域。这些对声音要求比较高的领域,传统的扩声系统有着佩戴复杂,容易啸叫,声音不清晰等缺点。随着技术的不断进步,高清晰,带降噪,防啸叫,低延时的音频扩音系统逐渐成为行业的风向标。基于此,无感扩声这个概念被提到了一个新的层面并逐渐成为现实。 作为长期深......
  • NMEA 0183协议消息内容解析(HTZN TTL / HT1818Z3G5L)
    1.简单介绍​ HT1818Z3G5L(HTZNTTL)基于杭州中科微电子AT6558D芯片所打造的一款GPS+BDS北斗+GLONASS格洛纳斯卫星定位授时导航模块,该模块采用3.3V供电,串口TTL接收NMEA0183协议消息。​ NMEA0183协议的消息内容大致可分为两种,定位与时间消息、卫星状态消息,这里我们主要介绍一......
  • 深入学习 XML 解析器及 DOM 操作技术
    所有主要的浏览器都内置了一个XML解析器,用于访问和操作XMLXML解析器在访问XML文档之前,必须将其加载到XMLDOM对象中所有现代浏览器都有一个内置的XML解析器,可以将文本转换为XMLDOM对象解析文本字符串以下示例将一个文本字符串解析为XMLDOM对象,并使用JavaScript从中提取......