前往小程序,Get更优阅读体验!
立即前往
首页
学习
活动
专区
工具
TVP
发布
社区首页 >专栏 >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

作者头像
GreatSQL社区
发布2024-03-25 09:53:45
1020
发布2024-03-25 09:53:45
举报

* GreatSQL社区原创内容未经授权不得随意使用,转载请联系小编并注明来源。 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T1T2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T1T2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T1T2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)

本文参与 腾讯云自媒体分享计划,分享自微信公众号。
原始发表:2024-03-17,如有侵权请联系 cloudcommunity@tencent.com 删除

本文分享自 GreatSQL社区 微信公众号,前往查看

如有侵权,请联系 cloudcommunity@tencent.com 删除。

本文参与 腾讯云自媒体分享计划  ,欢迎热爱写作的你一起参与!

评论
登录后参与评论
0 条评论
热度
最新
推荐阅读
目录
  • * GreatSQL社区原创内容未经授权不得随意使用,转载请联系小编并注明来源。 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
相关产品与服务
云数据库 MySQL
腾讯云数据库 MySQL(TencentDB for MySQL)为用户提供安全可靠,性能卓越、易于维护的企业级云数据库服务。其具备6大企业级特性,包括企业级定制内核、企业级高可用、企业级高可靠、企业级安全、企业级扩展以及企业级智能运维。通过使用腾讯云数据库 MySQL,可实现分钟级别的数据库部署、弹性扩展以及全自动化的运维管理,不仅经济实惠,而且稳定可靠,易于运维。
领券
问题归档专栏文章快讯文章归档关键词归档开发者手册归档开发者手册 Section 归档