一、修改MySQL表varchar类型 字段长度会锁表嘛?
凭借"经验"给出回答:如果字段长度超过256个字符就会锁表。
这样的回答错误 。看看MySQL 官方文档如何介绍:
Extending VARCHAR column size
The number of length bytes(字节) required by a VARCHAR column must remain the same. For VARCHAR columns of 0 to 255 bytes in size, one length byte is required to encode the value.
For VARCHAR columns of 256 bytes in size or more, two length bytes are required.
As a result, in-place ALTER TABLE only supports increasing VARCHAR column size from 0 to 255 bytes, or from 256 bytes to a greater size.
In-place ALTER TABLE does not support increasing the size of a VARCHAR column from less than 256 bytes to a size equal to or greater than 256 bytes.
In this case, the number of required length bytes changes from 1 to 2, which is only supported by a table copy (ALGORITHM=COPY).
上面的意思如下:
字符串的字段是以字节为单位存储的,utf8 一个字符需要三个字节,utf8mb4 一个字符需要4个字节。对于小于等于255字节以内的长度可以使用一个byte 存储。大于255个字节的长度则需要使用2个byte存储。
online ddl in-place 模式(不锁表)只支持字段的字节长度从0到255之间 或者256到更大值之间变化。
如果修改字段的长度,导致字段的字节长度无法使用 1 byte表示,得使用2个byte才能表示,比如从 240 修改为 256 ,如果在默认字符集为utf8mb4的情况下,varchar(60) 修改为 varchar(64),则DDL需要以copy模式,也即会锁表,阻塞写操作。
另外就是不支持以 in-place 方式缩小字段长度:
Decreasing VARCHAR size using in-place ALTER TABLE is not supported. Decreasing VARCHAR size requires a table copy (ALGORITHM=COPY).
二、实践验证官方提供的说法
演示环境:演示的环境是MySQL5.7.22 二进制安装的主从复制。
下面演示的sql都是在主库上进行执行sql的。
测试表说明:
s_api_filter_abcpool_data表的字符集为utf8mb4,初始字段长度为20 ,80个字节,可以使用1byte表示。分别修改字符串长度为 60--->64--->128。当字段的字节数变动 跨越了256 则会锁表。
测试表结构如下:
root@tidb06 21:59: [test001]> show create table s_api_filter_abcpool_data\G
*************************** 1. row ***************************
Table: s_api_filter_abcpool_data
Create Table: CREATE TABLE `s_api_filter_abcpool_data` (
`id` bigint(20) NOT NULL AUTO_INCREMENT COMMENT '自增主键',
`bill_business_id` bigint(20) NOT NULL DEFAULT '0' COMMENT 'API主表单据ID',
`finance_core_id` bigint(20) NOT NULL DEFAULT '0' COMMENT 'Pro业务单据ID',
`bill_no` varchar(100) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '业务单据号',
`bill_type_code` int(10) NOT NULL DEFAULT '0' COMMENT 'Pro业务单据类型,对应api业务单据类型枚举',
`bill_type` varchar(64) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '单据名称',
`bill_date` varchar(128) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '',
`status` tinyint(2) NOT NULL DEFAULT '0' COMMENT '状态,0=未推送,1=已推送',
`is_red` tinyint(2) NOT NULL DEFAULT '0' COMMENT '是否红冲 0-否,1-是',
`is_wash` tinyint(2) NOT NULL DEFAULT '0' COMMENT '是否冲销 0-否,1-是',
`push_date` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '推送时间',
`order_no` varchar(64) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '订单号',
`bill_fund_type` tinyint(2) NOT NULL DEFAULT '0' COMMENT '款项类型',
`amount` decimal(10,2) NOT NULL DEFAULT '0.00' COMMENT '金额',
`customer_id` int(11) NOT NULL DEFAULT '0' COMMENT '客户ID',
`start_date` int(20) NOT NULL DEFAULT '0' COMMENT '起始单元日',
`original_start_date` int(20) NOT NULL DEFAULT '0' COMMENT '原起始单元日',
`bill_id` bigint(20) NOT NULL DEFAULT '0' COMMENT '账单ID',
`original_bill_id` bigint(20) NOT NULL DEFAULT '0' COMMENT '原账单ID',
`rule_id` int(20) NOT NULL DEFAULT '0' COMMENT '对应过滤器规则ID',
`strategy_code` tinyint(2) NOT NULL DEFAULT '0' COMMENT '策略类型:1-过滤数据中的字段值,2-过滤数据中字段的计算,3-过滤数据中的字段值判断逾期天数-原财务pro过滤池策略,4-通过判断过滤子集的数据,过滤掉父集主表数据',
`rule_name` varchar(100) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT '' COMMENT '对应过滤器名称',
`document_info` mediumtext COLLATE utf8mb4_unicode_ci COMMENT 'pro推送单据完整信息',
`completion_document_info` mediumtext COLLATE utf8mb4_unicode_ci COMMENT '补全后单据完整信息',
`create_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
`update_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '最后修改时间',
PRIMARY KEY (`id`) USING BTREE,
KEY `idx_financeCoreId` (`finance_core_id`) USING BTREE COMMENT '核心主键索引',
KEY `idx_billBusinessId` (`bill_business_id`) USING BTREE COMMENT 'api业务表单据ID索引',
KEY `idx_billNo` (`bill_no`) USING BTREE COMMENT '单据号索引',
KEY `idx_startDate` (`start_date`) USING BTREE COMMENT '起始单元日索引',
KEY `ide_bill_type_date` (`bill_type_code`,`bill_date`) USING HASH COMMENT '单据类型+操作时间',
KEY `idx_create_time` (`create_time`) USING BTREE COMMENT '创建时间索引',
KEY `idx_customer_id` (`customer_id`) USING BTREE COMMENT '客户ID索引',
KEY `idex_billId` (`bill_id`) USING BTREE COMMENT 'billID索引'
) ENGINE=InnoDB AUTO_INCREMENT=1387698785715310595 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci COMMENT='API-过滤池存储原AB池数据表'
表字符集utf8mb4 和表记录为246万行
root@tidb06 22:01: [test001]> select count(*) from s_api_filter_abcpool_data;
+----------+
| count(*) |
+----------+
| 2466560 |
+----------+
开始验证测试:
ALGORITHM=INPLACE 方式 修改字段 bill_date 从 varchar(20) 到 varchar(52)
root@tidb06 22:57: [test001]> alter table s_api_filter_abcpool_data modify bill_date varchar(52) NOT NULL DEFAULT '', ALGORITHM=INPLACE;
Query OK, 0 rows affected (3 min 7.67 sec)
Records: 0 Duplicates: 0 Warnings: 0
同时查看当前事务执行情况:
root@tidb06 22:59: [(none)]> select * from information_schema.innodb_trx\G
*************************** 1. row ***************************
trx_id: 15290
trx_state: RUNNING
trx_started: 2021-05-12 22:57:23
trx_requested_lock_id: NULL
trx_wait_started: NULL
trx_weight: 0
trx_mysql_thread_id: 1905130
trx_query: alter table s_api_filter_abcpool_data modify bill_date varchar(52) NOT NULL DEFAULT '', ALGORITHM=INPLACE
trx_operation_state: reading clustered index
trx_tables_in_use: 1
trx_tables_locked: 0
trx_lock_structs: 0
trx_lock_memory_bytes: 1136
trx_rows_locked: 0
trx_rows_modified: 0
trx_concurrency_tickets: 0
trx_isolation_level: REPEATABLE READ
trx_unique_checks: 1
trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
trx_adaptive_hash_latched: 0
trx_adaptive_hash_timeout: 0
trx_is_read_only: 0
trx_autocommit_non_locking: 0
1 row in set (0.00 sec)
发现确实没有添加锁
当前执行 SQL 的行锁数量0
trx_tables_locked: 0
事务保留的锁数量0
trx_lock_structs: 0
ALGORITHM=INPLACE 修改字段 bill_date 从 varchar(52) 到 varchar(64):
root@tidb06 23:02: [test001]> alter table s_api_filter_abcpool_data modify bill_date varchar(64) NOT NULL DEFAULT '', ALGORITHM=INPLACE;
ERROR 1846 (0A000): ALGORITHM=INPLACE is not supported. Reason: Cannot change column type INPLACE. Try ALGORITHM=COPY.
此时只能时采用ALGORITHM=copy方式来修改:此过程非常的慢
root@tidb06 23:02: [test001]> alter table s_api_filter_abcpool_data modify bill_date varchar(64) NOT NULL DEFAULT '', ALGORITHM=copy;
root@tidb06 23:17: [test001]> select * from information_schema.INNODB_TRX\G
*************************** 1. row ***************************
trx_id: 15562
trx_state: RUNNING
trx_started: 2021-05-12 23:18:59
trx_requested_lock_id: NULL
trx_wait_started: NULL
trx_weight: 6756
trx_mysql_thread_id: 1905130
trx_query: alter table s_api_filter_abcpool_data modify bill_date varchar(64) NOT NULL DEFAULT '', ALGORITHM=copy
trx_operation_state: fetching rows
trx_tables_in_use: 2
trx_tables_locked: 3
trx_lock_structs: 284
trx_lock_memory_bytes: 24784
trx_rows_locked: 6752
trx_rows_modified: 6472
trx_concurrency_tickets: 0
trx_isolation_level: REPEATABLE READ
trx_unique_checks: 1
trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
trx_adaptive_hash_latched: 0
trx_adaptive_hash_timeout: 0
trx_is_read_only: 0
trx_autocommit_non_locking: 0
1 row in set (0.00 sec)
同时查看当前事务执行情况:
当前执行 SQL 的行锁数量3
trx_tables_locked: 3
事务保留的锁数量284
trx_lock_structs: 284
把字段bill_date 扩大到varchar(64)此时查看主库进程:发现在copy to tmp table
root@tidb06 23:20: [test001]> show full processlist\G
*************************** 1. row ***************************
Id: 37601
User: testrep
Host: 172.16.0.246:60158
db: NULL
Command: Binlog Dump
Time: 936618
State: Master has sent all binlog to slave; waiting for more updates
Info: NULL
*************************** 2. row ***************************
Id: 1905130
User: root
Host: localhost
db: test001
Command: Query
Time: 1019
State: copy to tmp table
Info: alter table s_api_filter_abcpool_data modify bill_date varchar(64) NOT NULL DEFAULT '', ALGORITHM=copy
*************************** 3. row ***************************
Id: 1906899
User: root
Host: localhost
db: test001
Command: Query
Time: 0
State: starting
Info: show full processlist
3 rows in set (0.00 sec)
root@tidb06 23:03: [test001]> alter table s_api_filter_abcpool_data modify bill_date varchar(64) NOT NULL DEFAULT '', ALGORITHM=copy;
Query OK, 2466560 rows affected (22 min 32.83 sec)
Records: 2466560 Duplicates: 0 Warnings: 0
此时查看从库会造成延迟:(延迟时间和主库的执行时间差不多)
root@tidb05 23:27: [(none)]> show slave status\G
...............
...........
........
Seconds_Behind_Master: 51
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 0
Last_IO_Error:
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 172160247
Master_UUID: da9d5c3e-a8aa-11eb-9c17-00163e0eced4
Master_Info_File: mysql.slave_master_info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: copy to tmp table
Master_Retry_Count: 86400
...............
...........
........
root@tidb05 23:30: [(none)]> show full processlist;
+---------+-------------+-----------+---------+---------+--------+----------------------------------+--------------------------------------------------------------------------------------------------------+
| Id | User | Host | db | Command | Time | State | Info |
+---------+-------------+-----------+---------+---------+--------+----------------------------------+--------------------------------------------------------------------------------------------------------+
| 33 | system user | | NULL | Connect | 937180 | Waiting for master to send event | NULL |
| 34 | system user | | test001 | Connect | 1581 | copy to tmp table | alter table s_api_filter_abcpool_data modify bill_date varchar(64) NOT NULL DEFAULT '', ALGORITHM=copy |
| 1872763 | root | localhost | NULL | Query | 0 | starting | show full processlist |
+---------+-------------+-----------+---------+---------+--------+----------------------------------+--------------------------------------------------------------------------------------------------------+
3 rows in set (0.00 sec)
nline ddl in-place 模式 字节数大于256时,也是不会锁表的
验证如下:
root@tidb06 23:32: [test001]> alter table s_api_filter_abcpool_data modify bill_date varchar(128) NOT NULL DEFAULT '', ALGORITHM=inplace;
Query OK, 0 rows affected (2 min 16.39 sec)
Records: 0 Duplicates: 0 Warnings: 0
root@tidb06 23:32: [test001]> select * from information_schema.INNODB_TRX\G
*************************** 1. row ***************************
trx_id: 15684
trx_state: RUNNING
trx_started: 2021-05-12 23:32:31
trx_requested_lock_id: NULL
trx_wait_started: NULL
trx_weight: 0
trx_mysql_thread_id: 1905130
trx_query: alter table s_api_filter_abcpool_data modify bill_date varchar(128) NOT NULL DEFAULT '', ALGORITHM=inplace
trx_operation_state: reading clustered index
trx_tables_in_use: 1
trx_tables_locked: 0
trx_lock_structs: 0
trx_lock_memory_bytes: 1136
trx_rows_locked: 0
trx_rows_modified: 0
trx_concurrency_tickets: 0
trx_isolation_level: REPEATABLE READ
trx_unique_checks: 1
trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
trx_adaptive_hash_latched: 0
trx_adaptive_hash_timeout: 0
trx_is_read_only: 0
trx_autocommit_non_locking: 0
1 row in set (0.00 sec)
另外就是不支持以 ALGORITHM=inplace 方式缩小字段长度。 降低varchar长度,只能采用ALGORITHM=COPY的方式
Decreasing VARCHAR size using in-place ALTER TABLE is not supported. Decreasing VARCHAR size requires a table copy (ALGORITHM=COPY).
验证如下:
order_no varchar(64) 长度从64降低到32
不支持ALGORITHM=inplace方式:
root@tidb06 22:17: [test001]> alter table s_api_filter_abcpool_data modify order_no varchar(32) NOT NULL DEFAULT '', ALGORITHM=inplace;
ERROR 1846 (0A000): ALGORITHM=INPLACE is not supported. Reason: Cannot change column type INPLACE. Try ALGORITHM=COPY.
耗时22分钟,主库期间一直锁表:
root@tidb06 22:33: [test001]> alter table s_api_filter_abcpool_data modify order_no varchar(32) NOT NULL DEFAULT '', ALGORITHM=copy;
Query OK, 2466560 rows affected (22 min 23.97 sec)
Records: 2466560 Duplicates: 0 Warnings: 0
root@tidb06 22:34: [(none)]> select * from information_schema.INNODB_TRX\G
*************************** 1. row ***************************
trx_id: 49734
trx_state: RUNNING
trx_started: 2021-05-16 22:34:11
trx_requested_lock_id: NULL
trx_wait_started: NULL
trx_weight: 9587
trx_mysql_thread_id: 2590351
trx_query: alter table s_api_filter_abcpool_data modify order_no varchar(32) NOT NULL DEFAULT '', ALGORITHM=copy
trx_operation_state: inserting
trx_tables_in_use: 2
trx_tables_locked: 3
trx_lock_structs: 1572
trx_lock_memory_bytes: 139472
trx_rows_locked: 9583
trx_rows_modified: 8015
trx_concurrency_tickets: 0
trx_isolation_level: REPEATABLE READ
trx_unique_checks: 1
trx_foreign_key_checks: 1
trx_last_foreign_key_error: NULL
trx_adaptive_hash_latched: 0
trx_adaptive_hash_timeout: 0
trx_is_read_only: 0
trx_autocommit_non_locking: 0
1 row in set (0.00 sec)
此时从库也出现了 锁表,延迟,cp临时表
Seconds_Behind_Master: 857
Slave_SQL_Running_State: copy to tmp table
trx_tables_locked: 3
trx_lock_structs: 254
参考文档连接:
https://mp.weixin.qq.com/s/R9j58nSXgKNK94PYMB-Otg