多线程备份工具mydumper怎样用

这篇文章主要介绍了多线程备份工具mydumper怎么用，具有一定借鉴价值，感兴趣的朋友可以参考下，希望大家阅读完这篇文章之后大有收获，下面让小编带着大家一起了解一下。
 
安装
 
点击(此处)折叠或打开
 
依赖安装
debian
apt-get install libglib2.0-dev
apt-get install libmysqlclient-dev
apt-get install libghc-zlib-dev
apt-get install libpcre++-dev
centos
yum install glib2-devel mysql-devel zlib-devel pcre-devel
# wget https://launchpad.net/ubuntu/+archive/primary/+files/mydumper_0.9.1.orig.tar.gz
# tar xzvf mydumper_0.9.1.orig.tar.gz
# cd mydumper-0.9.1/
# cmake .
# make && make
 
说明
 
点击(此处)折叠或打开
 
# mydumper –help
Usage:
  mydumper [OPTION…] multi-threaded MySQL dumping
Help Options:
  -?, –help                  Show help options
Application Options:
  -B, –database (指定要导出的数据库)
  -T, –tables-list           (指定要导出的表，多个表逗号分隔，不支持正则表达式）
  -o, –outputdir (指定输出目录)
  -s, –statement-size (导出的insert语句长度指定，单位字节，默认100wziji)
  -r, –rows (即水平拆分表进行存储，单位行数，这个选项会使–chunk-filesize失效)
  -F, –chunk-filesize (即水平拆分表进行存储，单位 MB)
  -c, –compress              (压缩输出文件)
  -e, –build-empty-files (空数据的表也生成文件)
  -x, –regex (正则表达式 匹配'db.table')
  -i, –ignore-engines        (设定需要忽略的存储引擎表，多个以逗号分隔)
  -m, –no-schemas (不要导出schema库中有数据的表)
  -d, –no-data (仅仅导出表结构，不要数据)
  -G, –triggers              (导出触发器)
  -E, –events                (导出event)
  -R, –routines (导出存储过程)
  -k, –no-locks (导出数据过程中不要临时使用读锁). WARNING: This will cause inconsistent backups
  –less-locking              Minimize locking time on InnoDB tables.
  -l, –long-query-guard (定义超过多久的查询为long querY ，默认60，单位 s)
  -K, –kill-long-queries (kill掉long query)(instead of aborting)
  -D, –daemon                (使用守护进程的模式)
  -I, –snapshot-interval     (dump快照之间的间隔设置，需要带上–daemon参数启动，默认60，单位minutes)
  -L, –logfile               Log file name to use, by default stdout  is used
  –tz-utc                    SET TIME_ZONE='+00:00' at top of dump to allow dumping of TIMESTAMP data when a server has data in different time zones or data is being moved between servers with different time zones, defaults to on use –skip-tz-utc to disable.
  –skip-tz-utc
  –use-savepoints            Use savepoints to reduce metadata locking issues, needs SUPER privilege
  –success-on-1146 Not increment error count and Warning instead of Critical in case of table doesn
  –lock-all-tables           Use LOCK TABLE for all, instead of FTWRL (锁定所有表)
  -U, –updated-since         Use Update_time to dump only tables updated in the last U days
  –trx-consistency-only      Transactional consistency only (事务一致)
  -t, –threads               Number of threads to use, default 4(备份线程数，默认为4)
 -C, –compress-protocol     Use compression on the MySQL connection(连接压缩)
  -v, –verbose               Verbosity of output, 0 = silent, 1 = errors, 2 = warnings, 3 = info, default 2
  
默认连接参数
  -h, –host                  The host to connect to
  -u, –user                  Username with privileges to run the dump
  -p, –password              User password
  -P, –port                  TCP/IP port to connect to
  -S, –socket                UNIX domain socket file to use for connection
 
测试
 
点击(此处)折叠或打开
 
mydumper -u root -p xxx -B db_slave -e -o db_slave_dir
目录文件如下:
多线程备份工具mydumper怎么用
 
其中各文件说明
 
建库文件 db_slave-schema-create.sql (通用格式dbname-schema-create.sql)
 
 
点击(此处)折叠或打开
 
# cat db_slave-schema-create.sql
CREATE DATABASE `db_slave` /*!40100 DEFAULT CHARACTER SET latin1 */;
建表文件 db_slave.t_check_flag-schema.sql(通用格式dbname.tbname-schema.sql)
 
 
点击(此处)折叠或打开
 
# cat db_slave.t_check_flag-schema.sql
/*!40101 SET NAMES binary*/;
/*!40014 SET FOREIGN_KEY_CHECKS=0*/;
CREATE TABLE `t_check_flag` (
  `db_ip` char(16) NOT NULL DEFAULT '',
  `check_flag` tinyint(4) DEFAULT '0' COMMENT '1 monitor  0 not '
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
数据文件 db_slave.t_check_flag.sql(通用格式dbname.tbname.sql)
 
点击(此处)折叠或打开
 
# cat db_slave.t_check_flag.sql
/*!40101 SET NAMES binary*/;
/*!40014 SET FOREIGN_KEY_CHECKS=0*/;
/*!40103 SET TIME_ZONE='+00:00' */;
INSERT INTO `t_check_flag` VALUES
("192.168.111.129",1),
("192.168.111.130",1),
("192.168.111.26",1);
整体导出信息文件 metadata(这里记录了开始时间，结束时间，以及导出开始时的binlog信息，对于建立从库比较方便)
 
点击(此处)折叠或打开
 
# cat metadata
Started dump at: 2016-11-19 13:38:56
SHOW MASTER STATUS:
    Log: 2104.014976
    Pos: 85965870
    GTID:(null)
Finished dump at: 2016-11-19 13:38:56
 
过程描述
打开general_log可以看到如下日志
 
 
点击(此处)折叠或打开
 
148901 Connect    root@localhost on db_slave
        148901 Query    SET SESSION wait_timeout = 2147483
        148901 Query    SET SESSION net_write_timeout = 2147483
        148901 Query    SHOW PROCESSLIST
        148901 Query    FLUSH TABLES WITH READ LOCK
        148901 Query    SELECT @@tokudb_version
        148901 Query    START TRANSACTION /*!40108 WITH CONSISTENT SNAPSHOT */
        148901 Query    /*!40101 SET NAMES binary*/
        148901 Query    SHOW MASTER STATUS
        148901 Query    SELECT @@gtid_current_pos
        148901 Query    SELECT @@default_master_connection
        148901 Query    SHOW SLAVE STATUS
        148902 Connect    root@localhost on
        148902 Query    SET SESSION wait_timeout = 2147483
        148902 Query    SET SESSION TRANSACTION ISOLATION LEVEL REPEATABLE READ
        148902 Query    START TRANSACTION /*!40108 WITH CONSISTENT SNAPSHOT */
        148902 Query    /*!40103 SET TIME_ZONE='+00:00' */
        148902 Query    /*!40101 SET NAMES binary*/
        148903 Connect    root@localhost on
        148903 Query    SET SESSION wait_timeout = 2147483
        148903 Query    SET SESSION TRANSACTION ISOLATION LEVEL REPEATABLE READ
        148903 Query    START TRANSACTION /*!40108 WITH CONSISTENT SNAPSHOT */
        148903 Query    /*!40103 SET TIME_ZONE='+00:00' */
        148903 Query    /*!40101 SET NAMES binary*/
        148904 Connect    root@localhost on
        148904 Query    SET SESSION wait_timeout = 2147483
        148904 Query    SET SESSION TRANSACTION ISOLATION LEVEL REPEATABLE READ
        148904 Query    START TRANSACTION /*!40108 WITH CONSISTENT SNAPSHOT */
        148904 Query    /*!40103 SET TIME_ZONE='+00:00' */
        148904 Query    /*!40101 SET NAMES binary*/
        148905 Connect    root@localhost on
        148905 Query    SET SESSION wait_timeout = 2147483
        148905 Query    SET SESSION TRANSACTION ISOLATION LEVEL REPEATABLE READ
        148905 Query    START TRANSACTION /*!40108 WITH CONSISTENT SNAPSHOT */
        148905 Query    /*!40103 SET TIME_ZONE='+00:00' */
        148905 Query    /*!40101 SET NAMES binary*/
        148901 Init DB    db_slave
        148901 Query    SHOW TABLE STATUS
        148901 Query    SHOW CREATE DATABASE `db_slave`
        148901 Query    UNLOCK TABLES /* FTWRL */
        148902 Query    SELECT /*!40001 SQL_NO_CACHE */ * FROM `db_slave`.`t_check_flag`
        148903 Query    SELECT /*!40001 SQL_NO_CACHE */ * FROM `db_slave`.`t_master_instance`
        148901 Quit    
        148904 Query    SHOW CREATE TABLE `db_slave`.`t_check_flag`
        148905 Query    SHOW CREATE TABLE `db_slave`.`t_master_instance`
①连接数据库
②判断长查询相关，–long-query-guard和–kill-long-queries
③flush tables with read locks ，备份mysiam表完成会立即unlock tables; 对于innodb表，mydumper·使用single transaction的方式进行备份。因此mysiam表的大小及数量影响着mysiam的备份时间进而影响着业务是否只读。
④通过 -t 参数指定的线程数量进行创建，并进行工作worker子线程
⑤确定当前要导出的表，并把待导出表加入到队列中。
⑥备份完成退出(其中mysiam完成会unlock tables;)
 
总结：
①可以使用 -t 参数指定线程数，多线程同时备份速度可以提升，但是也依赖于磁盘的IO性能
②导出过程要加读锁，建议在业务低峰期进行。
③在从库备份的时候，metadata中也是记录当前从库已执行到的binlog文件以及位置
④定时任务备份可使用 -v 参数输出相应等级的信息，作为备份日志。
⑤当库过于庞大的时候，建议使用 -c 参数对结果进行压缩。
⑥单表过于庞大的时候，可以使用 -r 或者 -F 参数分解，避免单表备份时间过长(不拆的话就是 one thread 进行导出)
 
感谢你能够认真阅读完这篇文章，希望小编分享的“多线程备份工具mydumper怎么用”这篇文章对大家有帮助。