zoukankan      html  css  js  c++  java
  • pt_tools04 pt-archiver

    TOC

    https://www.cnblogs.com/zhoujinyi/p/9925508.html


    1. 语法

    注意:pt-archiver操作的表必须有主键。

    1.1 选项:

    Specify at least one of "--dest","--file", or "--purge".

    下面几个参数都是互斥的,只能选其一

        "--ignore"and "--replace" are mutually exclusive. 
        "--txn-size"and "--commit-each" are mutually exclusive.
        "--low-priority-insert"and "--delayed-insert" are mutually exclusive.
        "--share-lock"and "--for-update" are mutually exclusive.
        "--analyze"and "--optimize" are mutually exclusive.
        "--no-ascend"and "--no-delete" are mutually exclusive.

    1.2 常用参数

    --limit10000 每次取1000行数据用pt-archive处理,Number of rows to fetch and archive per statement.
    --txn-size 1000 设置1000行为一个事务提交一次,Number of rows pertransaction.
    --where‘id<3000‘ 设置操作条件
    --progress5000 每处理5000行输出一次处理信息
    --statistics 输出执行过程及最后的操作统计。(只要不加上--quiet,默认情况下pt-archive都会输出执行过程的)
    --charset=UTF8 指定字符集为UTF8
    --bulk-delete 批量删除source上的旧数据(例如每次1000行的批量删除操作)
    --bulk-insert 批量插入数据到dest主机 (看dest的general log发现它是通过在dest主机上LOAD DATA LOCAL INFILE插入数据的)
    --replace 将insert into 语句改成replace写入到dest库
    --sleep120 每次归档了limit个行记录后的休眠120秒(单位为秒)
    --file‘/root/test.txt‘
    --purge 删除source数据库的相关匹配记录
    --header 输入列名称到首行(和--file一起使用)
    --no-check-charset 不指定字符集
    --check-columns 检验dest和source的表结构是否一致,不一致自动拒绝执行(不加这个参数也行。默认就是执行检查的)
    --no-check-columns 不检验dest和source的表结构是否一致,不一致也执行(会导致dest上的无法与source匹配的列值被置为null或者0)
    --chekc-interval 默认1s检查一次
    --local 不把optimize或analyze操作写入到binlog里面(防止造成主从延迟巨大)
    --retries 超时或者出现死锁的话,pt-archiver进行重试的间隔(默认1s)
    --no-version-check 目前为止,发现部分pt工具对阿里云RDS操作必须加这个参数
    --analyze=ds 操作结束后,优化表空间(d表示dest,s表示source)
    

    默认情况下,pt-archiver操作结束后,不会对source、dest表执行analyze或optimize操作,因为这种操作费时间,并且需要你提前预估有足够的磁盘空间用于拷贝表。一般建议也是pt-archiver操作结束后,在业务低谷手动执行analyze table用以回收表空间。


    2. 例子

    2.1 删除老数据

    pt-archiver --txn-size=2000 --sleep=1 --limit=2000 --statistics --no-check-charset --no-version-check --purge --bulk-delete --source h=rm-bp191tt4364tbtpqb143.mysql.rds.aliyuncs.com,P=3306,u=,p='',D=esjobdata,t=job_execution_log,i=idx_complete_time --where "complete_time < '2019-04-1 00:00:00'"

    2.2 归档

    pt-archiver --txn-size=1000 --limit=1000 --sleep=1 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp13xy75phh7c60em206.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=bx_app_chatroom,t=t_chatroom_behavior_log --dest h=rm-bp13xy75phh7c60em206.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=bx_app_chatroom,t=t_chatroom_behavior_log_tmp --where "create_time < '2019-03-28 11:00:00'"
    # 37---> 存放历史数据 rm-bp1vwq001c60rav2o.mysql.rds.aliyuncs.com [ ypp_app_2016 ]  
    
    ## 反归档
    pt-archiver --txn-size=1000 --sleep=1 --limit=1000 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp1vwq001c60rav2o.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=ypp_app_2016,t=risk_hit_data_20190119 --dest h=rm-bp1p8i0ia67e44cbi.mysql.rds.aliyuncs.com,L=1,P=3306,u=inception_rw,p='',D=ypp_fengkong,t=risk_hit_data --where "play_order_create_time = '1970-01-01 00:00:00'"
    
    ## 删除控制延迟
    pt-archiver --txn-size=3000 --limit=6000 --statistics --no-check-charset --no-version-check --purge --bulk-delete --source u=,p='',h=rm-bp1p8i0ia67e44cbi.mysql.rds.aliyuncs.com,P=3306,D=ypp_fengkong,t=risk_message_record --where "created_at < '2019-03-01 00:00:00'" --max-lag=30 --check-slave-lag u=,p='',h=rr-bp18xi82q15s3ahao.mysql.rds.aliyuncs.com,P=3306

    2.3 pt-archiver 测试对表数据重复插入

    #test表3条数据, test_tmp空表
    #pt-archiver --txn-size=1000 --limit=1000 --sleep=1 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=test_ince,t=test --dest h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=test_ince,t=test_tmp --where "id < 4"
    Started at 2019-03-28T17:27:07, ended at 2019-03-28T17:27:08
    Source: D=test_ince,L=1,P=3306,h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,p=...,t=test,u=
    Dest: D=test_ince,L=1,P=3306,h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,p=...,t=test_tmp,u=
    SELECT 3
    INSERT 3
    DELETE 0
    Action Count Time Pct
    sleep 1 1.0002 98.67
    bulk_inserting 1 0.0021 0.21
    commit 2 0.0021 0.21
    select 2 0.0017 0.17
    print_bulkfile 3 0.0000 0.00
    other 0 0.0076 0.75
    
    
    #test 表3条数据, test_tmp 3条数据
    #pt-archiver --txn-size=1000 --limit=1000 --sleep=1 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=test_ince,t=test --dest h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=test_ince,t=test_tmp --where "id < 4"
    Started at 2019-03-28T17:27:40, ended at 2019-03-28T17:27:41
    Source: D=test_ince,L=1,P=3306,h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,p=...,t=test,u=
    Dest: D=test_ince,L=1,P=3306,h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,p=...,t=test_tmp,u=
    SELECT 3
    INSERT 0
    DELETE 0
    Action Count Time Pct
    sleep 1 1.0002 98.67
    bulk_inserting 1 0.0022 0.21
    select 2 0.0019 0.19
    commit 2 0.0014 0.14
    print_bulkfile 3 0.0000 0.00
    other 0 0.0080 0.79
    
    
    #mysql> delete from test_tmp where id = 2;
    #pt-archiver --txn-size=1000 --limit=1000 --sleep=1 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=test_ince,t=test --dest h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=test_ince,t=test_tmp --where "id < 4"
    Started at 2019-03-28T17:29:35, ended at 2019-03-28T17:29:36
    Source: D=test_ince,L=1,P=3306,h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,p=...,t=test,u=
    Dest: D=test_ince,L=1,P=3306,h=rm-bp1q926uimur046k5905.mysql.rds.aliyuncs.com,p=...,t=test_tmp,u=
    SELECT 3
    INSERT 1
    DELETE 0
    Action Count Time Pct
    sleep 1 1.0002 98.51
    select 2 0.0027 0.27
    bulk_inserting 1 0.0024 0.24
    commit 2 0.0023 0.22
    print_bulkfile 3 0.0000 0.00
    other 0 0.0077 0.75

    结论: 已有数据不会重复插入

    主键非递增类型

    建议强制走索引

    # 默认语句
    pt-archiver --txn-size=2000 --limit=4000 --sleep=1 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp13xy75phh7c60em206.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=bx_app_chatroom,t=t_chatroom_behavior_log_old --dest h=rm-bp13xy75phh7c60em206.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=bx_app_chatroom,t=t_chatroom_behavior_log --where "create_time >= '2019-03-28 18:45:00' "
    
    #执行的语句是全表扫描
    mysql> explain SELECT /*!40001 SQL_NO_CACHE */ `id`,`room_id`,`user_id`,`platform`,`app_version`,`bundle_id`,`content`,`type`,`status`,`memo`,`create_time`,`app_id` FROM `bx_app_chatroom`.`t_chatroom_behavior_log_old` FORCE INDEX(`PRIMARY`) WHERE (create_time >= '2019-03-28 18:45:00' ) AND ((`id` > '12881a67935244d6b4b0190bea499822')) ORDER BY `id` LIMIT 4000;
    +----+-------------+-----------------------------+-------+---------------+---------+---------+------+----------+-------------+
    | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
    +----+-------------+-----------------------------+-------+---------------+---------+---------+------+----------+-------------+
    | 1 | SIMPLE | t_chatroom_behavior_log_old | range | PRIMARY | PRIMARY | 98 | NULL | 11111765 | Using where |
    +----+-------------+-----------------------------+-------+---------------+---------+---------+------+----------+-------------+
    
    
    
    # i指定索引
    pt-archiver --txn-size=2000 --limit=4000 --sleep=1 --bulk-insert --no-delete --statistics --no-check-charset --no-version-check --source h=rm-bp13xy75phh7c60em206.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=bx_app_chatroom,t=t_chatroom_behavior_log_old,i=idx_create_time --dest h=rm-bp13xy75phh7c60em206.mysql.rds.aliyuncs.com,L=1,P=3306,u=,p='',D=bx_app_chatroom,t=t_chatroom_behavior_log --where "create_time >= '2019-03-28 18:45:00' "
    
    
    mysql> explain SELECT /*!40001 SQL_NO_CACHE */ `id`,`room_id`,`user_id`,`platform`,`app_version`,`bundle_id`,`content`,`type`,`status`,`memo`,`create_time`,`app_id` FROM `bx_app_chatroom`.`t_chatroom_behavior_log_old` FORCE INDEX(`idx_create_time`) WHERE (create_time >= '2019-03-28 18:45:00' ) AND ((`create_time` > '2019-03-28 18:55:04')) ORDER BY `create_time` LIMIT 4000;
    +----+-------------+-----------------------------+-------+-----------------+-----------------+---------+------+--------+-----------------------+
    | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
    +----+-------------+-----------------------------+-------+-----------------+-----------------+---------+------+--------+-----------------------+
    | 1 | SIMPLE | t_chatroom_behavior_log_old | range | idx_create_time | idx_create_time | 4 | NULL | 231168 | Using index condition |
    +----+-------------+-----------------------------+-------+-----------------+-----------------+---------+------+--------+-----------------------+

    <wiz_tmp_tag id="wiz-table-range-border" contenteditable="false" style="display: none;">





  • 相关阅读:
    Neurosurgeon: Collaborative Intelligence Between the Cloud and Mobile Edge
    Adversarial Attack Type I: Cheat Classifiers by Significant Changes
    Federated Optimization for Heterogeneous Networks
    On the Convergence of FedAvg on Non-IID Data
    联邦学习综述
    Federated Learning: Challenges, Methods, and Future Directions
    Hop: Heterogeneity-aware Decentralized Training
    C++文件操作
    c++: internal compiler error: Killed (program cc1plus)
    SCAFFOLD: Stochastic Controlled Averaging for On-Device Federated Learning
  • 原文地址:https://www.cnblogs.com/jesper/p/10855871.html
Copyright © 2011-2022 走看看