##innodb correlate
 
innodb_flush_log_at_trx_commit
value: 0,[1],2
effect: control the flush operations of redo log buffer to redo logfile on disk.
detail: 
0,redo log buffer won't be flushed to disk after transaction commit.
1,guarantee the redo log buffer will be flushed to disk immediately as soon as transaction finish commit.it's recommend value in high consistency required system.whereas,the performance will be a little bit reduced.
2,the flush time depends on the OS cache.It's possible to loss transaction when OS crashs.
 
sync_binlog
value: 0,[1]~2**32-1
effect: control the amount of binlog commit groups which will be synchronized to disk.  
detail:
0,innodb won't synchronize the binlog to disk spontaneously merely depends on the machenism of OS cache to flush them in a certain time which cannot ensure the transaction to be safe.
1,all the binlogs will be flushed to disk as soon as transactions have committed.
2 or above,flush operations base on the number of group commit taking place.
as the existence of binlog group commit in MySQL 5.6,notice that we cannot say the number is equal to the transactions.let's suppose in the scenario that there're three sessions who have started transaction in turn.they have different begin time of transaction respectively.that is,the "sequence number" is also different,but they implement commit simultaneously what means they have the same "last commit" number.thus,they can be committed together by means of group commit.eventually,we'll get three transactions but only one group commit here.if the value of the "sync_bin" is 1,the binlog will be flushed to disk only once instead of three times.in any case the value is more than 1,the flush operation is still based on the number of group commit(not number of single transaction).
 
innodb_force_recovery
value: [0],1,2,3,4,5,6
effect: control the startup mode of innodb,try to begin with 0 then increase the number every time to start successfully.
detail:
1 (SRV_FORCE_IGNORE_CORRUPT)  ignore corrupt index and data pages.
2 (SRV_FORCE_NO_BACKGROUND) prevent master thread & purge threads from running.
3 (SRV_FORCE_NO_TRX_UNDO) won't rollback transaction after recovery
4 (SRV_FORCE_NO_IBUF_MERGE) stop running insert buffer,need to recreate all the secondary indexes.
5 (SRV_FORCE_NO_UNDO_LOG_SCAN) ignore undo log,transanctions will be forcedly commited even if they havn't committed.
6 (SRV_FORCE_NO_LOG_REDO) ignore redo log,method like failover in oracle database.
look out,value of 4~6 will corrupt the datafiles,need to set innodb read only.
 
innodb_fast_shutdown
value: 0,[1],2
effect: control the operation when shutdown,value 0 is the slowest mode while shutting down.
detail:
0,a full purge and change buffer merge before shutting down.
1,skip above two operations,but still will flush dirty page due to starp checkpoint.
2,only flush log buffers to logfiles and shutdown like crash,don't flush dirty pages which will lead to quite a long time to startup netxtime to rollback the uncommited transactions.
 
innodb_io_capacity
value: 100~[200]~2**64-1(64 bit)
effect: control the gross of IOPS such as flush pages(shared by all innodb buffer pool instances).
detail:
specify it as an appropiate value according to the disk performance is better than a rather hige value.only if the IOPS is the bottleneck while flushing pages,try to increase it little by little is a practical way.usually,200~400 is recommended value.when using regular disk drive.
notice,the parameter is relevent with "innodb_flush_sync" which can lead to being ignored.
 
innodb_io_capacity_max
value: depends on the OS platform
effect: the upper limit of "innodb_io_capacity".
detail:
it will be twice as the velue of "innodb_io_capacity" and the default minimum value is "2000" if you don't specify it explicitly in "my.cnf",it cannot be set lower than "innodb_io_capacity".
 
innodb_flush_sync
value: [on],off
effect: prevent bursts IO operations caused by a high value in "innodb_io_capacity" when doing checkpoint.
detail:
it may result in an inefficient value which you've set in "innodb_io_capacity",disable it can be a workaround,but the influence may occur while doing checkpoint(especially there're huge amount of dirty pages to be flushed).
 
to be continueously updated...

MySQL常用参数说明(持续更新)的更多相关文章

  1. mysql日常笔记(持续更新)

    常用场景 sql_mode问题:http://blog.csdn.net/ccccalculator/article/details/70432123 连续日期补全/数据补零操作 在不使用存储过程和函 ...

  2. MySQL问题总结(持续更新)

    CHAR和VARCHAR的区别 存储方式和检索方式不同: 1.CHAR固定长度字符类型.CHAR存储定长数据,CHAR字段上的索引效率高,比如定义char(10),那么不论你存储的数据是否达到了10个 ...

  3. mysql问题汇总——持续更新

    1.this is incompatible with sql_mode=only_full_group_by set @@sql_mode='STRICT_TRANS_TABLES,NO_ZERO_ ...

  4. MySql报错(持续更新)

    目录 MySql报错 1. 重复键报错1062- duplicate entry '0' for key 'xxx' 1.1 报错场景 1.2 报错原因 1.3 解决方法 1.4 具体举例 2. VS ...

  5. MySQL问题汇总(持续更新)

    1.This function has none of DETERMINISTIC, NO SQL 原因: Mysql如果开启了bin-log, 我们就必须指定我们的函数是否是 1 DETERMINI ...

  6. Mysql操作笔记(持续更新)

    1.mysqldump备份导出 备份成sql mysqldump -hlocalIp -uuserName -p --opt --default-character-set=utf8 --hex-bl ...

  7. iOS 常用三方(持续更新)

    iOS 常用三方 1.ZWMSegmentController 分页控制器 https://github.com/weiming4219/ZWMSegmentController

  8. linux常用指令集-持续更新...

    0.查看所有java进程GC情况:for i in `jps|egrep -v "Jps|Launcher" |cut -d" " -f1`;do pwdx $ ...

  9. Linux常用命令——持续更新(2018-05-09)

    此命令默认是在centos环境下执行,除非特殊标明. 1.查看ip: ifconfig 2.创建指定用户并分配到某个组:创建用户user并分配到root组 useradd -g root user 3 ...

随机推荐

  1. 阿里云更懂你的数据库,免费提供DBA服务

    阿里云更懂你的数据库,免费提供DBA服务   阿里云云数据库(RDS)管理控制台近期将全面升级为云数据库管家.云数据库管家的使命是提供便捷的操作.贴心的服务.专业的处理建议,帮助用户管理好云数据库. ...

  2. Qt 之模型/视图(自定义按钮)

    https://blog.csdn.net/liang19890820/article/details/50974059 简述 衍伸前面的章节,我们对QTableView实现了数据显示.自定义排序.显 ...

  3. Linux 命令备忘(持续更新中……)

    Linux命令 grep 1. 使用grep 筛选内容,多条件筛选用 grep - E "条件1|条件2" (满足条件1或条件2的均展示) 2. grep '条件3'|grep - ...

  4. java音频播放器

    java音频播放器备份,支持wav,mp3 都是摘抄于网络,wav播放,mp3播放 播放wav版本 包: 不需要其他jar包 代码: package com; import javax.sound.s ...

  5. Log4Net使用教程

    简介 为方便跟踪程序运行情况,我们可以记录系统运行异常日志,winform和web都可以通过继承异常或者try来实现. 官方网站:http://logging.apache.org/log4net/ ...

  6. IP及DNS设置(Netsh)

    #根据连接状态查找使用中网卡gwmi win32_networkadapter -filter "NetConnectionStatus = 2"#根据是否配置网关查找使用中网卡$ ...

  7. Python学习---重点模块之subprocess

    subprocess是用来执行系统程序,查看系统的模块, 查看当前目录 第一种方法: import subprocess # subprocess会单独自己开辟一个线程,内部是多线程 # stdout ...

  8. Hadoop学习---Ubuntu中hadoop完全分布式安装教程

    软件版本 Hadoop版本号:hadoop-2.6.0-cdh5.7.0: VMWare版本号:VMware 9或10 Linux系统:CentOS 6.4-6.5 或Ubuntu版本号:ubuntu ...

  9. linux:相关命令以及笔记

    首先,在linux 或者 unix里没有任何的反馈就叫成功,例如: (1)命令执行成功的样子:返回了登录用户信息 (2)命令执行失败的样子:没有返回任何东西 1.按键出现状况 解决:可以使用ctrl+ ...

  10. selenium 滑动解锁(drag_and_drop_by_offset)

    from selenium import webdriver from selenium.webdriver.common.action_chains import ActionChains from ...