MySQL IO线程及相关参数调优 update (进行大量数据更新引起的性能低)
日志内容
I/O thread 0 state: waiting for completed aio requests (insert buffer thread)
I/O thread 1 state: waiting for completed aio requests (log thread)
I/O thread 2 state: waiting for completed aio requests (read thread)
I/O thread 3 state: waiting for completed aio requests (read thread)
I/O thread 4 state: waiting for completed aio requests (read thread)
I/O thread 5 state: waiting for completed aio requests (read thread)
I/O thread 6 state: waiting for completed aio requests (read thread)
I/O thread 7 state: waiting for completed aio requests (read thread)
I/O thread 8 state: waiting for completed aio requests (read thread)
I/O thread 9 state: waiting for completed aio requests (read thread)
I/O thread 10 state: waiting for completed aio requests (write thread)
I/O thread 11 state: waiting for completed aio requests (write thread)
I/O thread 12 state: waiting for completed aio requests (write thread)
I/O thread 13 state: waiting for completed aio requests (write thread)
I/O thread 14 state: waiting for completed aio requests (write thread)
I/O thread 15 state: waiting for completed aio requests (write thread)
I/O thread 16 state: waiting for completed aio requests (write thread)
I/O thread 17 state: waiting for completed aio requests (write thread)
先说优化结果,增加了IO线程后,性能提升很大,批量update 很快就完成了
一、关于一个SQL的简单的工作过程
1、工作前提描述
1、启动MySQL,在内存中分配一个大空间innodb_buffer_pool(还有log_buffer)
2、多用户线程连接MySQL,从内存分配用户工作空间(其中排序空间)
3、磁盘上有数据库文件、ib_logfile、tmp目录、undo
2、SQL的简易流程
1、DQL操作
1、首先进行内存读
2、如果buffer pool中没有所需数据,就进行物理读
3、物理读数据读入buffer pool,再返回给用户工作空间
2、DML操作(例update)
1、内存读,然后进行物理读,读取所需修改的数据行
2、从磁盘调入undo页到buffer pool中
3、修改前的数据存入undo页里,产生redo
4、修改数据行(buffer pool中数据页成脏页),产生redo
5、生成的redo先是存于用户工作空间,择机拷入log_buffer中
6、log线程不断的将log_buffer中的记录写入redo logfile中
7、修改完所有数据行,提交事务,刻意再触发一下log线程
8、待log_buffer中的相关信息都写完,响应事务提交成功
至此,日志写入磁盘,内存脏块还在buffer pool中(后台周期写入磁盘,释放buffer pool空间)。
二、影响SQL执行性能的因素,及具体看方式
1、大量物理读
mysql> show global status like 'i%read%';
| Innodb_buffer_pool_reads | 647 |
| Innodb_data_read | 48402944 |
| Innodb_data_reads | 2996 |
| Innodb_pages_read | 2949 |
| Innodb_rows_read | 1002172 |
1、Innodb_buffer_pool_reads:物理读次数
2、Innodb_data_read:物理读数据字节量
3、Innodb_data_reads:物理读IO请求次数
4、Innodb_pages_read:物理读数据页数
5、Innodb_rows_read:物理读数据行数
2、Log写性能
复制代码
mysql> show engine innodb status \G
LOG
Log sequence number 144064129 //已经生成的日志量(累计值)/单位:字节
Log flushed up to 144064129 //已经写入的日志量(累计值)
Pages flushed up to 144064129 //已经写入的脏页量(累计值)
Last checkpoint at 144064120 //检查点
0 pending log flushes, 0 pending chkp writes
92 log i/o's done, 0.00 log i/o's/second
复制代码
关于redo log的写入:
1、Innodb_os_log_written:日志刷盘的字节数,如果在commit不怎么变化的情况下,这个值出现暴增,说明系统出现大事务了(处理:kill线程,必要情况kill掉mysql进程);
2、Innodb_log_writes:日志写的次数。
3、磁盘排序
复制代码
mysql> show status like 'Sort_merge_passes';
+-------------------+-------+
| Variable_name | Value |
+-------------------+-------+
| Sort_merge_passes | 0 |
+-------------------+-------+
1 row in set (0.00 sec)
复制代码
用户所需数据,如果没有内存buffer pool中,就发生物理读;
如果需要过滤掉很多数据,就会影响物理读和内存读,因为返回很多的数据(物理读),在内存中需要过滤掉很多数据(内存读);
如果涉及到group/order by,会在用户工作空间完成排序等,如果结果集过大,用户空间过小,进行磁盘排序,Sort_merge_passes>0 ,这就很影响数据库性能了。
三、MySQL线程及其工作
MySQL的工作机制是单进程多线程:IO线程=一个log线程+四个read线程+四个write线程
复制代码
mysql> show engine innodb status \G
FILE I/O
I/O thread 0 state: waiting for i/o request (insert buffer thread)
I/O thread 1 state: waiting for i/o request (log thread)
I/O thread 2 state: waiting for i/o request (read thread)
I/O thread 3 state: waiting for i/o request (read thread)
I/O thread 4 state: waiting for i/o request (read thread)
I/O thread 5 state: waiting for i/o request (read thread)
I/O thread 6 state: waiting for i/o request (write thread)
I/O thread 7 state: waiting for i/o request (write thread)
I/O thread 8 state: waiting for i/o request (write thread)
I/O thread 9 state: waiting for i/o request (write thread)
复制代码
1、读操作:innodb_read_io_threads
1、发起者:用户线程发起读请求
2、完成者:读线程执行请求队列中的读请求操作
3、如何调整读线程的数量
复制代码
mysql> show variables like 'innodb_read_io_threads';
+------------------------+-------+
| Variable_name | Value |
+------------------------+-------+
| innodb_read_io_threads | 4 |
+------------------------+-------+
1 row in set (0.01 sec)
复制代码
默认是开启4个读线程,静态参数,修改至配置文件中
4、如何确定是否需要增加读线程的数量
查看线程的状态:I/O thread 2 state: waiting for i/o request (read thread)
2、写操作:innodb_write_io_threads
1、发起者:page_cleaner线程发起
2、完成者:写线程执行请求队列中的写请求操作
3、如何调整写线程的数量
复制代码
mysql> show variables like 'innodb_write_io_threads';
+-------------------------+-------+
| Variable_name | Value |
+-------------------------+-------+
| innodb_write_io_threads | 4 |
+-------------------------+-------+
1 row in set (0.01 sec)
复制代码
默认是开启4个写线程,静态参数,修改至配置文件中
4、如何确定是否需要增加写线程的数量
查看线程的状态:I/O thread 6 state: waiting for i/o request (write thread)
关于innodb_purge_threads:page cleaner 线程
作用:
1、负责对 undo 数据页的清空
2、数据页中 delete 标志行的清除
3、清理 innodb buffer pool,负责把内存中的脏页发起写请求,write 线程负载把脏页刷新到磁盘上。
3、日志线程
3.1、只有一个日志线程
1、是否繁忙
I/O thread 1 state: waiting for i/o request (log thread):闲
2、日志写性能
复制代码
mysql> show global status like 'Innodb_log_waits';
+------------------+-------+
| Variable_name | Value |
+------------------+-------+
| Innodb_log_waits | 0 |
+------------------+-------+
1 row in set (0.00 sec)
复制代码
如果log buffer太小,就很容易满,导致无法写入,产生日志等待。
3、日志写压力
1、每秒吞吐量
复制代码
mysql> show global status like 'Innodb_os_log_written'; #redo log写的字节数
+-----------------------+-------+
| Variable_name | Value |
+-----------------------+-------+
| Innodb_os_log_written | 57856 |
+-----------------------+-------+
1 row in set (0.01 sec)
复制代码
2、每秒写入次数
复制代码
mysql> show global status like 'Innodb_log_writes';
+-------------------+-------+
| Variable_name | Value |
+-------------------+-------+
| Innodb_log_writes | 59 |
+-------------------+-------+
1 row in set (0.01 sec)
复制代码
3.2、对于日志监控来说,三个经典参数
1、Innodb_log_waits #redo写入的等待次数
2、Innodb_log_writes #redo写入的次数
3、Innodb_os_log_written #写入redo logfile中的字节量
3.3、日志写入异常判断
复制代码
mysql> show engine innodb status \G
Pending flushes (fsync) log: 0;
mysql> show global status like 'Innodb_os_log_pending_fsyncs';
+------------------------------+-------+
| Variable_name | Value |
+------------------------------+-------+
| Innodb_os_log_pending_fsyncs | 0 |
+------------------------------+-------+
1 row in set (0.00 sec)
复制代码
1、fsync:绕过文件系统缓存,直接将内存中的数据写入存储中,实现数据真正写入可靠的介质磁盘里。(对于redo log来说,通过fsync方式写入磁盘才是可靠的保证,因为写入文件系统缓存的提交成功响应并不是真正的将redo写入磁盘的logfile中)(sync:同步)
2、pending:挂起(写不动),redo写入存储cache过程中,某种原因io繁忙,cache被占满,超时响应,就会被挂起;
3、>0就说明系统IO出现问题,=0说明is OK。
四、log buffer调整依据
log buffer:日志缓存,一般都很小,调整100M足够使用
mysql> show variables like "%log_buffer%";
+------------------------+----------+
| Variable_name | Value |
+------------------------+----------+
| innodb_log_buffer_size | 16777216 |
+------------------------+----------+
5.7默认是16M,5.6默认是8M
1、文件中(log file):每次写的时候全局都写,不会挑着捡着写
1、日志写线程每一秒redo日志缓冲刷新到重做日志文件
2、每个事务提交时会将重做日志缓冲刷新到重做日志文件
3、每当重做日志缓冲池剩余空间小于1/2时,重做日志缓冲刷新到重做日志文件。
2、在企业中往往设置50-100M,最多设置为2、300M,调整依据:
1、内存空间足够大
2、日志产生量大,系统io阻塞了,系统的io占用的是一个带宽,log_writes线程被阻塞,log buffer满了,数据库会hang住。
3、Innodb_log_waits(状态值)
The number of times that the log buffer was too small and a wait was required for it to be flushed before continuing.
java新手自学群 626070845
java/springboot/hadoop/JVM 群 4915800
Hadoop/mongodb(搭建/开发/运维)Q群481975850
GOLang Q1群:6848027
GOLang Q2群:450509103
GOLang Q3群:436173132
GOLang Q4群:141984758
GOLang Q5群:215535604
C/C++/QT群 1414577
单片机嵌入式/电子电路入门群群 306312845
MUD/LIB/交流群 391486684
Electron/koa/Nodejs/express 214737701
大前端群vue/js/ts 165150391
操作系统研发群:15375777
汇编/辅助/破解新手群:755783453
大数据 elasticsearch 群 481975850
本文版权归作者和博客园共有,欢迎转载,但未经作者同意必须保留此段声明,且在文章页面明显位置给出原文连接,否则保留追究法律责任的权利。