InnoDB has a good source of information about its status which can be requested every time you need to know “what’s up” with that in your environment. The SHOW ENGINE INNODB STATUS will inform you the last x seconds of its operation, leveraging system or database administrator with the best – as possible – position about what is happening with data pages which are being manipulated in a sense of maintain those in the Buffer Pool as more as possible.

$ mysql -u  -p -e 'SHOW ENGINE INNODB STATUS\G' > file

Buffer Pool is the privileged main memory area where InnoDB will maintain all the last recently used data pages, regardless of the page’s size, in rotation, based on LRU algorithm. This area will serve well for SELECT, UPDATE and DELETE, SQL commands which will use more data from memory than that on disk. Pages will be cycling between young and old status, more used and less used, respectively…

----------------------
BUFFER POOL AND MEMORY
----------------------
Total memory allocated 79121448960; in additional pool allocated 0
Dictionary memory allocated 776119
Buffer pool size 4718590
Free buffers 4682063
Database pages 36395
Old database pages 13627
Modified db pages 23223
Pending reads 0
Pending writes: LRU 0, flush list 0, single page 0
Pages made young 672, not young 0
2.90 youngs/s, 0.00 non-youngs/s
Pages read 36066, created 329, written 323
75.09 reads/s, 1.50 creates/s, 0.00 writes/s
Buffer pool hit rate 985 / 1000, young-making rate 0 / 1000 not 0 / 1000
Pages read ahead 0.00/s, evicted without access 0.00/s, Random read ahead 5.00/s
LRU len: 36395, unzip_LRU len: 0
I/O sum[0]:cur[80], unzip sum[0]:cur[0]

As you can see above, the total allocated main memory for Buffer Pool is 79121448960, with some space for the InnoDB’s dictionary, the actual size of the buffer pool, the amount of space, that is, 4682063, what is 292629 in terms of data pages, considering 16kb pages, the amount of old pages the remains in the buffer pool and all the modified or dirty pages – those that were modified by an UPDATE, for example, and haven’t flushed to disk yet. Pending reads and writes indicates the amount of pages which were written to the buffer pool and haven’t flushed yet as the flush list and the amount in terms of pages.

A good point that called my attention was the read ahead and evictions noticed by the output above. “The read ahead request is an I/O request to prefetch multiple pages in the buffer pool asynchronously, in anticipation that these pages will be needed soon”. This will tell us how many pages were copied into the buffer pool and were evicted without being accessed anytime. I think it costs a little bit to get more pages than necessary into the buffer pool as the mechanism must discard pages which are not being accessed, even being this process asynchronous.

Recently, I’ve got very curious about the spin rounds behavior and I realized that if you have many transactions in sleep state inside InnoDB, waiting to be executed, perhaps it may be a spin round problem. The output of SHOW ENGINE INNODB STATUS will show you that…

----------
SEMAPHORES
----------
OS WAIT ARRAY INFO: reservation count 13701
--Thread 140549419812608 has waited at log0log.ic line 321 for 0.00 seconds the semaphore:
Mutex at 0x7c10f4b8 created file log0log.cc line 737, lock var 1
waiters flag 1
OS WAIT ARRAY INFO: signal count 15206
Mutex spin waits 607605, rounds 3114855, OS waits 8383
RW-shared spins 9396, rounds 101453, OS waits 1626
RW-excl spins 6569, rounds 137971, OS waits 3191
Spin rounds per wait: 5.13 mutex, 10.80 RW-shared, 21.00 RW-excl

What does it mean, so?

• Mutex spin waits 607605 is the number of times a thread tried to get a mutex and it wasn’t available, so it waited in a spin-wait;
• rounds 3114855 is the number of times threads looped in the spin-wait cycle, checking the mutex.
• OS waits 8383 is the number of times the thread gave up spin-waiting and went to sleep state instead.

In the SEMAPHORES output above we can observe a case of a fine tuning is needed to avoid context switches. It costs lots of computational resources to maintain information about the actual executing status to restore it as soon as possible. The RW-shared is high, but this is not the real problem. The real problem is happening around RW-excl which acquires locks and make the amount of rounds higher even on the OS level. The final result, 21 waits in the last five seconds.

转自:http://wagnerbianchi.com/blog/?p=376

InnoDB Status Output – Buffer Pool and Spin Rounds的更多相关文章

  1. Innodb之监控Buffer pool Load progress

    你可以使用PERFORMANCE SCHEMA中的相关信息监控BUFFER POOL状态加载进程. 1. 启用 stage/innodb/buffer pool load instrument: 2. ...

  2. MySQL InnoDB缓冲池(Buffer Pool)

    InnoDB缓冲池并不仅仅缓存索引,它还会缓存行的数据.自适应哈希索引.插入缓冲(Insert Buffer).锁,以及其他内部数据结构. InnoDB还使用缓冲池来帮助延迟写入,这样就能合并多个写入 ...

  3. 关于MySQL buffer pool的预读机制

    预读机制 两种预读算法 1.线性预读 2.随机预读 对预读的监控 一.预读机制 InnoDB在I/O的优化上有个比较重要的特性为预读,预读请求是一个i/o请求,它会异步地在缓冲池中预先回迁多个页面,预 ...

  4. 在线调整InnoDB Buffer Pool Size

    InnoDB Buffer Pool主要是用来缓存数据表和索引数据的内存区域,它的默认值为134217728字节(128MB).最大值取决于CPU架构;32位系统上的最大值为4294967295(23 ...

  5. Using SHOW PROCESSLIST and mysqladmin debug Output in Conjunction with SHOW INNODB STATUS

    When InnoDB appears hung, I know the natural reaction is to check SHOW ENGINE INNODB STATUS. In fact ...

  6. 14.6.3.5 Configuring InnoDB Buffer Pool Flushing

    14.6.3.5 Configuring InnoDB Buffer Pool Flushing InnoDB 执行某些任务在后台, 包括脏叶的刷新(那些已经发生改变的pages 但是没有写入到数据文 ...

  7. 一文了解MySQL的Buffer Pool

    摘要:Innodb 存储引擎设计了一个缓冲池(Buffer Pool),来提高数据库的读写性能. 本文分享自华为云社区<MySQL 的 Buffer Pool,终于被我搞懂了>,作者:小林 ...

  8. InnoDB Spin rounds per wait在>32位机器上可能为负

    今天发现一个系统innodb的spin rounds per wait为负,感觉很奇怪,原来是个bug: For example (output from PS but we have no patc ...

  9. [转]MySQL innodb buffer pool

    最近在对公司的 MySQL 服务器做性能优化, 一直对 innodb 的内存使用方式不是很清楚, 乘这机会做点总结. 在配置 MySQL 的时候, 一般都会需要设置 innodb_buffer_poo ...

随机推荐

  1. ajax 调用asp.net后台方法

    ajax 调用asp.net后台方法  这种做法有好几种,如调用xx.asxh 页面,或者直接调用xx.aspx也面,在page_Load中进行一些判断然后调用后面的其他方法, 或者你可以直接调用we ...

  2. 打造安全的App!iOS安全系列之 HTTPS

    如何打造一个安全的App?这是每一个移动开发者必须面对的问题.在移动App开发领域,开发工程师对于安全方面的考虑普遍比较欠缺,而由于iOS平台的封闭性,遭遇到的安全问题相比于Android来说要少得多 ...

  3. html5引用公共头尾

    <embed type="text/html" src="head.html" />

  4. 如何在 Linux 中清除缓存(Cache)

              如何在 Linux 中清除缓存(Cache)            方法一: http://mp.weixin.qq.com/s?__biz=MjM5ODAzODgyMQ==&am ...

  5. POJ 1028题目描述

    Description Standard web browsers contain features to move backward and forward among the pages rece ...

  6. 如何在APICloud平台使用腾讯X5引擎

    目前APICloud与腾讯X5引擎已经达成全方位的深度合作,APICloud在多个产品线深度集成X5引擎,广大APICloud开发者们即日起可通过以下几方面在你的APP中使用X5引擎,享受X5引擎带来 ...

  7. Map小记

    Corelocation地理定位Map Kit地图展示 在移动互联网的时代,移动app能解决用户的很多生活琐事,比如 导航:去任意陌生的地方 周边找餐馆找酒店,找银行,找电影院在上述应用中都用到了地图 ...

  8. Nuget程序包 使用log4net

    Nuget程序包不用细介绍,网上资源很多,有个项目使用了log4net,为项目打log,功能很多,足够一般使用. 使用时候需要在配置文件中对其进行相关配置,我自己的配置文件放在App.config文件 ...

  9. HBase shell

    进入命令行 ./hbase shell 查看HBase shell帮助 help 查看命令帮助 直接输入命令回撤 创建命名空间 create_namespace 'ns1' 查看命名空间 list_n ...

  10. iOS 瀑布流的Demo

    /** * 瀑布流Demo的主要代码,若想看完整的代码请到下面链接去下载 * * 链接: https://pan.baidu.com/s/1slByAHB 密码: r3q6 */ #import &l ...