MySQL需要很长时间才能开始

Question

我想弄明白mysql为什么要花2000万才能开始。

我正在运行5.7.17-13-log，而且我有一个大的ibdata1文件(20‘m)，即使每个表都激活了文件。它的内容如下：

# innochecksum -S /data/mysql/ibdata1
File::/data/mysql/ibdata1
================PAGE TYPE SUMMARY==============
#PAGE_COUNT     PAGE_TYPE
===============================================
  781526        Index page
 1031230        Undo log page
   11836        Inode page
   25060        Insert buffer free list page
   22645        Freshly allocated page
     114        Insert buffer bitmap
     107        System page
       1        Transaction system page
       2        File Space Header
     116        Extent descriptor page
       3        BLOB page
       0        Compressed BLOB page
       0        Other type of page
===============================================

下面是我在错误日志中看到的内容：

2017-06-05T19:08:31.307808Z 0 [Note] InnoDB: Highest supported file format is Barracuda.
2017-06-05T19:08:31.719378Z 0 [Note] InnoDB: Created parallel doublewrite buffer at /data/mysql/xb_doublewrite, size 31457280 bytes
2017-06-05T19:29:09.747406Z 0 [Note] InnoDB: Creating shared tablespace for temporary tables
2017-06-05T19:29:09.747507Z 0 [Note] InnoDB: Setting file './ibtmp1' size to 12 MB. Physically writing the file full; Please wait ...
2017-06-05T19:29:09.763021Z 0 [Note] InnoDB: File './ibtmp1' size is now 12 MB.
2017-06-05T19:29:09.869445Z 0 [Note] InnoDB: 96 redo rollback segment(s) found. 96 redo rollback segment(s) are active.
2017-06-05T19:29:09.869470Z 0 [Note] InnoDB: 32 non-redo rollback segment(s) are active.
2017-06-05T19:29:09.869574Z 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 1238576ms. The settings might not be optimal. (flushed=0, during the time.)

我看到一个提到的网页清洁剂后，采取2000万，在一个初创公司，这听起来怪怪的！

下面是mysqld上的系统输出，它显示它一直在读取ibdata1：

11706 13:01:38.019123953 0 mysqld (94026) > pread fd=4(<f>/data/mysql/ibdata1) size=16384 pos=9304113152
11707 13:01:38.019163204 0 mysqld (94026) > switch next=0 pgft_maj=1 pgft_min=4116243 vm_size=45320516 vm_rss=14609036 vm_swap=0
11709 13:01:38.019292874 0 mysqld (94026) < pread res=16384 data=.......F...E..._...n....E..............6?a..=..h................................
11712 13:01:38.019337538 0 mysqld (94026) > pread fd=4(<f>/data/mysql/ibdata1) size=16384 pos=9232695296
11713 13:01:38.019355435 0 mysqld (94026) > switch next=0 pgft_maj=1 pgft_min=4116247 vm_size=45320516 vm_rss=14609036 vm_swap=0
11715 13:01:38.019399597 0 mysqld (94026) < pread res=16384 data=.......?...9...G...n....E..............5?1.j....................................
11718 13:01:38.040605854 0 mysqld (94026) > pread fd=4(<f>/data/mysql/ibdata1) size=16384 pos=9304522752
11719 13:01:38.040644306 0 mysqld (94026) > switch next=0 pgft_maj=1 pgft_min=4116438 vm_size=45320516 vm_rss=14609560 vm_swap=0
11721 13:01:38.040774631 0 mysqld (94026) < pread res=16384 data=69....._...F...n...n....E..............*?A...u. .....w..........................
11872 13:01:38.062582012 0 mysqld (94026) > pread fd=4(<f>/data/mysql/ibdata1) size=16384 pos=9232826368
11873 13:01:38.062633600 0 mysqld (94026) > switch next=0 pgft_maj=1 pgft_min=4116541 vm_size=45320516 vm_rss=14610084 vm_swap=0
11875 13:01:38.062748273 0 mysqld (94026) < pread res=16384 data=.te=...G...?...O...n.(.gE..............X?8.i.R.......f.c........................

更新1

innodb-flush-method            = O_DIRECT
innodb-log-files-in-group      = 2
innodb-log-file-size           = 512M
innodb-flush-log-at-trx-commit = 1
innodb-file-per-table          = 1
innodb-buffer-pool-size        = 32G
innodb_log_buffer_size         = 128M
innodb_buffer_pool_instances    = 8
innodb_max_dirty_pages_pct  = 5
innodb_io_capacity      = 2000
innodb_buffer_pool_dump_at_shutdown = 1
innodb_buffer_pool_load_at_startup = 1

有什么想法吗？

谢谢,

Answer 1

原来mysql扫描所有撤销，寻找未提交的事务。

由于ibdata中有很大的撤消空间，所以我移动到使用撤销表空间。设置撤消截断选项使我的撤消表空间非常小。

现在我的分贝在2-3mn的顶部打开了！谢谢,