InnoDB undo log物理结构的初始化

作者按：一直以来，未好好学习 的undo，最近刚好有点时间准备学习一下，本文通过阿里内核月报和自己看代码进行综合总结，如果有误，欢迎拍砖~

水平有限，如果有误请指出。

一直以来未对 的undo进行好好的学习，最近刚好有点时间准备学习一下，通过阿里内核月报和自己看代码的综合总结一下。本文环境：

本文描述使用如上参数的设置。

一、undo 表空间物理文件的建立

本过程调用函数_init进行，栈帧如下：

#0  srv_undo_tablespaces_init (create_new_db=true, n_conf_tablespaces=4, n_opened=0x2ef55b0)at /root/mysqlc/percona-server-locks-detail-5.7.22/storage/innobase/srv/srv0start.cc:824
#1  0x0000000001bbd7e0 in innobase_start_or_create_for_mysql () at /root/mysqlc/percona-server-locks-detail-5.7.22/storage/innobase/srv/srv0start.cc:2188
#2  0x00000000019ca74e in innobase_init (p=0x2f2a420) at /root/mysqlc/percona-server-locks-detail-5.7.22/storage/innobase/handler/ha_innodb.cc:4409
#3  0x0000000000f7ec2a in ha_initialize_handlerton (plugin=0x2fca110) at /root/mysqlc/percona-server-locks-detail-5.7.22/sql/handler.cc:871
#4  0x00000000015f9edf in plugin_initialize (plugin=0x2fca110) at /root/mysqlc/percona-server-locks-detail-5.7.22/sql/sql_plugin.cc:1252

本过程主要有如下几个步骤：

for (i = 0; create_new_db && i < n_conf_tablespaces; ++i) //n_conf_tablespaces 为innodb_undo_tablespaces的配置的个数/** Default undo tablespace size in UNIV_PAGEs count (10MB). */
const ulint SRV_UNDO_TABLESPACE_SIZE_IN_PAGES =((1024 * 1024) * 10) / UNIV_PAGE_SIZE_DEF;
...err = srv_undo_tablespace_create(name, SRV_UNDO_TABLESPACE_SIZE_IN_PAGES); //建立undo文件
...

本步骤会有一个注释如下：

/* Create the undo spaces only if we are creating a newinstance. We don't allow creating of new undo tablespacesin an existing instance (yet).  This restriction exists becausewe check in several places for SYSTEM tablespaces to be less thanthe min of user defined tablespace ids. Once we implement savingthe location of the undo tablespaces and their space ids thisrestriction will/should be lifted. */

简单的讲就是建立undo 只能在初始化实例的时候，因为space id已经固定了。

for (i = 0; i < n_undo_tablespaces; ++i) {
....
err = srv_undo_tablespace_open(name, undo_tablespace_ids[i]); //打开UNDO文件 建立 file node
...
}

for (i = 0; i < n_undo_tablespaces; ++i) {fsp_header_init( //初始化fsp header 明显 space id 已经写入undo_tablespace_ids[i],SRV_UNDO_TABLESPACE_SIZE_IN_PAGES, &mtr); //SRV_UNDO_TABLESPACE_SIZE_IN_PAGES 默认的undo大小 10MB}

其中部分代码如下：

mlog_write_ulint(header + FSP_SPACE_ID, space_id, MLOG_4BYTES, mtr);mlog_write_ulint(header + FSP_NOT_USED, 0, MLOG_4BYTES, mtr);mlog_write_ulint(header + FSP_SIZE, size, MLOG_4BYTES, mtr);mlog_write_ulint(header + FSP_FREE_LIMIT, 0, MLOG_4BYTES, mtr);mlog_write_ulint(header + FSP_SPACE_FLAGS, space->flags,MLOG_4BYTES, mtr);mlog_write_ulint(header + FSP_FRAG_N_USED, 0, MLOG_4BYTES, mtr);flst_init(header + FSP_FREE, mtr);flst_init(header + FSP_FREE_FRAG, mtr);flst_init(header + FSP_FULL_FRAG, mtr);flst_init(header + FSP_SEG_INODES_FULL, mtr);flst_init(header + FSP_SEG_INODES_FREE, mtr);

这些都是fsp的内容。

做完这个步骤只是生成了4个大小为10MB的 undo 文件，并且已经加入到文件体系，但是里面没有任何类容。

二、中 的初始化

本步骤调用 ages->进行，本步骤除了初始化 以外还会初始化其( page no 5)信息如下：

/* Create the trx sys file block in a new allocated file segment */block = fseg_create(TRX_SYS_SPACE, 0, TRX_SYS + TRX_SYS_FSEG_HEADER,mtr); //建立segmentbuf_block_dbg_add_level(block, SYNC_TRX_SYS_HEADER);ut_a(block->page.id.page_no() == TRX_SYS_PAGE_NO);page = buf_block_get_frame(block); //获取内存位置mlog_write_ulint(page + FIL_PAGE_TYPE, FIL_PAGE_TYPE_TRX_SYS, //写入block 的类型MLOG_2BYTES, mtr);.../* Start counting transaction ids from number 1 up */mach_write_to_8(sys_header + TRX_SYS_TRX_ID_STORE, 1); // 初始化TRX_SYS_TRX_ID_STORE/* Reset the rollback segment slots.  Old versions of InnoDBdefine TRX_SYS_N_RSEGS as 256 (TRX_SYS_OLD_N_RSEGS) and expectthat the whole array is initialized. */ptr = TRX_SYS_RSEGS + sys_header;len = ut_max(TRX_SYS_OLD_N_RSEGS, TRX_SYS_N_RSEGS)* TRX_SYS_RSEG_SLOT_SIZE;//TRX_SYS_OLD_N_RSEGS 为256个memset(ptr, 0xff, len); //将slot的信息的全部初始化为ffptr += len;ut_a(ptr <= page + (UNIV_PAGE_SIZE - FIL_PAGE_DATA_END));/* Initialize all of the page.  This part used to be uninitialized. */memset(ptr, 0, UNIV_PAGE_SIZE - FIL_PAGE_DATA_END + page - ptr); //将剩下的空间设置为0x00mlog_log_string(sys_header, UNIV_PAGE_SIZE - FIL_PAGE_DATA_END+ page - sys_header, mtr);/* Create the first rollback segment in the SYSTEM tablespace */slot_no = trx_sysf_rseg_find_free(mtr, false, 0);page_no = trx_rseg_header_create(TRX_SYS_SPACE, univ_page_size,ULINT_MAX, slot_no, mtr); //将第一个slot固定在ibdata中

完成了这一步过后的 block 5 就初始化完了，而且我们看到所有的 slots 都初始化完成(源码所示有256个，实际上最多只会有128个，其中0号solt固定在中)，注意这里的槽大小是ZE设置的大小为8字节，4字节space id ,4字节 page no，它们会指向 所在的位置。

/** Transaction system header */
/*------------------------------------------------------------- @{ */
#define TRX_SYS_TRX_ID_STORE    0   /*!< the maximum trx id or trxnumber moduloTRX_SYS_TRX_ID_UPDATE_MARGINwritten to a file page by anytransaction; the assignment oftransaction ids continues fromthis number rounded up byTRX_SYS_TRX_ID_UPDATE_MARGINplusTRX_SYS_TRX_ID_UPDATE_MARGINwhen the database isstarted */  //最大的事物ID，下次实例启动会加上TRX_SYS_TRX_ID_UPDATE_MARGIN启动
#define TRX_SYS_FSEG_HEADER 8   /*!< segment header for thetablespace segment the trxsystem is created into */
#define TRX_SYS_RSEGS       (8 + FSEG_HEADER_SIZE)/*!< the start of the array ofrollback segment specificationslots *///指向rollback segment header的槽
/*------------------------------------------------------------- @} */

三、进行 的初始化

调用 进行：

根据注释和代码已经是个淘汰的参数，应该用ents代替。

这两个参数默认是就是及 128 其实不用设置的。本文也用128进行讨论。

参数 ents

static MYSQL_SYSVAR_ULONG(rollback_segments, srv_rollback_segments,PLUGIN_VAR_OPCMDARG,"Number of rollback segments to use for storing undo logs.",NULL, NULL,TRX_SYS_N_RSEGS,  /* Default setting */1,            /* Minimum value */TRX_SYS_N_RSEGS, 0);  /* Maximum value */

参数

static MYSQL_SYSVAR_ULONG(undo_logs, srv_undo_logs,PLUGIN_VAR_OPCMDARG,"Number of rollback segments to use for storing undo logs. (deprecated)",NULL, innodb_undo_logs_update,TRX_SYS_N_RSEGS,  /* Default setting */1,            /* Minimum value */TRX_SYS_N_RSEGS, 0);  /* Maximum value */

就是128

下面是注释和代码

 /* Deprecate innodb_undo_logs.  But still use it if it is set tonon-default and innodb_rollback_segments is default. */if (srv_undo_logs < TRX_SYS_N_RSEGS) {ib::warn() << deprecated_undo_logs;if (srv_rollback_segments == TRX_SYS_N_RSEGS) {srv_rollback_segments = srv_undo_logs;}}

n_noredo_created = trx_sys_create_noredo_rsegs(n_tmp_rsegs); //创建 32个 临时rollback segments

我们这里不准备考虑临时

ulint   new_rsegs = n_rsegs - n_used; //eg:128 -33 = 95for (i = 0; i < new_rsegs; ++i) { //对每个rollback segment进行初始化ulint   space_id;space_id = (n_spaces == 0) ? 0: (srv_undo_space_id_start + i % n_spaces); //获取 undo space_id 采用 取模的方式循环初始化 1 2 3 4ut_ad(n_spaces == 0|| srv_is_undo_tablespace(space_id));if (trx_rseg_create(space_id, 0) != NULL)

我们能够注意到这里是i % 的取模方式为我们ces参数设置的值，因此每个 是轮序的方式分布到4个不同的undo 中的。

如上是调用te完成的。步骤大概如下：

1、建立

block = fseg_create(space, 0, TRX_RSEG + TRX_RSEG_FSEG_HEADER, mtr); //建立一个回滚段，返回段头所在的块

2、初始化和E信息

/* Initialize max size field */mlog_write_ulint(rsegf + TRX_RSEG_MAX_SIZE, max_size,MLOG_4BYTES, mtr);/* Initialize the history list */mlog_write_ulint(rsegf + TRX_RSEG_HISTORY_SIZE, 0, MLOG_4BYTES, mtr);flst_init(rsegf + TRX_RSEG_HISTORY, mtr);

3、初始化每个undo 所在的page no

 for (i = 0; i < TRX_RSEG_N_SLOTS; i++) { //TRX_RSEG_N_SLOTS 为1024 初始化每个槽 值为 4字节指向 undo segment header的page notrx_rsegf_set_nth_undo(rsegf, i, FIL_NULL, mtr);}

初始化的情况下我们看到指向的page no都是 ，说明没有分配任何实际的undo 。

4、整个 初始化完成后将space id和page no 写回到 中。

sys_header = trx_sysf_get(mtr); //获取 5号 block指针 跳过  FIL_PAGE_DATA 38U trx_sysf_rseg_set_space(sys_header, rseg_slot_no, space, mtr); //设置spacetrx_sysf_rseg_set_page_no(sys_header, rseg_slot_no, page_no, mtr); //设置 no

/* Transaction rollback segment header */
/*-------------------------------------------------------------*/
#define TRX_RSEG_MAX_SIZE   0   /* Maximum allowed size for rollbacksegment in pages */
#define TRX_RSEG_HISTORY_SIZE   4   /* Number of file pages occupiedby the logs in the history list */ //history 链表大小
#define TRX_RSEG_HISTORY    8   /* The update undo logs for committedtransactions */ //链表头base node  他们通常调用include/fut0lst.ic中的函数进行更改
#define TRX_RSEG_FSEG_HEADER    (8 + FLST_BASE_NODE_SIZE)/* Header for the file segment wherethis page is placed */
#define TRX_RSEG_UNDO_SLOTS (8 + FLST_BASE_NODE_SIZE + FSEG_HEADER_SIZE)/* Undo log segment slots */ //
/*-------------------------------------------------------------*/

作为 base node的 ，我们可以看到定义如下：

/* We define the field offsets of a base node for the list */
#define FLST_LEN    0   /* 32-bit list length field */
#define FLST_FIRST  4   /* 6-byte address of the first elementof the list; undefined if empty list */
#define FLST_LAST   (4 + FIL_ADDR_SIZE) /* 6-byte address of thelast element of the list; undefinedif empty list */#define FIL_ADDR_PAGE   0   /* first in address is the page offset */
#define FIL_ADDR_BYTE   4   /* then comes 2-byte byte offset within page*/
#endif /* !UNIV_INNOCHECKSUM */
#define FIL_ADDR_SIZE   6   /* address size is 6 bytes */

多了一个长度

到这里128 已经初始化完成，并且，每个都包含1024个 undo slots。

四、整个过程初始化完成后的分布图

为了让图更加美观和好理解，我这里使用的是ces=2的情况下作图，也就是只有2个 undo 的情况。其实4个也是同样的道理，因为 slot是轮询在表空间分配的。

undo phy5.jpg

最终，我们看到初始化完成后undo slot指向的都是，及没有指向，当实际分配的时候这些slot就会指向我们的undo 。

同时我们可以看看到底包含哪些类型块，使用自制的小工具读取如下：

./myblock undo001 -d|morecurrent read blocks is : 0 --This Block is file space header blocks!
current read blocks is : 1 --This Block is insert buffer bitmap  blocks!
current read blocks is : 2 --This Block is inode blocks!
current read blocks is : 3 --This Block is system   blocks!
current read blocks is : 4 --This Block is system   blocks!
current read blocks is : 5 --This Block is system   blocks!
current read blocks is : 6 --This Block is system   blocks!
current read blocks is : 7 --This Block is system   blocks!
current read blocks is : 8 --This Block is system   blocks!
current read blocks is : 9 --This Block is system   blocks!
current read blocks is : 10 --This Block is system   blocks!
current read blocks is : 11 --This Block is system   blocks!
current read blocks is : 12 --This Block is system   blocks!
current read blocks is : 13 --This Block is system   blocks!
current read blocks is : 14 --This Block is system   blocks!
current read blocks is : 15 --This Block is system   blocks!
current read blocks is : 16 --This Block is system   blocks!
current read blocks is : 17 --This Block is system   blocks!
current read blocks is : 18 --This Block is system   blocks!
current read blocks is : 19 --This Block is system   blocks!
current read blocks is : 20 --This Block is system   blocks!
current read blocks is : 21 --This Block is system   blocks!
current read blocks is : 22 --This Block is system   blocks!
current read blocks is : 23 --This Block is system   blocks!
current read blocks is : 24 --This Block is system   blocks!
current read blocks is : 25 --This Block is system   blocks!
current read blocks is : 26 --This Block is system   blocks!
current read blocks is : 27 --This Block is undo blocks!
current read blocks is : 28 --This Block is undo blocks!
current read blocks is : 29 --This Block is undo blocks!
current read blocks is : 30 --This Block is undo blocks!
current read blocks is : 31 --This Block is undo blocks!
current read blocks is : 32 --This Block is undo blocks!
current read blocks is : 33 --This Block is undo blocks!
current read blocks is : 34 --This Block is undo blocks!
current read blocks is : 35 --This Block is undo blocks!
current read blocks is : 36 --This Block is undo blocks!
current read blocks is : 37 --This Block is undo blocks!
current read blocks is : 38 --This Block is new allocate blocks!
current read blocks is : 39 --This Block is new allocate blocks!
current read blocks is : 40 --This Block is new allocate blocks!
current read blocks is : 41 --This Block is new allocate blocks!
current read blocks is : 42 --This Block is new allocate blocks!

这里 - 就是我们的 block。我这里当然是 4个undo 的情况，看的是undo 1。看来没有问题。分析正确。

五、总结

参考文献：

阿里内核月报：

对本文有任何疑问可扫码添加原文作者微信

加入知数堂

挑战40万+年薪！

知数堂

叶金荣与吴炳锡联合打造

领跑IT精英培训

行业资深专家强强联合，倾心定制

MySQL实战/MySQL优化//

/ SQL优化/+ELK

数门精品课程