05 mysql innodb page
前言
最近看到了 何登成 大佬的 "深入MySQL源码 -- Step By Step" 的 pdf 呵呵 似乎是找到了一些 方向
之前对于 mysql 方面的东西, 更多的仅仅是简单的使用[业务中的各种增删改查], 以及一些面试题的背诵
这里会参照 MySQL Internals Manual 来大致的看一下 innodb 里面的 page 的存储相关, 这些是深入了解 mysql 的基础
本文内容对应的是 mysql innodb record 里面的 查看一下 rec 的数据信息, 除了 rec 之外还有那些信息呢 ?
MySQL Internals Manual - 22.2 InnoDB Page Structure
以下截图参照自 MySQL Internals Manual
总的来说 Page 拆分成 FileHeader, PageHeader, Infimum & Supremum, UserRecords, FreeSpace, PageDirectory, FileTrailer
FileHeader 里面主要是包含了当前文件的相关信息, 这里面的我们需要关注的就 FIL_PAGE_OFFSET, FIL_PAGE_PREV, FIL_PAGE_NEXT
PageHeader 里面主要包含的是当前 page 的元数据信息
PAGE_N_DIR_SLOTS : 当前 page pageDirectory 的数量
PAGE_HEAP_TOP : 当前 FreeSpace 的开始的偏移
PAGE_N_HEAP : 当前 page 记录的数量(UserRecords + Infimum + Supremum)
PAGE_FREE : 当前 page 被删除的记录数量
PAGE_FREE : 当前 page 被删除的记录占用的字节数
PAGE_LAST_INSERT : 当前 page 最后一个新增的记录的偏移
PAGE_DIRECTION : 当前 page 最后一个新增的记录的偏移
PAGE_N_DIRECTION : 当前 page 最后一个新增的记录的偏移
PAGE_N_RECS : UserRecords 记录数量
PAGE_MAX_TRX_ID : 最大的 trxId
PAGE_LEVEL : pageLevel
PAGE_INDEX_ID : pageIndexId
PAGE_BTR_SEG_LEAF : btrSegLeaf
PAGE_BTR_SEG_TOP : btrSegTop
Infimum & Supremum : Page 里面两条预制的记录 infimum 和 supremum, 分别表示 record 链表逻辑上的 head 和 tail
UserRecords : 各个用户的记录信息, 可以参见 mysql innodb record
FreeSpace : 当前 page 未使用的空间
PageDirectory : key order 排序的 稀疏 record指针 列表
FilTrailer : page 的 checksum + FileHeader 的 FIL_PAGE_LSN 后四字节
22.2.2 Example 里面介绍了一个实例的案例, 剖析一个实际的记录在内存中的数据分布情况, 以及拆解每一个字节的逻辑意义, 可以移步文档看一下, 这里就不截图了, 请自行查阅文档
Page 实际案例剖析
我们这里需要着重注意的是 PageHeader 的这部分的数据信息
PAGE_N_DIR_SLOTS : 当前 page pageDirectory 的数量
PAGE_HEAP_TOP : 当前 FreeSpace 的开始的偏移
PAGE_N_HEAP : 当前 page 记录的数量(UserRecords + Infimum + Supremum)
PAGE_FREE : 当前 page 被删除的记录数量
PAGE_FREE : 当前 page 被删除的记录占用的字节数
PAGE_LAST_INSERT : 当前 page 最后一个新增的记录的偏移
PAGE_DIRECTION : 当前 page 最后一个新增的记录的偏移
PAGE_N_DIRECTION : 当前 page 最后一个新增的记录的偏移
PAGE_N_RECS : UserRecords 记录数量
PAGE_MAX_TRX_ID : 最大的 trxId
PAGE_LEVEL : pageLevel
PAGE_INDEX_ID : pageIndexId
PAGE_BTR_SEG_LEAF : btrSegLeaf
PAGE_BTR_SEG_TOP : btrSegTop
# user 对应的数据 当前页的数据信息, 拆解
(lldb) x 0x1286cc000 -c 0x120
0x1286cc000: 10 aa fb 30 00 00 00 03 ff ff ff ff ff ff ff ff .��0....��������
0x1286cc010: 00 00 00 00 00 1a a2 7d 45 bf 00 00 00 00 00 00 ......�}E�......
0x1286cc020: 00 00 00 00 00 06 00 02 00 b9 80 04 00 00 00 00 .........�......
0x1286cc030: 00 a0 00 02 00 01 00 02 00 00 00 00 00 00 00 00 .�..............
0x1286cc040: 00 00 00 00 00 00 00 00 00 16 00 00 00 06 00 00 ................
0x1286cc050: 00 02 00 f2 00 00 00 06 00 00 00 02 00 32 01 00 ...�.........2..
0x1286cc060: 02 00 1c 69 6e 66 69 6d 75 6d 00 03 00 0b 00 00 ...infimum......
0x1286cc070: 73 75 70 72 65 6d 75 6d 05 00 00 00 10 00 21 80 supremum......!.
0x1286cc080: 00 00 01 00 00 00 00 2b 07 04 00 00 01 56 04 7b .......+.....V.{
0x1286cc090: 80 00 00 1c 6a 65 72 72 79 04 00 00 00 18 ff d0 ....jerry.....��
0x1286cc0a0: 80 00 00 02 00 00 00 00 35 04 83 00 00 01 36 01 ........5.....6.
0x1286cc0b0: 10 80 00 00 16 6c 75 63 79 00 00 00 00 00 00 00 .....lucy.......
0x1286cc0c0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x1286cc0d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x1286cc0e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x1286cc0f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x1286cc100: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x1286cc110: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
# page 的记录信息
0x1286cc000: 10 aa fb 30 00 00 00 03 ff ff ff ff ff ff ff ff .��0....��������
0x1286cc010: 00 00 00 00 00 1a a2 7d 45 bf 00 00 00 00 00 00 ......�}E�......
0x1286cc020: 00 00 00 00 00 06 00 02 00 b9 80 04 00 00 00 00 .........�......
0x1286cc030: 00 a0 00 02 00 01 00 02 00 00 00 00 00 00 00 00 .�..............
0x1286cc040: 00 00 00 00 00 00 00 00 00 16 00 00 00 06 00 00 ................
0x1286cc050: 00 02 00 f2 00 00 00 06 00 00 00 02 00 32 01 00 ...�.........2..
# FIL HEADERR : 38 bytes
FIL_PAGE_SPACE = 10 aa fb 30
FIL_PAGE_OFFSET = 00 00 00 03
FIL_PAGE_PREV = ff ff ff ff
FIL_PAGE_NEXT = ff ff ff ff
FIL_PAGE_LSN = 00 00 00 00 00 1a c4 cb
FIL_PAGE_TYPE = 45 bf
FIL_PAGE_FILE_FLUSH_LSN = 00 00 00 00 00 00 00 00
FIL_PAGE_ARCH_LOG_NO = 00 00 00 06
# PAGE_HEADER : 56 byte
PAGE_N_DIR_SLOTS = 00 02
PAGE_HEAP_TOP = 00 b9
PAGE_N_HEAP = 80 04
PAGE_FREE = 00 00
PAGE_GARBAGE = 00 00
PAGE_LAST_INSERT = 00 a0
PAGE_DIRECTION = 00 02
PAGE_N_DIRECTION = 00 01
PAGE_N_RECS = 00 02
PAGE_MAX_TRX_ID = 00 00 00 00 00 00 00 00
PAGE_LEVEL = 00 00
PAGE_INDEX_ID = 00 00 00 00 00 00 00 16
PAGE_BTR_SEG_LEAF = 00 00 00 06 00 00 00 02 00 f2
PAGE_BTR_SEG_TOP = 00 00 00 06 00 00 00 02 00 32
# Page Directory & Fil Trailer, 地址不一样是因为是不同的进程里面 inspect 的, 但是不影响数据
(lldb) x 0x12663bfe0 -c 0x20
0x12663bfe0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663bff0: 00 00 00 00 00 70 00 63 ac c3 1c c6 00 1a c4 cb .....p.c��.�..��
PAGE_N_DIR_SLOTS = [0x0070, 0x0063]
FIL_PAGE_END_LSN = 63 ac c3 1c c6 00 1a c4 cb
CHECK_SUM = ac c3 1c c6
FIL_PAGE_LSN = 00 1a c4 cb
UserRecords 的组织
从下面的 step2 的数据组织上来看, 我们这里的 record 列表物理上是需要 record 在 FreeSpace 里面申请 record 的空间, 但是逻辑 [以 next offset 来路由]来说 是根据 id 排序的
jerry -> lucy -> jerry5 -> jerry7 -> jerry8 -> jerry9
# 这里是 原有数据 jerry, lucy
# 物理上不是按照 id 进行排序的阿, 但是逻辑上来说 是按照 key order 排序的[根据 next 单链表]
# 1. 先新增了 jerry7, jerry8, jerry9
INSERT INTO `test`.`user`(`id`, `age`, `name`) VALUES (7, 22, 'jerry7');
INSERT INTO `test`.`user`(`id`, `age`, `name`) VALUES (8, 22, 'jerry8');
INSERT INTO `test`.`user`(`id`, `age`, `name`) VALUES (9, 22, 'jerry9');
# 2. 先新增了 jerry5
INSERT INTO `test`.`user`(`id`, `age`, `name`) VALUES (5, 22, 'jerry5');
# step1
(lldb) x 0x12663807f -c 0x120
0x12663807f: 80 00 00 01 00 00 00 00 2b 07 04 00 00 01 56 04 ........+.....V.
0x12663808f: 7b 80 00 00 1c 6a 65 72 72 79 04 00 00 00 18 00 {....jerry......
0x12663809f: 20 80 00 00 02 00 00 00 00 35 04 83 00 00 01 36 ........5.....6
0x1266380af: 01 10 80 00 00 16 6c 75 63 79 06 00 00 00 20 00 ......lucy.... .
0x1266380bf: 22 80 00 00 07 00 00 00 00 3d 0b 84 00 00 01 37 "........=.....7
0x1266380cf: 01 10 80 00 00 16 6a 65 72 72 79 37 06 00 00 00 ......jerry7....
0x1266380df: 28 00 22 80 00 00 08 00 00 00 00 3d 0c 85 00 00 (."........=....
0x1266380ef: 01 38 01 10 80 00 00 16 6a 65 72 72 79 38 06 00 .8......jerry8..
0x1266380ff: 00 00 30 ff 6c 80 00 00 09 00 00 00 00 3d 0d 86 ..0�l........=..
0x12663810f: 00 00 01 39 01 10 80 00 00 16 6a 65 72 72 79 39 ...9......jerry9
0x12663811f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663812f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663813f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663814f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663815f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663816f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663817f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663818f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
# step2
(lldb) x 0x12663807f -c 0x120
0x12663807f: 80 00 00 01 00 00 00 00 2b 07 04 00 00 01 56 04 ........+.....V.
0x12663808f: 7b 80 00 00 1c 6a 65 72 72 79 04 00 00 00 18 00 {....jerry......
0x12663809f: 86 80 00 00 02 00 00 00 00 35 04 83 00 00 01 36 .........5.....6
0x1266380af: 01 10 80 00 00 16 6c 75 63 79 06 00 00 00 20 00 ......lucy.... .
0x1266380bf: 22 80 00 00 07 00 00 00 00 3d 0b 84 00 00 01 37 "........=.....7
0x1266380cf: 01 10 80 00 00 16 6a 65 72 72 79 37 06 00 00 00 ......jerry7....
0x1266380df: 28 00 22 80 00 00 08 00 00 00 00 3d 0c 85 00 00 (."........=....
0x1266380ef: 01 38 01 10 80 00 00 16 6a 65 72 72 79 38 06 00 .8......jerry8..
0x1266380ff: 00 00 30 ff 6c 80 00 00 09 00 00 00 00 3d 0d 86 ..0�l........=..
0x12663810f: 00 00 01 39 01 10 80 00 00 16 6a 65 72 72 79 39 ...9......jerry9
0x12663811f: 06 00 00 00 38 ff 9a 80 00 00 05 00 00 00 00 3d ....8�.........=
0x12663812f: 14 8a 00 00 01 91 01 10 80 00 00 16 6a 65 72 72 ............jerr
0x12663813f: 79 35 00 00 00 00 00 00 00 00 00 00 00 00 00 00 y5..............
0x12663814f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663815f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663816f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663817f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
0x12663818f: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
参考
MySQL Internals Manual
mysql innodb record