对PostgreSQL xmax的理解
xmax
The identity (transaction ID) of the deleting transaction, or zero for an undeleted row version. It is possible for this column to be nonzero in a visible row version. That usually indicates that the deleting transaction hasn't committed yet, or that an attempted deletion was rolled back.
http://www.postgresql.org/docs/9.1/static/ddl-system-columns.html
作一个实验:
我数据库关闭前,最后一条transaction的id是 1874。
我开一个终端A,此时终端A的当前transactionId为 1875。
[pgsql@localhost bin]$ ./psql
psql (9.1.)
Type "help" for help. pgsql=# begin;
BEGIN pgsql=# select xmin,xmax,* from tab01;
xmin | xmax | id | cd
------+------+----+----
| | |
| | |
| | |
| | |
| | |
| | |
| | |
( rows) pgsql=#
我再开一个终端B,此时,终端B的transactionId为:1876。
[pgsql@localhost bin]$ ./psql
psql (9.1.)
Type "help" for help. pgsql=# begin;
BEGIN
pgsql=# select xmin,xmax,* from tab01;
xmin | xmax | id | cd
------+------+----+----
| | |
| | |
| | |
| | |
| | |
| | |
| | |
( rows) pgsql=#
回到终端A,执行 delete 操作:
pgsql=# delete from tab01 where id=;
DELETE
pgsql=#
此时,在终端A中,已经看不到删除后的数据:
pgsql=# select xmin,xmax,* from tab01;
xmin | xmax | id | cd
------+------+----+----
| | |
| | |
| | |
| | |
| | |
| | |
( rows) pgsql=#
此时,由于终端A尚未提交,所以,可以在终端B中看到如下的情形:
pgsql=# select xmin,xmax,* from tab01;
xmin | xmax | id | cd
------+------+----+----
| | |
| | |
| | |
| | |
| | |
| | |
| | |
( rows) pgsql=#
也就是说,id为9的那条记录,其xmax为1875,表明其为 transactionid为 1875的事务所删除。
回到终端A,进行提交:
pgsql=# commit;
COMMIT
pgsql=# select xmin,xmax,* from tab01;
xmin | xmax | id | cd
------+------+----+----
| | |
| | |
| | |
| | |
| | |
| | |
( rows) pgsql=#
再回到终端B,查看:
pgsql=# select xmin,xmax,* from tab01;
xmin | xmax | id | cd
------+------+----+----
| | |
| | |
| | |
| | |
| | |
| | |
( rows) pgsql=#
让我来再进一步,看一看:
重新开两个终端:
终端A和终端B。
在终端A中:
[pgsql@localhost bin]$ ./psql
psql (9.1.)
Type "help" for help. pgsql=# select xmin,xmax,cmin,cmax,* from tab01;
xmin | xmax | cmin | cmax | id | cd
------+------+------+------+----+----
| | | | |
| | | | |
( rows) pgsql=# \q
可以看到两条由不同的事务提交所形成的记录。
然后再次使用psql: 经过update 后,自己所看到的是 xmin的变化,这时尚未提交,别的终端看到就不一样了。
终端A:
[pgsql@localhost bin]$ ./psql
psql (9.1.)
Type "help" for help. pgsql=# begin;
BEGIN
pgsql=# update tab01 set id= where cd='';
UPDATE
pgsql=# select xmin,xmax,cmin,cmax,* from tab01;
xmin | xmax | cmin | cmax | id | cd
------+------+------+------+----+----
| | | | |
| | | | |
( rows) pgsql=#
此时,进入终端B:
[pgsql@localhost bin]$ ./psql
psql (9.1.)
Type "help" for help. pgsql=# select xmin,xmax,cmin,cmax,* from tab01;
xmin | xmax | cmin | cmax | id | cd
------+------+------+------+----+----
| | | | |
| | | | |
( rows) pgsql=#
我推测,update的时候,自身事务所看到的是内存中的影像。同时它也已经提交到了物理文件上。而别的事务是从物理文件来读取的。
在A终端提交以后:
pgsql=# commit;
COMMIT
pgsql=#
B终端所看到的:
pgsql=# select xmin,xmax,cmin,cmax,* from tab01;
xmin | xmax | cmin | cmax | id | cd
------+------+------+------+----+----
| | | | |
| | | | |
( rows) pgsql=#
继续从代码上进行分析:
/*
* heap_update - replace a tuple
*
* NB: do not call this directly unless you are prepared to deal with
* concurrent-update conditions. Use simple_heap_update instead.
*
* relation - table to be modified (caller must hold suitable lock)
* otid - TID of old tuple to be replaced
* newtup - newly constructed tuple data to store
* ctid - output parameter, used only for failure case (see below)
* update_xmax - output parameter, used only for failure case (see below)
* cid - update command ID (used for visibility test, and stored into
* cmax/cmin if successful)
* crosscheck - if not InvalidSnapshot, also check old tuple against this
* wait - true if should wait for any conflicting update to commit/abort
*
* Normal, successful return value is HeapTupleMayBeUpdated, which
* actually means we *did* update it. Failure return codes are
* HeapTupleSelfUpdated, HeapTupleUpdated, or HeapTupleBeingUpdated
* (the last only possible if wait == false).
*
* On success, the header fields of *newtup are updated to match the new
* stored tuple; in particular, newtup->t_self is set to the TID where the
* new tuple was inserted, and its HEAP_ONLY_TUPLE flag is set iff a HOT
* update was done. However, any TOAST changes in the new tuple's
* data are not reflected into *newtup.
*
* In the failure cases, the routine returns the tuple's t_ctid and t_xmax.
* If t_ctid is the same as otid, the tuple was deleted; if different, the
* tuple was updated, and t_ctid is the location of the replacement tuple.
* (t_xmax is needed to verify that the replacement tuple matches.)
*/
HTSU_Result
heap_update(Relation relation, ItemPointer otid, HeapTuple newtup,
ItemPointer ctid, TransactionId *update_xmax,
CommandId cid, Snapshot crosscheck, bool wait)
{
HTSU_Result result;
TransactionId xid = GetCurrentTransactionId();
Bitmapset *hot_attrs;
ItemId lp;
HeapTupleData oldtup;
HeapTuple heaptup;
Page page;
Buffer buffer,
newbuf;
bool need_toast,
already_marked;
Size newtupsize,
pagefree;
bool have_tuple_lock = false;
bool iscombo;
bool use_hot_update = false;
bool all_visible_cleared = false;
bool all_visible_cleared_new = false; ... ////////////////First Phase marked by gaojian newtup->t_data->t_infomask &= ~(HEAP_XACT_MASK);
newtup->t_data->t_infomask2 &= ~(HEAP2_XACT_MASK);
newtup->t_data->t_infomask |= (HEAP_XMAX_INVALID | HEAP_UPDATED);
HeapTupleHeaderSetXmin(newtup->t_data, xid);
HeapTupleHeaderSetCmin(newtup->t_data, cid);
HeapTupleHeaderSetXmax(newtup->t_data, ); /* for cleanliness */
newtup->t_tableOid = RelationGetRelid(relation); ... if (!already_marked)
{
/* Clear obsolete visibility flags ... */
oldtup.t_data->t_infomask &= ~(HEAP_XMAX_COMMITTED |
HEAP_XMAX_INVALID |
HEAP_XMAX_IS_MULTI |
HEAP_IS_LOCKED |
HEAP_MOVED);
/* ... and store info about transaction updating this tuple */ ///HeapTupleHeaderSetXmax(oldtup.t_data, xid);
/////>>>>>added by gaojian for testing.
////xid = (TransactionId)8888; fprintf(stderr,"x-----1,xid is :%d \n",(int)xid); HeapTupleHeaderSetXmax(oldtup.t_data, xid);
HeapTupleHeaderSetCmax(oldtup.t_data, cid, iscombo); sleep();
} ... return HeapTupleMayBeUpdated;
}
可以看到,第一段的 :
newtup->t_data->t_infomask &= ~(HEAP_XACT_MASK);
newtup->t_data->t_infomask2 &= ~(HEAP2_XACT_MASK);
newtup->t_data->t_infomask |= (HEAP_XMAX_INVALID | HEAP_UPDATED);
HeapTupleHeaderSetXmin(newtup->t_data, xid);
HeapTupleHeaderSetCmin(newtup->t_data, cid);
HeapTupleHeaderSetXmax(newtup->t_data, ); /* for cleanliness */
newtup->t_tableOid = RelationGetRelid(relation);
写的很明白, HeapTupleHeaderSetXmin(newtup->t_data, xid);
而第二段的 :
if (!already_marked)
{
/* Clear obsolete visibility flags ... */
oldtup.t_data->t_infomask &= ~(HEAP_XMAX_COMMITTED |
HEAP_XMAX_INVALID |
HEAP_XMAX_IS_MULTI |
HEAP_IS_LOCKED |
HEAP_MOVED);
/* ... and store info about transaction updating this tuple */ HeapTupleHeaderSetXmax(oldtup.t_data, xid);
HeapTupleHeaderSetCmax(oldtup.t_data, cid, iscombo);
}
然后,我再来验证一下,加点调试代码,这样,我执行sql时会出错:
if (!already_marked)
{
/* Clear obsolete visibility flags ... */
oldtup.t_data->t_infomask &= ~(HEAP_XMAX_COMMITTED |
HEAP_XMAX_INVALID |
HEAP_XMAX_IS_MULTI |
HEAP_IS_LOCKED |
HEAP_MOVED);
/* ... and store info about transaction updating this tuple */ ///HeapTupleHeaderSetXmax(oldtup.t_data, xid);
/////>>>>>added by gaojian for testing.
////xid = (TransactionId)8888; fprintf(stderr,"x-----1,xid is :%d \n",(int)xid); HeapTupleHeaderSetXmax(oldtup.t_data, xid);
HeapTupleHeaderSetCmax(oldtup.t_data, cid, iscombo); }
执行结果:
pgsql=# begin;
BEGIN
pgsql=# update tab01 set id= where cd = '';
ERROR: could not access status of transaction
DETAIL: Could not read from file "pg_subtrans/0000" at offset : Success.
pgsql=# \q
[pgsql@loca
再次登陆进来看看,哈,这证明我所看到的代码部分,就是实现写入xmax的部分。
总结来说,就是,在update(可能delete也是),用向oldtup写入transaction id的方式,来记录哪个transaction改动了记录。
[pgsql@localhost bin]$ ./psql
psql (9.1.)
Type "help" for help. pgsql=# select xmin,xmax,cmin,cmax, * from tab01;
xmin | xmax | cmin | cmax | id | cd
------+------+------+------+-----------+----
| | | | |
| | | | |
| | | | |
( rows) pgsql=#
对PostgreSQL xmax的理解的更多相关文章
- 腾讯云数据库团队:PostgreSQL TOAST技术理解
作者介绍:胡彬 腾讯云高级工程师 TOAST是"The Oversized-Attribute Storage Technique"的缩写,主要用于存储一个大字段的值.要理解TOA ...
- 关于linux上postgresql的一些理解
刚开始接触postgresql,安装后就有一个默认用户postgres,而且在启动postgresql后只能通过切换到linux的postgres用户才能登录数据库进行操作,和Mysql的登录认证居然 ...
- PostgreSQL相关的软件,库,工具和资源集合
PostgreSQL相关的软件,库,工具和资源集合. 备份 wal-e - Simple Continuous Archiving for Postgres to S3, Azure, or Swif ...
- PostgreSQL安装详细步骤(windows)
原文地址:http://blog.chinaunix.net/uid-354915-id-3498734.html PostgreSQL安装:一.windows下安装过程安装介质:postgresql ...
- PostgreSQL安装详细步骤windows
PostgreSQL安装:一.windows下安装过程安装介质:postgresql-9.1.3-1-windows.exe(46M),安装过程非常简单,过程如下:1.开始安装: 2.选择程序安装目录 ...
- PostgreSQL安装详细步骤(windows)[转]
PostgreSQL安装: 一.windows下安装过程 安装介质:postgresql-9.1.3-1-windows.exe(46M),安装过程非常简单,过程如下: 1.开始安装: 2.选择程序安 ...
- 2.PostgreSQL安装详细步骤(windows)【转】
感谢 Junn9527 PostgreSQL安装:一.windows下安装过程安装介质:postgresql-9.1.3-1-windows.exe(46M),安装过程非常简单,过程如下:1.开始安装 ...
- PostgreSQL Replication之第五章 设置同步复制(1)
到目前为止,我们已经处理了基于文件的复制(或日志传送)和简单的基于流复制的设置.在两种情况中,在master上事务被提交之后,数据被提交,由slave接收.在master提交和slave实际上完全地接 ...
- PostgreSQL在win7上安装详细步骤
原文:PostgreSQL在win7上安装详细步骤 PostgreSQL安装: 一.windows下安装过程 安装介质:postgresql-9.1.3-1-windows.exe(46M),安装过程 ...
随机推荐
- Vim的撤销与重做
命令模式下 u:撤销 Ctrl+r:重做(撤销撤销)
- Android 权限查寻表
示例 <?xml version="1.0" encoding="utf-8"?> <manifest xmlns:android=" ...
- pku,杨建武:文本挖掘技术
http://webkdd.org/course/ http://www.icst.pku.edu.cn/lcwm/course/WebDataMining/ http://www.icst.pku. ...
- MySQL open table
背景: MySQL经常会遇到Too many open files,MySQL上的open_files_limit和OS层面上设置的open file limit有什么关系? 源码中也会看到不同 ...
- 使用Unity3d的Physics.Raycast()的用法做子弹射击
Class Functions 1)static function Raycast (origin : Vector3, direction : Vector3, distance : float = ...
- Thread: BooleanRT : Realtime 3D boolean operations for (Runtime,Editor)
A Product by Mixed Dimensions What is BooleanRT? BooleanRT is a real-time 3D boolean operations exte ...
- sencha touch tabsidebar 源码扩展
先上图看效果 没错,这是一个sencha touch 项目,而这里的右边推出效果(下文叫做tabsiderbar),使用插件tabsiderbar来扩展的. 插件js下载地址:http://www.m ...
- Java [leetcode 32]Longest Valid Parentheses
题目描述: Given a string containing just the characters '(' and ')', find the length of the longest vali ...
- 优秀it博客和文章
优秀博客 综合 杨文博(供职于百度公司,任复合搜索部资深研发工程师,目前作为tech lead,负责垂直行业搜索后端架构研发.) 杨远骋 徐宥(Google 软件工程师. 这个中文博客是我的思考记录, ...
- window2008 64位系统没有office组件问题分析及解决
服务器是windows server2008 64位系统, 我的系统需要用到Microsoft.Office.Interop.Excel组件 在上传Excel单据遇到错误:检索 COM 类工厂中 CL ...