估计大部分都是邓超的粉丝。我对明星不感冒，不过看下来，邓超确实演得好，而且也没有因为自己大牌而偷懒，戏份很多。

这部戏如果没有邓超的表演只能算一般般。孙俪没现身，来的是孙俪她妈……

当LMT表空间中文件被offline后，如果该文件中包含有segment header，那么从dba_extents中不能看到该segment。

来看一个测试：

SQL> select file_id from dba_data_files where tablespace_name=’TEST’;

FILE_ID
———-
10
11

SQL> create table eagle_fan(x int) tablespace test;

Table created.

SQL> select header_file from dba_segments where segment_name=’EAGLE_FAN’ and owner=’SYS’;

HEADER_FILE
———–
10

SQL> select extent_id ,file_id from dba_extents where segment_name=’EAGLE_FAN’ and owner=’SYS’;

EXTENT_ID FILE_ID
———- ———-
0 10

SQL> select distinct segment_name from dba_extents where file_id=10;

SEGMENT_NAME
——————————————————————————–
TEST
EAGLE_FAN

SQL> alter database datafile 10 offline ;

Database altered.

SQL> select extent_id ,file_id from dba_extents where segment_name=’EAGLE_FAN’ and owner=’SYS’;

SQL> select distinct segment_name from dba_extents where file_id=10;

SEGMENT_NAME
——————————————————————————–
TEST

no rows selected

可以看到datafile 10被offline后，从dba_extents已经看不到该segment。推想而知，对于LMT tablespace，dba_extents会读取文件头信息。
那么也就是说，如果有文件被offline后，我们无法单从dba_extents中知道该文件中包含有哪些segment。

那么我们应该再从dba_segments中得到segment header在该文件中的segment:

SQL> select distinct segment_name from dba_extents where file_id=10
2 union
3 select distinct segment_name from dba_segments where header_file=10;

SEGMENT_NAME
——————————————————————————–
EAGLE_FAN
TEST

而DMT tablespace不存在这个问题，我们来看一看dba_extents的视图定义就很清楚了：

union all以上的部分对应于DMT tablespace，下面的部分对应于LMT tablespace。

DMT tablespace从uet$中读取extent信息，所以不受offline datafile的影响。

LMT tablespace从x$ktfbue中读取extent信息，对于x$ktfbue，描述为”Used extent bitmap in file header for LMT (equivalent to uet$ in DMT)”

我们来看看datafile offline前后，该表有什么变化:

– online datafile 10,多分配几个extent给表eagle_fan，并记录下一些信息

SQL> recover datafile 10
Media recovery complete.
SQL> alter database datafile 10 online;

Database altered.

SQL> alter table eagle_fan allocate extent;

Table altered.

SQL> /

Table altered.

SQL> /

Table altered.

SQL> /

Table altered.

SQL> select extent_id,file_id,block_id from dba_extents where segment_name=’EAGLE_FAN’ order by extent_id;

EXTENT_ID FILE_ID BLOCK_ID
———- ———- ———-
0 10 7049
1 11 7177
2 10 7177
3 11 7305
4 10 7305

SQL> select ktfbuefno file_id,count(*) from sys.x$ktfbue where ktfbuefno in (10,11) group by ktfbuefno;

FILE_ID COUNT(*)
———- ———-
10 58
11 58

SQL> create table t as select * from sys.x$ktfbue where ktfbuefno in (10,11);

Table created.

– offline后再来比较，可以发现关于Eagle_Fan表的信息丢失

SQL> alter database datafile 10 offline;

Database altered.

SQL> select ktfbuefno file_id,count(*) from sys.x$ktfbue where ktfbuefno in (10,11) group by ktfbuefno;

FILE_ID COUNT(*)
———- ———-
11 56
10 55

SQL> select e.ktfbuefno,e.ktfbuesegfno,e.ktfbuesegbno,e.ktfbuesegtsn from t e where ktfbuefno in (10,11) and
2 (ktfbuefno,ktfbuesegfno,ktfbuesegbno,ktfbuesegtsn)
3 not in (select ktfbuefno,ktfbuesegfno,ktfbuesegbno,ktfbuesegtsn from sys.x$ktfbue where ktfbuefno in (10,11))
4 ;

KTFBUEFNO KTFBUESEGFNO KTFBUESEGBNO KTFBUESEGTSN
———- ———— ———— ————
10 10 7052 9
11 10 7052 9
10 10 7052 9
11 10 7052 9
10 10 7052 9

1* select header_file,header_block,segment_name from dba_segments where header_file=10 and header_block=7052
SQL> /

HEADER_FILE HEADER_BLOCK SEGMENT_NAME
———– ———— ——————————
10 7052 EAGLE_FAN

SQL> select * from dba_extents where segment_name=’EAGLE_FAN’;

no rows selected

数据库仍然可以正常运行，但是LNS进程报错称找不到被删除的logfile。我们使用lgwr async模式传输redo到远端physical standby，physical standby无法得到被删除的logfile。

从pfiles中可以看出，因为lgwr进程每时每刻都获得所有logfile的文件句柄，所以虽然该logfile被删除，lgwr还是可以正常读写该logfile，数据库还是可以正常的运行。

但是lns进程只是获得当前的文件句柄，所以当切换到被删除的logfile时，lns进程不能访问该文件。

XFAN primary$> ps -ef |grep lns1|grep XFAN
oracle 17579 1 0 03:02:06 ? 0:01 ora_lns1_XFAN
XFAN primary$> pfiles 17579 | grep redo
/oracle/SCRATCH/data03/XFAN/redo/.redo2.log::cdev:vxfs: — LNS process only hold redo2.log’s handler
XFAN primary$> ps -ef |grep lgwr|grep XFAN
oracle 16377 1 0 02:44:54 ? 0:01 ora_lgwr_XFAN
XFAN primary$> pfiles 16377 |grep redo
/oracle/SCRATCH/data03/XFAN/redo/.redo1.log::cdev:vxfs:
/oracle/SCRATCH/data03/XFAN/redo/.redo2.log::cdev:vxfs:
/oracle/SCRATCH/data03/XFAN/redo/.redo3.log::cdev:vxfs:
/oracle/SCRATCH/data03/XFAN/redo/.redo4.log::cdev:vxfs: — LGWR process hold all redo log files’ handler

首先做的是在primary数据库上清除该logfile，过程如下：

1. 切换到好的logfile group
alter system switch logfile;
2. checkpoint，使该logfile的状态从active变为inactive
alter system checkpoint;
3. 清除该logfile，因为不能被归档，所以要加上unarchived
alter database clear unarchived logfile group x;

因为中间少了一个archive log，所以physical standby不能继续recover。

使用incremental backup来修复该physical standby，大约花费6个小时。如果重新build standby，大约需要花费一个星期。

对于有些部分复制的logical standby来说，需要忽略这部分错误。对于grant select on table to user，oracle已经很聪明的忽略了这类错误，显然grant sequence的问题是设计上的一个bug。

还好我们可以通过dbms_logstdby.skip_error来自动忽略此错误。

首先创建一个procedure用来处理GRANT的DDL 错误

CREATE OR REPLACE PROCEDURE sys.handle_error_ddl (
old_stmt IN VARCHAR2,
stmt_type IN VARCHAR2,
schema IN VARCHAR2,
name IN VARCHAR2,
xidusn IN NUMBER,
xidslt IN NUMBER,
xidsqn IN NUMBER,
error IN VARCHAR2,
new_stmt OUT VARCHAR2

) AS
BEGIN
— Ignore any GRANT errors
new_stmt := old_stmt;
IF INSTR(UPPER(old_stmt),’GRANT’) > 0
THEN
new_stmt := NULL;
END IF;
END handle_error_ddl;
/

然后执行dbms_logstdby.skip_error，当遇到任何Non schema DDL的错误时，调用procedure sys.handle_error_ddl 来处理。

begin
DBMS_LOGSTDBY.SKIP_ERROR (‘NON_SCHEMA_DDL’, ”,”, ‘SYS.HANDLE_ERROR_DDL’);
end;
/

之前我有一篇文章尽量缩短oracle upgrade时间讲述了一些缩短升级过程的技巧，不过对于10g到11g的升级，怎么算下来都需要1个多小时（目前我们还在寻找缩短11g升级时间的方法）。11g中含有set echo on的文件有下面6个，都是在$ORACLE_HOME/rdbms/admin下面：

1. catocm.sql
2. prvthtdb.plb
3. prvtaddm.plb
4. prvtpspi.plb
5. utlrdt.sql
6. prvtbog2.plb

这样的话，对于我们2个小时的outage window，如果升级过程中出现问题想做downgrade（downgrade过程大约需要40分钟），极大可能会超出outage window。所以必须找到一种快速downgrade的方法。这里便要提到10g的new feature — flashback database。

经过我的测试，flashback database只用了9秒钟，enable flashback会增加大约1分钟upgrade的时间。flashback area的实际使用大小大约为1.6GB。

具体步骤如下：

1. modify compatible = 10.2.0.4
2. Update log_archive_format = TEST_%s.%t_%r_arc
3. set diagnostic_dest in 11g init.ora
#diagnostic_dest=/oracle/TEST/data14/TEST #enabled when switching to 11g
4. Turn on Flashback
DB_RECOVERY_FILE_DEST_SIZE=4G
DB_RECOVERY_FILE_DEST=’/oracle/TEST/data14/TEST/flashback’

– in 10g environment.

SQL> startup mount

SQL> alter database flashback on;

SQL> CREATE RESTORE POINT before_upgrade GUARANTEE FLASHBACK DATABASE;

Restore point created.
SQL> shtudown immediate;

– switching to 11g environment

SQL> startup upgrade

SQL> spool upgrade.log
SQL> set time on
SQL> @catupgrd.sql

SQL> startup

SYS@TEST:prod SQL> @?/rdbms/admin/utlu112s
.
Oracle Database 11.2 Post-Upgrade Status Tool 01-28-2010 00:49:30
.
Component Status Version HH:MM:SS
.
Oracle Server
. VALID 11.2.0.1.0 00:45:51
Oracle Real Application Clusters
. INVALID 11.2.0.1.0 00:00:02
Gathering Statistics
. 00:12:59
Total Upgrade Time: 00:58:55

PL/SQL procedure successfully completed.

SQL> @?/rdbms/admin/catuppst
SQL> @?/rdbms/admin/utlrp
SQL> shutdown immediate

– Flashback in 11g environment
SYS@TEST:prod SQL> shutdown immediate
Database closed.
Database dismounted.
ORACLE instance shut down.
SYS@TEST:prod SQL> startup mount

SYS@TEST:prod SQL> set time on timing on

00:55:47 SYS@TEST:prod SQL> select FLASHBACK_ON from v$database;

FLASHBACK_ON
——————
RESTORE POINT ONLY

Elapsed: 00:00:00.01
00:55:54 SYS@TEST:prod SQL> FLASHBACK DATABASE TO RESTORE POINT before_upgrade;

Flashback complete.

Elapsed: 00:00:09.00

Incomplete Recovery applied until change 3565405231814 time 01/27/2010 03:11:22
Flashback Media Recovery Complete
Completed: FLASHBACK DATABASE TO RESTORE POINT before_upgrade
SQL> shutdown immediate

– switch back to 10g version

SQL> startup mount

SQL> alter database open resetlogs;

Database altered.

SQL> select comp_name,version,status,modified from dba_registry;

SYS@USER10:prod SQL> select comp_name,version,status,modified from dba_registry;

COMP_NAME
————————————————————————————————————————————
VERSION STATUS MODIFIED
—————————— ——————————— ————————–
Oracle Database Catalog Views
10.2.0.4.0 VALID 01-MAY-2009 00:04:13

Oracle Database Packages and Types
10.2.0.4.0 VALID 01-MAY-2009 00:04:13

Oracle Real Application Clusters
10.2.0.4.0 INVALID 01-MAY-2009 00:04:13

你能一眼就找到我吗？

晕，学习轻松…..这也是夸奖吗

歌曲：Amazing Grace
演唱者：Declan Galbraith

You Raise Me Up

2005年，专辑《Face To Face》，歌手：Westlife

When I am down and, oh my soul, so weary;
When troubles come and my heart burdened be;
Then, I am still and wait here in the silence,
Until you come and sit awhile with me.

You raise me up, so I can stand on mountains;
You raise me up, to walk on stormy seas;
I am strong, when I am on your shoulders;
You raise me up… To more than I can be.

You raise me up, so I can stand on mountains;
You raise me up, to walk on stormy seas;
I am strong, when I am on your shoulders;
You raise me up… To more than I can be.

You raise me up, so I can stand on mountains;
You raise me up, to walk on stormy seas;
I am strong, when I am on your shoulders;
You raise me up… To more than I can be.
You raise me up, so I can stand on mountains;
You raise me up, to walk on stormy seas;
I am strong, when I am on your shoulders;
You raise me up… To more than I can be.
You raise me up… To more than I can be.