oracle10g/11g 新特性九(数据库诊断)

Data Recovery Advisor

Data Recovery Advisor

Data Failure: Examples

Listing Data Failures

Advising on Repair

Executing Repairs

Classifying (andClosing ) Failures

Data Recovery Advisor Views

Best Practice:  proactive Checks

Invoking proactve health check of the database and its components

1.Health Monitor or RMAN VALIDATE DATABASE command

2.Checking for logical and physical corruption

3.Findings logged in the ADR

11g 数据库恢复专家例子 1.tablespace--app_data2.backup--datafile--app_data01.dbf_rman   run{   allocate channel d1 type disk format 'F:\app\user\back\datafile_%s_%p.bus';   backup datefile 10;  }3.破坏4.recover--data  recover advise-rman进入rman list failure (会显示出系统检测到的问题) advise failure  分析故障 也可以通过动态性能试图看一些东西： select repair_id, advise_id,summay,rank, status from v$ir_repair; select * from v$ir_manual_checklist; 登录ramn:   repaire failure preview;  只是擦看 （注意这条命令必须和advise failure在同一个session）   repaire failure  ; 真正执行

What Is Block Corruption

Whenever a block is read or wrtten , a consistency check is performed

  a.Block version

  b.DBA(data block address) value in cache s compared to the DBA value in the block buffer

  c.Block-checksum , if enabled

A corrupt block is identified as being one of he following:

   a.Media corrupt

   b.Logically (or software ) corrupt

Block Corruption Symptoms : ORA-01578

The error ORA-01578:"ORACLE data block corrupted (file # %s, block # %s)":

1.Is generated when a corrupted data block is found

2.Always returns the relativefile numer and block number

3.Is returned to the session that issued the query being performed when the corruption was discovered

4.Appears in the  alert.log file

How to Handle Corruption

1.Cehck the alert log and perating system log file

2.Use available diagnostic tools to find out the type of corruption

3.Determine whether the error persists by running chedks multiple times

4.Recover data form the corrupted object if necessary

5.Resolve any hardware issues:

   a.Memory boards

   b.disk controller

   c.Disks

6.Recover or restore data from the corrupt object if necessary

Setting Parameters to Detect Corruption

Setting Parameters to Detect Corruption

坏块实验

1.config   archivelog 归档模式   mount模式下：   alter database flashback on;   select flashback_on from v$database;   alter system set db_ultra_safe='DATA_AND_INDEX'2.create tablespace---app_test    create tablespace app_test datafile '/xxx/xxx/xx/app_test.dbf' size 20M;create user user1 identified by user1 default tablespace app_test;grant connect, resource to user1;user1/user1 登录建几张表在app_test上3.backup_rman  坏块实验  rman target /nocatalog  run {  allocate channel d1 type disk format '/xxx/xxx/xxx/back/app_test_%s_%p.buf'  backup datafile  11  }  4.破坏  关闭系统 然后对app_test.dbf进行编辑  startup 可能能启动起来，可能在查某一张表时候才报错  可以在alert.log里看到  select * from v$databawse_block_corruption; 5.check block-dbv-rman   链接rman   6.startup7.recover block   recover datafile 11 block 185;8.check block

Automatic Diagnostic Workflow

Automatic Diagnostic Repository

The  V$DIAG_INFO View

Location for Diagnostic Traces

现在只需要配置一个参数就可以了 ： diagnostic_dest ,  这样，udump之类的参数  就不用配置了

Health Monitor:Overview

Running Health Checks Manually: PL/SQL Example