在oracle10g之前使用resetlogs打開數(shù)據(jù)庫(kù)之后，之前的的備份就不能用于恢復(fù)了。即不能進(jìn)行跨resetlogs時(shí)間點(diǎn)的恢復(fù)。所以要求執(zhí)行完之后馬上進(jìn)行全庫(kù)備份。

Oracle10g以后允許跨越resetlogs時(shí)間點(diǎn)進(jìn)行完全或者不完全恢復(fù)。實(shí)驗(yàn)如下：

1. 執(zhí)行全庫(kù)備份。

   RMAN> backup database plus archivelog delete all input;

   insert into test select * from test;

   SQL> alter system switch logfile;

   SQL> commit;

   Commit complete.

   SQL> alter system switch logfile;

   System altered.

   SQL> truncate table test;

   Table truncated.

   SQL> alter system switch logfile;

   System altered.

   SQL> select * from v$log_history;

   SQL> select recid,stamp,sequence#,first_change#,first_time,next_change# from v$log_history

   2 where recid >80;

   RECIDSTAMP FIRST_CHANGE# FIRST_TIME NEXT_CHANGE#

   ---------- ---------- ------------- ------------ ------------

   81 856723148 2130643 27-AUG-14 2130889

   82 856752323 2130889 27-AUG-14 2155374

   83 856753004 2155375 28-AUG-14 2175380

   84 856753636 2175380 28-AUG-14 2176030

   85 856754916 2176030 28-AUG-14 2177334

   86 856754945 2177334 28-AUG-14 2177347

   87 856755046 2177347 28-AUG-14 2178414

   88 856755142 2178414 28-AUG-14 2178471

   89 856755199 2178471 28-AUG-14 2178508

   90 856755243 2178508 28-AUG-14 2178543

   91 856755424 2178543 28-AUG-14 2178823

   RECIDSTAMP FIRST_CHANGE# FIRST_TIME NEXT_CHANGE#

   ---------- ---------- ------------- ------------ ------------

   92 856755456 2178823 28-AUG-14 2178836

   93 856755483 2178836 28-AUG-14 2178940

   13 rows selected.

2. 關(guān)閉數(shù)據(jù)庫(kù)、刪除數(shù)據(jù)模擬故障

   刪除之后，執(zhí)行如下

   RMAN> startup mount;

   Oracle instance started

   database mounted

   Total System Global Area 599785472 bytes

   Fixed Size 2022600 bytes

   Variable Size 268436280 bytes

   Database Buffers 322961408 bytes

   Redo Buffers 6365184 bytes

   RMAN> run{

   set until sequence 12 thread 1;

   restore database;

   recover database;

   }

   這里判斷具體到那個(gè)sequence，當(dāng)前查詢?nèi)缦拢瑲v史查看v$log_history的sequence#：

   Sql>archive log list;

   如果出現(xiàn)這個(gè)錯(cuò)誤：

   SQL> recover database until cancel;

   ORA-00283: recovery session canceled due to errors

   ORA-01610: recovery using the BACKUP CONTROLFILE option must be done

   SQL> recover database BACKUP CONTROLFILE

   ORA-00905: missing keyword

   解決：

   recover database using backup controlfile until cancel;

   RMAN> alter database open resetlogs;

   查看序號(hào)是否復(fù)位：

   SQL> archive log list;

   Database log mode Archive Mode

   Automatic archival Enabled

   Archive destination USE_DB_RECOVERY_FILE_DEST

   Oldest online log sequence 0

   Next log sequence to archive 1

   Current log sequence 1

   SQL> create conn xzsp create tt able tt as select * from dba_users;

   Table created.

   SQL> alter system switch logfile;

   System altered.

   SQL> select count(*) from tt;

   COUNT(*)

   ----------

   26

   SQL> inert in sert into tt select * from tt;

   26 rows created.

   SQL> /

   52 rows created.

   SQL> /

   104 rows created.

   SQL> commit

   2 ;

   Commit complete.

   SQL> alter system switch logfile;

   System altered.

   SQL> select recid,stamp,first_change#,first_time,next_change# from v$log_history;

   RECID STAMP SEQUENCE# FIRST_CHANGE# FIRST_TIME NEXT_CHANGE#

   ---------- ---------- ---------- ------------- ------------ ------------

   100 856759449 1 2179277 28-AUG-14 2179707

   101 856759476 2 2179707 28-AUG-14 2179788

   102 856761071 1 2179621 28-AUG-14 2179920

   103 856761108 2 2179920 28-AUG-14 2179945

   103 rows selected.

   Oracle保留了resetlogs之前的日志序列號(hào)，并且日志recid繼續(xù)增長(zhǎng)，控制文件也保留了歸檔日志的序列。所以可以跨RESETLOGS進(jìn)行恢復(fù)。10g以前 是不能夠使用當(dāng)前控制文件恢復(fù)之前的備份的。

   RMAN> run {

   2> restore database;

   3> recover database;

   4> }

   SQL> select count(*) from tt;

   COUNT(*)

   ----------

   208

   恢復(fù)能夠完成，還有l(wèi)og_archive_format有關(guān)系。

   Sql>show parameter log_archive_format

   log_archive_format string archive_%t_%s_%r.log

   這里的%R是oracle新增的參數(shù),是resetlogs的標(biāo)志號(hào).這一歸檔日志格式可以是不同Incarnation的數(shù)據(jù)庫(kù)歸檔日志避免相互覆蓋，從而跨越resetlogs恢復(fù)的日志基礎(chǔ)得以保證。

   ?

   ?

   思考：這里看到rman備份只有一次，是在resetlogs之前，第二次完全恢復(fù)是基于之前的備份。這里看到oracle10確實(shí)是越過(guò)了resetlog執(zhí)行了恢復(fù)。

ORACLE跨越時(shí)間點(diǎn)的恢復(fù)