Oracle10g New Feature:闪回恢复区空间管理

Oracle10g New Feature:闪回恢复区空间管理

作者：eygle | <script src="http://www.eygle.com/English.js" type="text/javascript"></script>English Version 【转载时请以超链接形式标明文章出处和作者信息及本声明】
链接：http://www.eygle.com/archives/2005/12/oracle10g_flash_recovery_area_space_management.html

---

站内相关文章|Related Articles

修改默认的undo_retention参数设置

How to maintain Oracle10g Recyclebin?

Oracle 10gR2 Study Notes:New View V$FLASH_RECOVERY_AREA_USAGE

工作日志：闪回查询及scn_to_timestamp

如何 DUMP ASH 信息至跟踪文件

Oracle在10g中引入了闪回区(flash recovery area)的概念，用以简化和完善备份，但是闪回区同样需要精心规划和设置，否则一样会遇到问题，从Oracle10gR2开始，Oracle还提供了一个新的视图V$FLASH_RECOVERY_AREA_USAGE，用以监控闪回区空间的耗用情况。本文简要介绍Oracle闪回区的警报和空间维护机制。

每次RMAN在闪回区(flash recovery area)创建文件时，会同时更新可删除文件列表。当闪回区存在空间压力时，Oracle会自动从闪回区中删除废弃文件，当没有更多空间可以释放时，Oracle会给出空间压力警报。

当空间使用达到100%，数据库将会因为无法归档等原因挂起。

闪回区的大小由:db_recovery_file_dest_size 参数指定。
路径由: db_recovery_file_dest 参赛指定。

SQL> show parameter db_recovery
            NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
db_recovery_file_dest                string      /msflsh
db_recovery_file_dest_size           big integer 65G 

这两个参数都是动态参数。

当闪回区空间使用达到85%时，Oracle会发出警告:

*** SERVICE NAME:(SYS$BACKGROUND) 2005-12-03 13:20:16.864
*** SESSION ID:(156.1) 2005-12-03 13:20:16.864
ORA-19815: WARNING: db_recovery_file_dest_size of 53687091200 bytes is 85.00% used, 
and has 8050696704 remaining bytes available.

当空间使用达到97%的时候，Oracle会发出Critical的警报:

ORA-19815: WARNING: db_recovery_file_dest_size of 53687091200 bytes is 97.02% used, 
and has 1602355712 remaining bytes available.

当空间使用达到100%的时候，数据库无法归档就会挂起了:

ORA-19815: WARNING: db_recovery_file_dest_size of 53687091200 bytes is 100.00% used, 
and has 0 remaining bytes available.

接下来就是这样的错误了:

ORA-19809: limit exceeded for recovery files
ORA-19804: cannot reclaim 9563136 bytes disk space from 53687091200 limit
*** 2005-12-04 13:59:14.011 52278 kcrr.c
ARC1: Error 19809 Creating archive log file to 
'/msflsh/MMSDB/archivelog/2005_12_04/o1_mf_1_17108_%u_.arc'
*** 2005-12-04 13:59:14.011 50725 kcrr.c
kcrrfail: dest:10 err:19809 force:0 blast:1
*** 2005-12-04 13:59:14.012 52278 kcrr.c
ARC1: All standby destinations failed; successful archival assumed
*** 2005-12-04 13:59:14.026 16432 kcrr.c
ORA-16038: log 1 sequence# 17108 cannot be archived

注意这里的一个词:reclaim,Oracle用了回收在这里，意思就是已经没有空间可以回收以满足归档的空间需求了。
当Oracle在reclaim空间时，你可能看到如下类似信息:

Sat Oct  1 21:20:54 2005
Deleted Oracle managed file +ORADG/danaly/backupset/2006_09_07/ncsnf0_tag20060907t192619_0.274
Deleted Oracle managed file +ORADG/danaly/archivelog/2006_09_08/thread_1_seq_35.276.600588049
Sun Oct  2 05:46:40 2005
Thread 1 advanced to log sequence 80
  Current log# 2 seq# 80 mem# 0: +ORADG/danaly/onlinelog/group_2.260.600173851
  Current log# 2 seq# 80 mem# 1: +ORADG/danaly/onlinelog/group_2.261.600173853
Sun Oct  2 05:46:41 2005
Deleted Oracle managed file +ORADG/danaly/archivelog/2006_09_08/thread_1_seq_36.277.600600509
Deleted Oracle managed file +ORADG/danaly/archivelog/2006_09_08/thread_1_seq_37.278.600625093
Deleted Oracle managed file +ORADG/danaly/archivelog/2006_09_09/thread_1_seq_38.279.600674413