This topic has been archived. It cannot be replied.
-
工作学习 / 专业技术讨论 / 请教个oracle archive log的问题?TB级数据库,硬盘是RAID 0,三个一组。坏掉了一个硬盘=烂掉3个,修复一把无效,只能重建文件系统. 不巧硬盘中的数据包括归档日志和若干个数据文件。
有10月1日的备份,目前已经恢复了那3个硬盘的数据(10.1)。但由于从10月10日到目前的archive log都没有啦,如何能让那些丢失的数据文件重新online.
难道非得把所有10月1日的datafile,controlfile全部restore? 超慢磁带机估计得做好几天。
-thisunreal(饭得志);
2005-11-2
{373}
(#2585897@0)
-
如果你不在乎10月后的数据,应该可以用RECOVER DATAFILE OR TABLESPACE TO
RECOVERY TO 10。1。
-fxid(民工模范);
2005-11-2
{8143}
(#2585936@0)
-
但这个前提是用备带把1T多的数据文件都恢复一次到某个时间点?
-thisunreal(饭得志);
2005-11-2
(#2585955@0)
-
the scenario may like this, uu have tons current datafiles and one old and one current control file, u use recover datafile(old) until controlfile, system will use controlfile to recognize the old and put new timetamp on it..
-fxid(民工模范);
2005-11-2
{196}
(#2585984@0)
-
if u are not so hurry , I can do a test for on NT platform or u have a testing database could do this.
-fxid(民工模范);
2005-11-2
(#2585986@0)
-
sorry, I couldn't, No oracle on my machine anymore.
-fxid(民工模范);
2005-11-2
(#2586000@0)
-
.
-thisunreal(饭得志);
2005-11-2
{650}
(#2586221@0)
-
hi, buddy, why did u try 'recover database' instead of 'recover datafile'? from current
control file, u have no change to recover old datafile with recover database.
-fxid(民工模范);
2005-11-2
(#2586302@0)
-
这两个是一样的,都失败。recover database = 同时recover几个datafile
-thisunreal(饭得志);
2005-11-2
(#2586475@0)
-
That's weird.Right now the biggest hard drives are 300GB. 3*300GB=900GB maximum less than 1 TB. How come you said it's TB level database?
I am not DB expert, but to me, it's so dumb because RAID 0 is with very high risk.
If you have 10.1 db, can you only restore the archived logs from tape to hard drive and do the recovery? How big is the archived logs from 10.1 to now?
-xczheng(图图);
2005-11-2
{363}
(#2586904@0)
-
.
-thisunreal(饭得志);
2005-11-2
{549}
(#2586955@0)
-
I think u can open the old datafiles with or without rebuild control file and needn't
to recover entire database. But I lost my oracle on my machine. I tried to download
10g but won't unzip it. any suggestion?
-fxid(民工模范);
2005-11-2
(#2586972@0)
-
我仔细想了一下,始终存在consistency的问题
数据无法同步,肯定会失败。2种方法:
1、把所有数据文件倒回同一个时间点,然后用归档日志恢复到某时间点 --- 以前常干这个
2、有足够多的归档日志可以把老早的离线文件恢复到当前状态
在归档日志损坏的情况下,只能把全部数据文件restore。
-thisunreal(饭得志);
2005-11-3
{263}
(#2587957@0)
-
我这么理解,你现在的数据库可以OPEN,然后作ALTER TABLESPACE OR DATAFILE
OFFLINE;然后把老的DATAFILE RESTORE,然后RECOVER WITH REBUILD CONTROL FILE, 不知道可行不?
-fxid(民工模范);
2005-11-3
(#2588061@0)
-
我的理解是,所有的恢复都是为了数据同步。归档日志和数据文件都丢了咋同步?还好机器没什么人用,系统烂在那里2个多星期了才有人看到。整套东西包括SAN存储unix+数据库也没人管,正好拿来练手。
-thisunreal(饭得志);
2005-11-3
{110}
(#2588098@0)
-
一种可能是那几个数据文件一直都没人用,ARCHIVE LOG 不涉及它们,不存在数据完整性问题,而是如何再RECOGNIZE 的问题。否则不会坏了很久才给人发现。
-fxid(民工模范);
2005-11-3
(#2588132@0)
-
, 你现在DB能起来么,如果能,如果你有从10.1到现在的ARCHIVE LOG 在磁带上的备份,就拿磁带机全倒进来. 就可以RECOVE到现在的. 如果没备份,就头大了你.
-judy_an(绿腰);
2005-11-3
(#2587983@0)
-
DB能起来,几个数据文件offline;archive log和烂掉的硬盘一起消失了。
-thisunreal(饭得志);
2005-11-3
(#2588103@0)
-
还要那几个数据文件/表空间做甚,直接DROP掉吧。反正不在乎数据,又不想再从磁带恢复,数据文件/archive log也没。
-llau(眉豆);
2005-11-3
{62}
(#2588309@0)
-
你说的是3个disk的卷没有了,那是这个表空间都没有了吗?如果别的表里有一些类似foreign key的关联, 那系统怎么处理呢?
-xczheng(图图);
2005-11-3
(#2588755@0)