Business Copy and Oracle Database

[Sreejith Kaliyam]

Hi,

We are using the HP EVA4000 business copy to take the snapshots of our oracle databasse. Please see the below procedure we are following. Please advice me how to recover the database.

Backup
Please see the below steps we are using to take the snapshots.
DB server
1) move the oracle database in to hot backup mode using the sql command â alter database begin backupâ ( Oracle is running on hprx6600 server with hp-ux 11iv2 â oracle database 10GR2)
Storage management server
2) start the snapshots of all the mirrors of production luns (data1, data2, redolog) using the CV EVA SSU script. The script first deletes the existing snapshot, fracture the mirror and then take a fully allocated snapshot of the fractured mirrors, resync the mirror.
DB server
3) move the oracle database in to normal mode using the command â alter database end backupâ followed by a log switch.



Recovery
The DB servers are in hp Serviceguard cluster using the file system to store the db files and archive logs. The data is lying on data1, the archive logs are in data2 luns and both are part of single volume group. Redolog is in different volume group. When we mount the snapshots to the secondary server to start the database, it does not start and it requests for an archive log which is not yet generated on the system and is next in the sequence for generation. To recover the database, we are using the following commands

Sql>startup mount
Sql>Recover database using backup controlfile until cancel

Here it asks for auto cancel or return and if we press â autoâ it asks for an archive logs which is not yet generated on the system and is next in the sequence for generation

We then did the following,

Sql> alter system set _allow_resetlogs_corruption=true scope=spfile
(and again mounted the db in order to let this parameter get set ,after this the below command was issued)
Sql>alter database open resetlogs;
Note: (database opened without any problem)

But the customer does not want the â alter system set _allow_resetlogs_corruption=true scope=spfileâ to be used for recovery due to his data integrity concerns. In fact the screen shots of the exact error we are getting while mounting the db without the above command is not available at this moment But can be produced at a later stage.

Could you please confirm that the above steps for taking the oracle db snap shot is the correct step we are following and what is the additional steps to recover the db from the snapshot mount points. we are referring the attached document for taking the snapshots and recovering the database.