3com database adapter module for networker nsr-ora v3.2 User Manual

System Administrator’s Guide

U42252-J-Z915-1-76

Recovery

   An automatic recovery is quite unsuitable if the fault in the database files 

is not a physical media failure but a logical error in the data of the 
database. The automatic recovery procedure has no way of detecting 
logical errors. 

An automatic recovery always reproduces the last backed up state prior 
to a physical fault. In order to reproduce an earlier condition, e.g. before 
the occurrence of a logical error, you can use the command ora_recut 
(recovery until time).

   If the data was distributed over a number of NetWorker servers by the 

backup then it cannot be recreated automatically.

Under ORACLE Parallel Server (OPS), the automatic recovery terminates 
directly after the tablespaces have been restored. The redolog files that are dis-
tributed over the various nodes of an OPS system cannot be recreated automat-
ically, instead the user must ensure that they can be accessed from a single 
OPS node and must then recreate them manually. You will find a description of 
this in the section 

.

In all cases in which an automatic recovery is not appropriate, you should carry 
out a manual recovery. For manual recovery you will need to know in which 
directories your redologs and the files or raw devices of your tablespaces are 
located.

The following sections describe how you can perform a recovery of an ORACLE 
database. The first section describes automatic recovery of an optional 
database level using ora_recut, the second section describes an fully-automatic 
recovery of the last saved level of the database, the third section describes a 
manual recovery and the fourth one a recovery with OPS.