在同一个RMAN脚本里面分别allocate两个物理节点的channel。
NMDA Admin文档上有相关说明:
Archived redo logs
Each node in a RAC system maintains a separate set of redo logs. Redo logs that
become full are archived on the local node. As a result, the archived redo logs are
divided among the nodes of the system.
To enable RMAN to back up and recover a RAC system, make all the archived redo
log files accessible by all nodes participating in the backup or recovery. The
appropriate Oracle RAC documentation provides information on how to share the
archived redo logs.
The following sections provide sample scripts to back up and restore all the archived
redo log files in a RAC system:
◆
“Backing up all archived logs from each node” on page 224
◆
“Restoring all archived logs from each node” on page 224
Note:
The archived logs should use the same NSR_CLIENT setting that is used to back up the
EMC NetWorker Module for Databases and Applications
224
Release 1.2 Administration Guide
Cluster Systems
Backing up all archived logs from each node
All the archived log files in a RAC system can be backed up from a single node, such
as a node named ops1.emc.com by using the following type of RMAN script:
run {
allocate channel t1 type ’SBT_TAPE’
connect ’user_name/user_passwd@connect_string_of_ops1’;
allocate channel t2 type ’SBT_TAPE’
connect ’user_name/user_passwd@connect_string_of_ops2’;
send ’NSR_ENV=(NSR_CLIENT=ops1.emc.com)’;
backup filesperset 10
(archivelog all delete input format ’al_%s_%p’);
release channel t1;
release channel t2;
}