cancel
Showing results for 
Search instead for 
Did you mean: 

Multiple Virtual Interfaces (Clusterconfig)

clinden
Level 2
Hi,

we have NBU 4.5 FP6 on HP-UX Server with in this case Solaris9 Client with Oracle9/RMAN on it.

On the Solaris Client there are running multiple Oracle instances, each of them with different clusteradresses.
Because of accounting-reasons, we have to seperate those two Instances in differen client_names, which ist no problem as we do backups.

We have an Gbit-Ethernet connection to the Backupserver. On it there are three IP-Adresses. The first one is the Real Adress of the Host, the second on is the cluster address of the Oracle Instance 1 (ORA1) and the third one, is the cluster address of the ORacle Instance 2 (ORA2).

All these IP-Adresses are in the same subnet. i.e. 10.20.100.0/24 with 10.20.100.90/91/92 as addresses.
The Masterserver sits in the same LAN-segment with 10.20.100.10/24.

We have three clientnames defined. In /usr/openv/netbackup/bp.conf, we have defined the real hostname ehsdav01bk, in the two different oracle user homes we have defined pdav01bk and pdav02bk.

So backup went fine, and everything works okay.

The Problem is restore of the second instance ORA2 with hostname pdav02bk.

bprd log:
10:25:49.121 <2> logconnections: BPRD ACCEPT FROM 10.20.100.92.37006 TO 10.20.100.10.13720
10:25:49.122 <2> bprd: socket fd from accept() is 7
10:25:49.123 <2> connected_peer: Connection from host pdav01bk, 10.20.100.92, on non-reserved port 37006
10:25:49.123 <2> listen_loop: request complete

As you can see, the client is connecting through the wrong interface.
Defining the Hostname in required interface does not face any solution, cause the three interfaces are in the same subnet.

We are in need to define i.e. ge0:2 as required interface.

Has anyone had such a behavior or a solution to this problem?
1 REPLY 1

Ilya_Voronin
Level 4
Try to add
REQUIRED_INTERFACE = pdav02bk
to bp.conf in ORA2 home.