Should first point out that both the BUE server as well as the Oracle box are mutlihomed, but I've tested with the credentials that the RMAN listener can be connected to over the network segment that's being used for backup traffic, and I used a custom selection list to specifically point to the correct interface on the Oracle machine, so I'm not sure that there'd be a good reason for this to fail.  Failure occurs immediately.

Here's the output from the RMAN script at the time of failure:

Recovery Manager: Release 11.1.0.7.0 - Production on Thu Nov 5 16:05:45 2009

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

 
 # -----------------------------------------------------------------
 # RMAN command section
 # -----------------------------------------------------------------
 
 RUN {
 ALLOCATE CHANNEL ch0
     TYPE 'SBT_TAPE';
 SEND 'BSA_SERVICE_HOST=172.16.4.102,NBBSA_TOTAL_STREAMS=1,NBBSA_JOB_COOKIE={58EA2080-0C0E-488E-B5EE-C407DDE83141},NBBSA_DB_DEVICE_NAME=Oracle-Linux::\\oraclebox-mgt\dbinstance';
 BACKUP
     INCREMENTAL LEVEL=0
     FORMAT 'BE_U'
     DATABASE FORCE;
 BACKUP
     FORMAT 'BE_U'
     ARCHIVELOG ALL FORCE NOT BACKED UP;
 BACKUP CURRENT CONTROLFILE
     FORMAT 'BE_U';
 
 RELEASE CHANNEL ch0;
 }
 
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
ORA-12547: TNS:lost contact

Recovery Manager complete.
 
Devices acquired: 1
Data streams used: 0