Oracle 10g Too Many INACTIVE Processes in V$SESSION

Question

Help.
I have a 10g Catelogue database that I use only for RMAN backups for other databases, it
sits on a Windows 2003 server.
My problem is that it is generating a lot of INACTIVE processes for user SYSMAN to such an
extent I had to increase the processes to 600 from 150 (SQL> Alter System Set Processes=600
Scope=spfile;), but by the rate the INACTIVE processes are coming I get database connection
errors and memory errors after a few minutes/hrs after restarting the database.
SQL> select count(*), username, status, state
2 from v$session
3 group by username,status,state
4 /
COUNT() USERNAME STATUS STATE*
2 SYS ACTIVE WAITED SHORT TIME
15 ACTIVE WAITING
234 SYSMAN INACTIVE WAITING
SYSMAN has about 234 INACTIVE processes above, but that figures rises sharply. After some
moments (1 hour) when I try to run the select statement again I get:
SQL> /
from v$session
ERROR at line 2:
ORA-04030: out of process memory when trying to allocate 123404 bytes (QERGH
hash-agg,kllcqas:kllsltba)
When I exit SQLPLUS and try to login I get this
C:\>sqlplus
SQL*Plus: Release 10.2.0.1.0 - Production on Wed Oct 8 12:55:14 2008
Copyright (c) 1982, 2005, Oracle. All rights reserved.
Enter user-name: / as sysdba
ERROR:
ORA-12560: TNS:protocol adapter error
Enter user-name:
So I have to restart the database.
Help me solve my problem.
Regards,
Kevin

Answer 1

These are the EVENTS and it seems they have the same count and also they constitute the majority
SELECT count(), event*
FROM v$session_event
WHERE SID in (Select SID From V$SESSION WHERE USERNAME = 'SYSMAN' AND Status = 'INACTIVE')
GROUP BY event
ORDER BY Event
COUNT() EVENT*
309 SQL*Net break/reset to client
309 SQL*Net message from client
309 SQL*Net message to client
309 control file sequential read
16 log file sync
The method of the RMAN backup:
I'm using windows scheduler to run RMAN scripts with a command like:
rman target / @livedb catalog / @catalogdb backup database;
Regards,
Kevin