Strange ORA-12542 issue

Question

Hi,
I'm facing a baffling ORA-12542 issue for batch processes run by our application.
OS Platform : Solaris 10 8/07 Sparc. Database runs in a Solaris sparse root zone
Database : Oracle 9.2.0.8
Listener : Uses default port of 1521
Background:
We have a daemon running on the database server that polls the database for jobs. This daemon queries a table with the database. If it finds a job it disconnects and spawns a new process based on the job definition. This process in turn calls a shell script that invokes SQL*Plus to generate a report.
Issue:
Periodically the daemon serves an ORA-12542. We've tried tracing the listener but no information comes up there.
In /var/tmp/.oracle we see only two socket files. So old socket files also do not seem to be causing a problem.
We've now increased the queue size for the listener. Also we've tried setting TCP_CONN_REQ_MAX_Q to 1024 on the Solaris server.
The only commonality between the daemon logs and listener logs we find is a step of service update. Could this be having any impact on a daemon which polls frequently? Below, the first line is from our daemon log and the lower lines are from the listener log.
tmd_TRN1[5683:1000]:17/07/2009 16:52:45 ORA-12542: TNS:address already in use
17-JUL-2009 16:52:44 * service_update * TRN1 * 0
17-JUL-2009 16:52:45 * service_update * DEV3 * 0
tmd_TRN1[5683:1000]:17/07/2009 11:32:38 ORA-12542: TNS:address already in use
17-JUL-2009 11:32:37 * service_update * AST3 * 0
17-JUL-2009 11:32:38 * service_update * AST1 * 0
17-JUL-2009 11:32:38 * service_update * DB115RPY * 0
17-JUL-2009 11:32:42 * service_update * AST2 * 0
17-JUL-2009 11:32:48 * (CONNECT_DATA=(SID=TRN1)(CID=(PROGRAM=tmd)(HOST=xxxxxx)(USER=tmbatch))) * (ADDRESS=(PROTOCOL=tcp)(HOST=127.0.0.1)(PORT=50057))
* establish * TRN1 * 0
17-JUL-2009 11:33:01 * service_update * DEV1 * 0
17-JUL-2009 11:33:02 * service_update * DEV3 * 0
17-JUL-2009 11:33:07 * service_update * TRN1 * 0
17-JUL-2009 11:33:18 * service_update * TRN2 * 0
tmd_TRN1[8165:1000]:10/07/2009 15:15:52 ORA-12542: TNS:address already in use
10-JUL-2009 15:15:46 * service_update * TRN2 * 0
10-JUL-2009 15:15:52 * (CONNECT_DATA=(SID=TRN1)(CID=(PROGRAM=tmd)(HOST=xxxxxx)(USER=tmbatch))) * (ADDRESS=(PROTOCOL=tcp)(HOST=127.0.0.1)(PORT=62888))
* establish * TRN1 * 0
10-JUL-2009 15:15:59 * service_update * AST2 * 0
10-JUL-2009 15:16:03 * service_update * DB115RPY * 0
10-JUL-2009 15:16:25 * service_update * TRN1 * 0
Any clues will be appreciated :-)
Cheers,
Aniruddha
Edited by: Aniruddha Khadkikar on Jul 20, 2009 11:29 AM
Edited by: Aniruddha Khadkikar on Jul 20, 2009 11:34 AM

Answer 1

Change this:
LISTENER =
(DESCRIPTION_LIST =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP)(PORT = 1521)(QUEUESIZE = 20))
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC))
)As:
LISTENER =
(DESCRIPTION_LIST =
(DESCRIPTION =
(ADDRESS_LIST =
*(ADDRESS = (PROTOCOL = TCP)(PORT = 1521)( HOST = ipaddress of any one NIC )(QUEUESIZE = 20))*
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC))
Restart the listener and try again. For more see DOC ID 114323.1 on metalink.
Reason may be the entries in /etc/hosts which have two entries one for loopback address and other for NIC's assigned address.
Regards
Edited by: orant575 on Jul 20, 2009 11:52 AM