Oracle 10G Listner hanging with HP UX 11.23

[Pradeep Kumar VR]

Hi
Can anybody help me in the below problem.
This week listner hangs twice and no cluster commands are wrkng during that time at active server. I have 2 Itanium servers in active-passive cluster. No error msg in syslog or cluster log. But following error msg in listner log,
" TNS-12500: TNS:listener failed to start a dedicated server process
TNS-12547: TNS:lost contact
TNS-12560: TNS:protocol adapter error
TNS-00517: Lost contact
HPUX Error: 32: Broken pipe"
Pls advice some remedy for this problem
rgds
Pradeep

[Steven E. Protter]

Shalom,

This is an Oracle configuration error or permissions issue.

Checklist:
1) See that the listener is being started by oracle and not root and that file permissions on all oracle binaries and configuration files are not root.

2) Examine the tnsnames.ora and listener.ora file for configuration errors.

3) Because syslog says nothing, I doubt networking is an issue, but its worth taking a look, especially if the configuration runs for a while and then fails.

4) Make sure that all requireds Oracle and HP-UX Oracle patches are installed. swverify \* and make sure no requires patches or software are in a corrupt state. There will be issues here but corrupt software must be removed and re-installed.

5) No cluster commands are working. cmviewcl -v doesn't work? If so there may be an inconsistency in the cluster configuration files between nodes. Normally this generates an error message however. Still its worth a check.

Its unclear from the information provided so far whether the listener fails over time or never starts. That needs to be determined.

SEP

[Yogeeraj_1]

hi Pradeep,

According to Metalink, Intermittent TNS-12500 errors are caused by a lack of resources on the server and finding which resource is depleted may be difficult.

1. The TNS-12500 can be a result of the 'processes' parameter in the init.ora file being too low.

2. If the init.ora 'processes' parameter seems ok then you may need to increase the Unix kernel parameters for the maximum number of processes or users (for example, nproc or maxuprc).

3. Check that you have adequate swap space.

4. Disable OTRACE. OTRACE is a tracing feature that can cause many problems. OTRACE is enabled by default. To disable OTRACE:
a. Stop the Oracle database.
b. Go to the $ORACLE_HOME/otrace/admin directory.
c. Delete all files with a '.dat' extension.
d. Restart the Oracle database.

5. Use the Multi-Threaded Server (MTS) option. MTS will fix the problem since the listener does not spawn dedicated server processes in this environment. MTS also cuts down on the amount of resources being utilized by having connections threaded through dispatchers and by having shared server processes read dispatcher requests from queues. To invoke MTS:

a. Stop the database
b. Modify the init.ora file to include these parameters:

mts_dispatchers=","
mts_max_dispatchers=
mts_servers=
mts_max_servers=
mts_service=
mts_listener_address=

An example of a working MTS configuration would be:

mts_dispatchers="tcp,3"
mts_max_dispatchers=10
mts_servers=2
mts_max_servers=4
mts_service=ORCL
mts_listener_address="(address=(protocol=tcp)(host=mars1)(port=1521))"

c. Restart the database.

hope this helps!

kind regards
yogeeraj

[Martin Decker]

Hello Pradeep,

this is a known issue. Check MetaLink Note 284602.1.

You are experiencing 2 tnslsnr processes, one of which is the parent process of the other. There are no connects possible as the listener is stuck in "fork()".

Regards,
Martin