Listener Cfg Question

[tman138]

I am running an Oracle 8.16 databse on HPUX 11.
I recently re-configured my production server to include a test database as well. I modifed the listener.ora file to include the new test SID and the listener when started shows both the test and production service handlers running sucessfully. I shutdown every night for backups using a script file that performs several shutdown routines including LSNRCTL stop and LSNRCTL dbsnmp_stop. This script has not changed from its original state when I was running only the one database SID. The listener log shows the listener halts @ midnight and attempts to restart @ 2:30 AM but gets a TNS-12502, TNS-12542 and other errors. The odd thing is that if I mannually shut down the listener - and it reports as not running, and then restart it all is normal again. Here is a portion of the listener log:

19-JUL-2005 00:50:15 * (CONNECT_DATA=(CID=(PROGRAM=)(HOST=nsm55dbs)(USER=root))(COMMAND=stop)(ARGUMENTS=64)(SERVICE=LISTENER)(VERSION=135290880)) * stop * 0
No longer listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=ipc)(KEY=EXTPROC)))
No longer listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=nsm55dbs)(PORT=1526)))
No longer listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=nsm55dbs)(PORT=2481))(PROTOCOL_STACK=(PRESENTATION=GIOP)(SESSION=RAW)))
19-JUL-2005 00:50:20 * service_quit * prod * 0
19-JUL-2005 00:51:34 * service_quit * test * 0
19-JUL-2005 02:30:15 * service_register * prod * 0
19-JUL-2005 02:30:27 * service_register * test * 0

TNSLSNR for HPUX: Version 8.1.6.0.0 - Production on 19-JUL-2005 02:30:30

(c) Copyright 1998, 1999, Oracle Corporation. All rights reserved.

Log messages written to /oracle/product/816/network/log/listener.log
Trace information written to /oracle/product/816/network/trace/listener.trc
Trace level is currently 0

Attempted to listen on: (ADDRESS=(PROTOCOL=ipc)(KEY=PNPKEY))
19-JUL-2005 02:30:30 * 12502
TNS-12502: TNS:listener received no CONNECT_DATA from client
TNS-12542: TNS:address already in use
TNS-12560: TNSrotocol adapter error
TNS-00512: Address already in use
HPUX Error: 226: Address already in use

Another oddity is that I have a mixture of Oracle 7.3.4 clients as well as 8i and 9i clients. The 7.3.4 clients can still connect & TNSPing works on those clients, but the 8 & 9i clients cannot connect until I manually stop and restart the listener & TNSPing cannot connect. It's almost at there is a ghosted process running. I am stymied.
Any thoughts?

 

[tman138]

I figured it out myself. The key is right there in the first log line. The stop command was issued by root rather than Oracle. Apparently the listener was stopped but the address was still in use. I added 'su - oracle -c' prior to the lsncrctl stop and voilá - problem solved! The listener shuts down properly, releases the address, and the dbshut commmand was executed as Oracle to begin with so that part was fine. The start up script is unchanged, and the result is much better. TNS errors all gone.