and how on earth would anyone discover this when a search, specifically for this string SUBSCRIBE_FOR_NODE_DOWN_EVENT returns no hits on metalink. I don't happen to agree with those who suggest that metalink should be google-searchable, but with the inability of the search engine to return hits that are there, I can see the arguments :(
Check Metalink article 284602.1. The parameter is actually:
where [LISTENER] is the name of the listener, which by default is "LISTENER".
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Fuad Arshad Sent: Friday, September 01, 2006 10:55 AM To: AGUERRA@xxxxxxxxx; oracle-l Subject: Re: Two 10g listeners at the same time
oracle calls this a feature i think it is a bug anyways the fix is subscribe_for_node_down_event=off in listener.ora
----- Original Message ---- From: "Guerra, Abraham J" <AGUERRA@xxxxxxxxx> To: oracle-l <oracle-l@xxxxxxxxxxxxx> Sent: Friday, September 1, 2006 10:46:25 AM Subject: Two 10g listeners at the same time
Twice this week the 10g listener spun a child process and hung all new connections to any database (10.2.0.2 on HPUX 11.11) This is what it looked like:
ps -ef ...
oracle 21170 16606 0 13:57:00 ? 0:00 /usr/oracle/rdbms/10.2/bin/tnslsnr LISTENER -inherit oracle 16606 1 0 Aug 28 ? 83:26 /usr/oracle/rdbms/10.2/bin/tnslsnr LISTENER -inherit
The only way to fix this was to kill the processes (lsnrctl did not work) and start the listener again... No error messages were found in the listener.log.
Any help would be appreciated.
Abraham Guerra American Family Insurance. -- //www.freelists.org/webpage/oracle-l -- //www.freelists.org/webpage/oracle-l
-- Niall Litchfield Oracle DBA http://www.orawin.info