ORA-609 TNS-12537 and TNS-12547 in 11g Alert.log

数据库alert日志出现如下错误

Fatal NI connect error 12537, connecting to:
 (LOCAL=NO)

  VERSION INFORMATION:
        TNS for IBM/AIX RISC System/6000: Version 11.2.0.2.0 - Production
        TCP/IP NT Protocol Adapter for IBM/AIX RISC System/6000: Version 11.2.0.2.0 - Production
        Oracle Bequeath NT Protocol Adapter for IBM/AIX RISC System/6000: Version 11.2.0.2.0 - Production
  Time: 21-AUG-2012 09:50:15
  Tracing not turned on.
  Tns error struct:
    ns main err code: 12537

TNS-12537: TNS:connection closed
    ns secondary err code: 12560
    nt main err code: 0
    nt secondary err code: 0
    nt OS err code: 0
opiodr aborting process unknown ospid (15204768) as a result of ORA-609

错误的原因

The ORA-609 error is thrown when a client connection of any kind failed to complete or 
aborted the connection process before the connection/authentication process was complete.

Very often, this connection abort is due to a timeout.  Beginning with 10gR2, a default value for 
inbound connect timeout has been set at 60 seconds.  
This time limit is often inadequate for the entire connection process to complete.    

We have also discovered that the ORA-609 occurs frequently in installations 
where the database is monitored by DB Console and the Enterprise Manager agent (emagent).
After the DB Console is started and as a matter of routine, the emagent will repeatedly try to 
connect to the target instances.  
We can see frequent emagent connections in the listener.log without error.  
However, on occasion it may have failed to complete the connection process at the database so an ORA-609 is thrown.  
The emagent will simply retry the connection and may be successful on the subsequent try.  
(Provided there is no real fault occurring at the listener or database).  
This temporary failure to connect will not be reported back to DB Console and there will be no indication,
 except for the ORA-609, that a fault occurred.

出现这个问题的主要原因是因为从10.2开始inbound connect timeout默认为60 seconds,而在很多建立连接过程可能超过这个时间从而出现类此错误,常见的诱因是DB Console 和 Enterprise Manager agent (emagent). EM会重复的尝试连接到数据库。其过程中会偶尔的出现连接超时的问题，但是接下来会继续尝试，并获得成功。这种临时的失败不会导致EM报错而只会以ora-609的形式记录在alert log中.

处理方法

For that reason, we often recommend increasing the values for INBOUND_CONNECT_TIMEOUT at both listener 
and server side sqlnet.ora file as a preventive measure.  
If the problem  is due to connection timeouts, an increase in the following parameters should eliminate 
or reduce the occurrence of the ORA-609s.

e.g.
Sqlnet.ora: SQLNET.INBOUND_CONNECT_TIMEOUT=180
Listener.ora: INBOUND_CONNECT_TIMEOUT_listener_name=120

These settings are in seconds.  Again, the default is 60.

问题跟踪方法

If the issue persists and inbound connect does not have any effect, the following steps are intended to 
help locate  the client that may be causing the errors.

1)  Suppress the TNS errors in the alert.log by setting the following listener.ora file parameter:  

DIAG_ADR_ENABLED_listener_name=OFF

This will cause the TNS errors to be posted to the ORACLE_HOME/network/log/sqlnet.log file that is 
local to the database and 
may yield useful information about the client's address.

For example, here's a snippet from a server side sqlnet.log where client address info was posted:

Production Time: 15-FEB-2010 07:15:01

Fatal NI connect error 12537, connecting to:
(DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(Host=yourhost)(Port=1521))(CONNECT_DATA=(SID=PROD1DR)
(CID=(PROGRAM=sqlplus)(HOST=client_host)(USER=client))))

Observe the PROGRAM and HOST fields on the last line.  This is where the connection originated.
Be sure to match timestamps in the sqlnet.log with the timestamps of the alert.log errors.  
Once you've located the offending client, 
you can enable client tracing to try and determine the cause:

TRACE_LEVEL_CLIENT=16
TRACE_DIRECTORY_CLIENT=<dir location>
TRACE_TIMESTAMP_CLIENT=TRUE
DIAG_ADR_ENABLED=off   <<<<<11g or newer client requirement

If you need assistance with client or server tracing, please open an SR with Global Customer Support.


2)  Check the listener.log for client connections that were logged at timestamps that match the ORA-609 
timestamps as they appear in the alert.log.  
The client information is recorded in each listener.log entry.  
Since this error occurs AFTER the listener has handled the connection, 
do not expect to see errors in the listener.log.

Here's an example snippet of an incoming client connection that was posted to the listener.log:



20-JAN-2009 17:08:45 (CONNECT_DATA=(SID=orcl)(CID=
(PROGRAM=D:\oracle\product\10.1.0\Db_1\perl\5.6.1\bin\MSWin32-x86\perl.exe)(HOST=myclient)

Note that the exact timestamp, program name and client host will often be recorded.  Again, 
once you've located the offending client, 
enable tracing (see above) to try to capture the connection failure.

3)  Enable server side Oracle Net tracing and capture the TNS error along with the incoming connection.
Match the PID that accompanies the ORA-609 to the server trace label.  e.g.  

ORA-609 : opiodr aborting process unknown ospid (4799_1)  *Note the PID

This PID would correspond to server trace labeled:  svr_4799.trc.  Check the server trace for either 
TNS error (the 609 will not appear) 
and try to locate the originating client address.  If assistance is needed for this investigation, 
please open an SR with Oracle Support.

See below for instuctions on enabling Oracle Net server tracing.

The following details the discovery of the source of an ORA-609 for a real case:

The alert.log reports the following messages intermittently but frequently:

Mon Nov 16 22:39:22 2009
ORA-609 : opiodr aborting process unknown ospid (nnnn)

Enabled Oracle Net server tracing:

TRACE_LEVEL_SERVER=16
TRACE_DIRECTORY_SERVER=<dir location>
TRACE_TIMESTAMP_SERVER=TRUE
DIAG_ADR_ENABLED=off

Reloaded listener and wait for error to appear again.:


ORA-609 : opiodr aborting process unknown ospid (5233_1)

Note that the server trace file set that corresponded to this event was named svr_5233*.trc.
Of course the timestamps of the alert.log event and the server trace creation matched as well.

A review of the server trace showed only an EOF failure and the  TNS-12537 error:


Read unexpected EOF ERROR 
nserror: nsres: id=0, op=68, ns=12537

In this particular case, there was no information about the client in the trace. 
This is atypical for a server trace.   
It may be that the client aborted before all the client information was posted to the file.  
However, there was post in the listener.log f
or an emagent connection that was established at the same point in time.

Here's an excerpt from a listener.log entry where an emagent establishes a connection:

PROGRAM=D:\oracle\product\10.1.0\Db_1\bin\emagent.exe)

Checked the EM Agent traces and logs and discovered the following entry:

Fatal NI connect error 12547, connecting to:
(LOCAL=NO)

VERSION INFORMATION:
TNS for Solaris: Version 11.1.0.7.0 - Production
Oracle Bequeath NT Protocol Adapter for Solaris: Version 11.1.0.7.0 - Production
TCP/IP NT Protocol Adapter for Solaris: Version 11.1.0.7.0 - Production
Time: 16-NOV-2009 22:39:22

****Tracing to file: /backup/sid_traces/sqlnetlog/svr_5233.trc 

Tns error struct:

ns main err code: 12547
TNS-12547: TNS:lost contact
ns secondary err code: 12560
nt main err code: 0
nt secondary err code: 0
nt OS err code: 0

****Note the name of the server trace which contains the PID:  svr_5233.trc 
Also, the timestamp of the agent event matches the timestamp of the alert.log error.



Check the following locations for EM Agent traces. If working with support on this issue and 
the EM Agent is suspected, upload ALL files under:

$ORACLE_HOME/sysman/log/emagent.trc < Single node agent trace location
$ORACLE_HOME/host/sysman/log/emagent.trc < RAC agent trace location

It was determined that in this case, the emagent was aborting the connection 
before it was complete and then simply reconnecting 
and succeeding on the subsequent try.  No errors were reported in the listener log or listener trace. 
No errors were returned to the DB Console.
There was no apparent outage of any kind.  No action was taken to correct the ORA-609 in this case.  
It was decided that the message was informational and completely benign. 

参考文档：
ORA-609 TNS-12537 and TNS-12547 in 11g Alert.log (Doc ID 1116960.1)
Troubleshooting Guide ORA-609 : Opiodr aborting process unknown ospid (Doc ID 1121357.1)