连接一个运行的实例显示："Connected to an Idle Instance"

[b]"Connected to an Idle Instance" Message when Connecting Bequeath to a Running Instance (文档 ID 435044.1)
[/b]
In this Document

Symptoms

Cause

Solution

References

APPLIES TO:

Oracle Database - Enterprise Edition - Version 8.0.6.0 to 11.2.0.1.0 [Release 8.0.6 to 11.2]
Information in this document applies to any platform.
***Checked for relevance on 25-Apr-2012***

SYMPTOMS

While connecting bequeath to a running instance the following message appears :

$ export ORACLE_SID=test 

$ sqlplus "/ as sysdba" 

SQL*Plus: Release 10.2.0.1.0 - Production on Wed Jun 6 12:17:14 2007 

Copyright (c) 1982, 2005, Oracle. All rights reserved. 

Connected to an idle instance. 

SQL>
The "test" instance was running and this was verified with the ps command:

$ ps -ef | grep pmon 

oracle 9714 1 0 May28 ? 00:00:07 ora_pmon_test 

oracle 32635 1 0 May30 ? 00:00:01 ora_pmon_orcl115b
Furthermore the test instance can be started up:

$ export ORACLE_SID=test 

$ sqlplus "/ as sysdba" 

SQL*Plus: Release 10.2.0.1.0 - Production on Wed Jun 6 12:17:14 2007 

Copyright (c) 1982, 2005, Oracle. All rights reserved. 

Connected to an idle instance. 

SQL> startup nomount 

ORACLE instance started. 

Total System Global Area 866291712 bytes 

Fixed Size 1302036 bytes 

Variable Size 239075820 bytes 

Database Buffers 620756992 bytes 

Redo Buffers 5156864 bytes 

SQL>
 
The alert.log was immediately written to with the informations pertaining to the startup of the operation. When a log switch was performed in the "real" test instance (within a SQL*NET connection)  then the alert.log
was appended with the corresponding information. So at that point on the system there were two instances "test" running from the same Oracle Home

$ ps -ef | grep pmon 

oracle 9714 1 0 May28 ? 00:00:07 ora_pmon_test 

oracle 32635 1 0 May30 ? 00:00:01 ora_pmon_orcl115b 

oracle 10382 1 0 12:27 ? 00:00:00 ora_pmon_test
 

CAUSE

The issue can be caused by a wrong setting of the ORACLE_HOME environment variable used when starting up the instance. This can easily be discovered if one checks the values of the environment variables set within
the background processes of the first "test"  instance started :

$ ps -ef | grep pmon 

oracle 9714 1 0 May28 ? 00:00:07 ora_pmon_test 

oracle 32635 1 0 May30 ? 00:00:01 ora_pmon_orcl115b 

oracle 10382 1 0 12:27 ? 00:00:00 ora_pmon_test

cat /proc/9714/environ

HOSTNAME=ro-rac4 

SHELL=/bin/bash 

TERM=xterm 

HISTSIZE=1000 

USER=oracle 

LS_COLORS=no=00:fi=00:di=00;34:ln=00;36:pi=40;33:so=00;35:bd=40;33;01:cd=40;33;01:... 

ORACLE_SID=test 

MAIL=/var/spool/mail/oracle 

PATH=INPUTRC=/etc/inputrc 

PWD=/home/oracle 

LANG=en_US.UTF-8SSH_ASKPASS=/usr/libexec/openssh/gnome-ssh-askpass 

SHLVL=1 

HOME=/home/oracleLOGNAME=oracleLESSOPEN=|/usr/bin/lesspipe.sh %s 

ORACLE_HOME=/u01/app/oracle/product//10.2.0/db_1 

G_BROKEN_FILENAMES=1_=/u01/app/oracle/product/10.2.0/db_1/bin/sqlplus 

ORA_NET2_DESC=8,11ORACLE_SPAWNED_PROCESS=1 

SKGP_HIDDEN_ARGS=<FATAL/S/x0/x1/x0/x71F6487C/12621/12621/x0>0

$ORACLE_HOME of the old test instance points to "ORACLE_HOME=/u01/app/oracle/product//10.2.0/db_1" . Note the double slash ("//").
whereas  $ORACLE_HOME of the new test instance points to "ORACLE_HOME=/u01/app/oracle/product/10.2.0/db_1" 

ORACLE_SID and ORACLE_HOME are used to generate the IDs of the shared memory segments for the SGA. When using a different ORACLE_HOME value a different ID will get generated and since this does not match any of the existing segments' IDs it is assumed that
this is a new idle instance.
The above example shows how to get the environment variables for a process running on Linux. In order to get these variables for other platforms please read Note
373303.1-How to Check the Environment Variables for an Oracle Process

The same issue can be caused by wrong permissions of the oracle executable. The problem will be discovered only when the bequeath connection will be executed as another OS user than the one used to startup the database.
The correct permissions should be :

-rwsr-s--x 1 oracle oinstall 187777421 Jun 13 19:16 oracle 

and they look like the following when encountering the problem:

-rwxr-xr-x 1 oracle oinstall 187777421 Jun 13 19:14 oracle
 

SOLUTION

Shutdown the instance and start it up from an OS session where all the environment variables are correctly set.
Note: If the instance is to be started via SRVCTL make sure that all the CRS configuration files are using a correct value for the ORACLE_HOME (also in $ORACLE_HOME/bin/racgwrap for 10g, 11.1). To set the Oracle
Home in the cluster's registry to the right value run "srvctl modify database -d <database> -o <Oracle Home>".
-OR-
If the running instance can't be shutdown, set the ORACLE_HOME to the value found in the environment variables set o
cee3
f the running instance.
 
To correct the permissions of the oracle executable the following command has to be executed after stopping all the instances running from that Oracle Home :

cd $ORACLE_HOME/bin

chmod 6751 oracle