将RAC软件转换为单实例软件

http://blog.itpub.net/26736162/viewspace-2155632/

1. Stop database and CRS on both node

$ srvctl stop database -d mydb

# crsctl stop crs

2. Turn Off RAC

SQL> startup

ORA-29702 error occurred in Cluster Group Service operation

Relink with the RAC OFF.

$ cd $ORACLE_HOME/rdbms/lib

$ /usr/ccs/bin/make -f ins_rdbms.mk rac_off

Relinking oracle

$ make -f ins_rdbms.mk ioracle

## OR , both working fine

$ cd $ORACLE_HOME/bin

$ relink oracle

If ASM Instance Exist, run below command as root

# /oracle/product/10.2.0/db/bin/localconfig delete

# /oracle/product/10.2.0/db/bin/localconfig add

3. Parameter(Pfile/spfile) & database changes

SQL> startup

SQL> alter database disable thread 2;

SQL> alter system set remote_listener='';

3a. Remove unwanted logfile

SQL> select thread#, group# from v$log;

SQL> alter database drop logfile group 3;

SQL> alter database drop logfile group 4;

3b. Remove unwanted tablespace

SQL> drop tablespace UNDOTBS2 including contents and datafiles;

3c. Rename instance name.

SQL> alter system set instance_name= scope=spfile;

SQL> shutdown immediate

SQL> startup

- Change your ORACLE_SID environment

4. Run $ORA_CRS_HOME/install/rootdelete.sh on both node

- This will stop and remove all CRS startup related file

5. Remove $ORA_CRS_HOME binary using Clusterware OUI installer

- Ignore any error if 2nd node already down

- rm -rf $ORA_CRS_HOME

6. Modify listener file

$ vi $ORACLE_HOME/network/admin/listener.ora

6a. Modify tnsname file

$ vi $ORACLE_HOME/network/admin/tnsnames.ora

应该还要把初始化参数中的cluster_database改成false

The node cloning procedure from a RAC database creates a cloned single instance RAC DB, so the new cloned DB contains an extra thread/instance as in the original RAC environment.

The following steps will remove the other node's files or any existing information in the DB that might contains information for the other instances that should not exist in this single instance environment:

1.  Shutdown all the instance in RAC environment.

2.  Shutdown all the listeners.

3.  Relink Oracle executable with rac_off option.

  $ make -f ins_rdbms.mk rac_off

  $ make -f ins_rdbms.mk ioracle

4.  Remove all the parameters from init.ora which points to second instance and it is on other node.

5.  Remove cluster_database parameter

6.  Remove undo_tablespace parameter which points to second instance

7.  Start-up the database in mount stage and execute..

alter database disable thread

8.  Open the database.

9.  After opening database you can drop the redo-log groups which are for other instance. You can also drop any other tablespace of other instance.

10.  Run the following command as the Oracle software install owner, and make sure the $ORACLE_HOME is defined before running the following command:

$ORACLE_HOME/oui/bin/runInstaller -updateNodeList ORACLE_HOME=$ORACLE_HOME

"CLUSTER_NODES={node_list}" CRS=FALSE

Note: Replace the {node_list} with the current node only as this is a single instance database so you should write down the existing only node name of the node that is hosting this database instance.

1.1  将RAC软件转换为单实例软件

http://blog.itpub.net/26736162/viewspace-2155632/

在rac环境中，只有在集群正常启动的情况下才能创建和启动单实例的数据库，否则报错：

[oracle@raclhr-11gR2-N1 ~]$ sqlplus / as sysdba

SQL*Plus: Release 11.2.0.3.0 Production on Tue Jun 5 09:40:26 2018

Copyright (c) 1982, 2011, Oracle.  All rights reserved.

Connected to an idle instance.

SYS@OGGS> startup

ORA-29702: error occurred in Cluster Group Service operation

SYS@OGGS>

解决办法：

NON-RAC Database Startup Giving Error ORA-29702 (文档 ID 433310.1)

1. Shutdown the database completely.

2. Relink with RAC OFF ，use oracle owner:

$ cd $ORACLE_HOME/rdbms/lib

$ make -f ins_rdbms.mk rac_off

$ make -f ins_rdbms.mk ioracle

1. Startup the database.

在执行编译后，集群仍然能正常启动，但是集群中的RAC数据库不能正常启动，会报错：

SYS@lhrrac11> startup

ORA-01078: failure in processing system parameters

ORA-00439: feature not enabled: Real Application Clusters

SYS@lhrrac11>

恢复操作：

$ cd $ORACLE_HOME/rdbms/lib

$ make -f ins_rdbms.mk rac_on

$ make -f ins_rdbms.mk ioracle

至于grid部分，我们可以删除，也可以保留。

In this Document

APPLIES TO:

Oracle Database - Enterprise Edition - Version 10.2.0.3 and later
Oracle Solaris on SPARC (64-bit)
***Checked for relevance on 15-Jun-2012***
***Checked for relevance on 27-Aug-2014***

SYMPTOMS

While starting up the database, Getting error ORA-29702

Messages in the Alert Log file:

Cluster communication is configured to use the following interface(s) for this instance  192.168.193.2 
Wed May 23 16:00:55 2007 
cluster interconnect IPC version:Oracle UDP/IP (generic) 
WARNING: Failed to set buffer limit on IPC interconnect socket Oracle requires that the
SocketReceive buffer size be tunable upto 1MB
Please make sure the kernel parameterwhich limits SO_RCVBUF value set by applications is atleast 1MB
MMNL started with pid=15, OS id=14908 
Wed May 23 16:00:56 2007 
starting up 1 shared server(s) ... 
USER: terminating instance due to error 29702 
Instance terminated by USER, pid = 14841 

CAUSE

RDBMS install is not running on RAC but still it is assuming it to be RAC instance.

Cluster information required according to alert.log

SOLUTION

1. Shutdown the database completely.

2. Relink with RAC OFF :

$ cd $ORACLE_HOME/rdbms/lib

$ make -f ins_rdbms.mk rac_off

$ make -f ins_rdbms.mk ioracle

3. Startup the database.