Step by Step Procedure

Please follow below steps for converting the RAW devices to ASMFD:

1. As a GRID user run below command to generate the which ASM Disk is associated with the OS disk:

$ sqlplus / as sysasm

set lines 200
col path for a30
col name for a30
col header_status for a30
select name, path,header_status from v$asm_disk;

2. As a Grid Infrastructure (GI) software owner, get the existing ASM diskstring value:

grid@cehaovmsp1129 ~]$ asmcmd dsget

3. As a GI software owner update the Oracle ASM disk discovery string to enable Oracle ASMFD to discover devices for migartion.

[grid@cehaovmsp1129 ~]$ asmcmd dsset '/dev/rdsk/*','AFD:*'

4. As a root user stop the cluster accessing the shared ASMLib disks:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/crsctl stop crs -f

5. As root, configure Oracle ASMFD to filter at the node level:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/asmcmd afd_configure

6. As the Oracle Grid Infrastructure / root owner verify the status of Oracle ASMFD:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/asmcmd afd_state


7. As a root user, with the help of the data gathered in step 1, migrate (label) all the RAW disks:

$ORACLE_HOME/bin/asmcmd afd_label --migrate

You can run below command to generate the automatic syntax:

e.g

SQL> select 'asmcmd afd_label '||ltrim(NAME)||' ' disk,PATH||' --migrate' path from V$ASM_DISK;

DISK PATH
--------------------- ------------------------------
asmcmd afd_label DATA_0004 /dev/xvdg1 --migrate
asmcmd afd_label DATA_0003 /dev/xvdf1 --migrate
asmcmd afd_label DATA_0002 /dev/xvde1 --migrate
asmcmd afd_label DATA_0001 /dev/xvdd1 --migrate
asmcmd afd_label DATA_0000 /dev/xvdc1 --migrate


8. Validate that all the disks have been migrated to the ASMFD:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/asmcmd afd_lsdsk

9. Stop the ACFS as a root user:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/acfsload stop

10. Stop the AFD as a root user:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/afdload stop

11. Start the ACFS as a root user:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/acfsload start

12. Start the AFD as a root user:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/afdload start

13. Validate the status of the disks once again:

[root@cehaovmsp1129 ~]# $ORACLE_HOME/bin/asmcmd afd_lsdsk

14. Check the disks details are populated under "/var/opt/oracle/oracleafd/disks/" folder.

15. Repeat all the except steps no. 7 (migrating the disks) on all the nodes of the cluster.

16. As a GI software owner update the Oracle ASM disk discovery string to enable Oracle ASMFD to discover devices in future

[grid@cehaovmsp1129 ~]$ asmcmd dsset 'AFD:*'

17. Re-Start the CRS as root user. This will bring the CRS and the databases automatically.


from support.oracle.com 발췌