By hamsun on Jun 20, 2013
ACFS Replication, new features in 188.8.131.52 enables replication of Oracle ACFS file systems across the network to a remote site, providing disaster recovery capability for the file system.
For this demonstration, I used the practice environment provided for the course Oracle Grid Infrastructure 11g: Manage Clusterware and ASM.
Because the disk space is limited and 4 GB is needed per node, I used only a two Nodes cluster, named host01 and host02.
1. Before you start, be sure your cluster is correctly configured and works
2. Using ASMCA, create the disk groups with the required attribute.
a. Create PRIM Diskgroup with external redundancy and use 5 ASM disks ORCL:ASMDISK05
to ORCL:ASMDISK09. Set ASM and ADVM Compatibility to 184.108.40.206.0
b. Create SEC Diskgroup with external redundancy and use 5 ASM disks ORCL:ASMDISK10 to ORCL:ASMDISK14. Set ASM and ADVM Compatibility to 220.127.116.11.0
3. Using ASMCA, create the volumes needed.
a. Create Volume PRIM on PRIM Diskgroup and use 11G Bytes for the volume (Remember
4 GB per node)
b. Create volume SEC on SEC Diskgroup and use 11G Bytes for the volume (Remember 4 GB per node)
4. Using Linux, create the required directories
a. On Host01, create the directory /prim
a. On host02, create the directory /sec
5. Using ASMCA, create the ASM Cluster File Systems needed:
a. Create an ASM Cluster File System in PRIM Volume and specify /prim as Mount Point
b. Create an ASM Cluster File System in SEC Volume and specify /sec as Mount Point
6. Using NETCA, create TNS Alias PRIM for PRIM Service on host01 and create TNS Alias SEC for SEC Service on host02. (do not specify Cluster-scan)
7. Verify the result in TNSNAMES.ora file
8. Using Linux, create a password file on host01 with oracle as password
9. Using Linux, create a password file on host02 with oracle as password
10. On Host01, using SQL*Plus, create an user named oracle with password oracle and grant the necessary privileges. The password file will be updated on both sites, Host01 and Host02
11. On Host01, connected on instance +ASM1, add a service PRIM
12. Verify the listener on host01 is listening for service PRIM on instance +ASM1
13. Test the service name PRIM
14. On Host02, connected on instance +ASM2, add a service SEC
15. Verify the listener on host01 is listening for service SEC on instance +ASM2
16. Test the service name SEC
17. Using Linux, using df on host01, check the file system /prim is mounted and dismount the unnecessary file system mounted (/sec)
18. Using Linux, using df on host02, check the file system /sec is mounted and dismount the unnecessary file system mounted (/prim)
19. Connected on host02, using acfsutil, initialize ACFS replication on standby site first
20. Only when the replication has been successfully initialized on standby site, connect to host01 and initialize ACFS on primary site using acfsutil.
21. Check ACFS Replication on primary site
22. Check ACFS Replication on standby site
23. On host01, create a file (named toto)
24. On host02, after a while, check if the file has been replicated
25. On host01, pause the replication and create a second file (named titi)
26. On host02, the new file is normally not replicated because replication is paused on host01
27. After a while, resume replication on host01
28. After few seconds, on host02, the new file is normally replicated
29. On host02, pause the replication
30. On host01, create a new file (named new)
31. On host02, the file is not created because the replication on standby site is paused
32. On host02, resume the replication and test if the new file is created
33. On host02 standby site, try to remove file
34. On host01 primary site, terminate and verify the replication status
35. On host02 standby site, terminate and verify the replication status
About the Author:
Bruno d’Agostini joined Oracle in 1994 and currently is a French Instructor for Oracle University specializing in the Oracle Database, Data Guard, Data Warehouse, Streams, GoldenGate, Parallelism, Partitioning, Clusterware, ASM, RAC and Exadata Database Machine.