2009-01-23 09:16 AM
I've setup my database volumes/LUNs/logical volumes/filesystems in the following manner, in an attempt to get a SMO fast restore (w/SnapRestore) to work. Alas, I have failed. Here is the setup:
FilerB: /vol/db_ora_data/lun2 hdisk5 fcs0 FCP 50g (53687091200) GOOD
FilerB: /vol/db_ora_data/lun1 hdisk4 fcs1 FCP 20g (21474836480) GOOD
My Oracle data files all sit on hdisk5, my control files, redo logs, archive logs and temp spaces all sit on hdisk4.
The SMO backup successfully snaps the data on hdisk5, but gives me the following error on the hdisk4 data:
--SMO-11034 [ERROR]: LUN "NS5200B:/vol/db_ora_data/lun1" contains partitions corresponding to "/dev/hdisk4" and "/dev/hdisk4". Cannot use multiple partitions on the same LUN - please reconfigure.
This error leads me to believe I will need distinct NetApp volumes (and thus hdisks) for EACH filesystem that holds the archive logs, redo logs, control files and temp spaces. This would be an additional eight volumes -- which leads to eight new hdisks and eight new AIX volume groups -- which is an administrative overhead I'd prefer to not have to do.
Is there another way? Am I mis-interpreting this error?
2009-06-01 12:38 PM
Historically, SMO required databases to be laid out in a single filesystem per volume group manner. With SMO 3.0, this is no longer the case.
I would suggest taking a look at the SMO 3.0 Best Practices Guide to start:
Specifically, check out the sample database volume layouts in Appendix L as well as the section on "Fast Restores" (volume snap restore - which is also a new feature supported in SMO 3.0).
SMO 3.0 for AIX requires SnapDrive for Unix 4.1.