Disk group with external redundancy

[satish]

Hi Erman,

We have 2 node RAC with 19.28 version. We have a query.

We have a disk group with external redundancy. We understand, the kfed utility is an Oracle-supplied tool used to read, write, and repair ASM disk headers. A backup copy of the header is automatically saved by Oracle within the same physical disk, typically at a different block location (e.g., block 510). If the internal backup copy is also corrupted, or if the corruption extends beyond the header (into data allocation units), the kfed repair method will not work. In this case,what options, do we have to proceed. Please share your expertise.

Thank You,
Satish

[ErmanArslansOracleBlog]

Contacting the Storage Admin may help  (since you are using External Redundancy , Storage admin may help you -- using backup snapshot and stuff like that -- so storage admin may restore a snapshot lun and then help you on the recovery in the storage level -- raid integrity and rebuild if needed)

You may use RMAN -- as you may already guess. (Recreate diskgroup + DB restore)
 
You can create an Oracle Support SR, and consider using Oracle AMDU (ASM Metadata Dump), and extract the datafiles, and then register them and reuse them with a proper disk group. This may be an option.

You can create an Oracle Support SR , and consider using kfed to copy a header of good healthy to this failing disk, and then manually edit it in place to make it work.. (this is last resort, and requires lots of skills and low level knowledge)

So,

Storage Snapshot - first choice
RMAN Restore Up to last Archivelog - standard choice
AMDU Extraction -- If no backups exist
Manual kfed Edit -- only as a last resort

There are some notes , blog post that may give you insights on this.. For instance: https://oraclehandson.wordpress.com/2010/08/06/cloning-oracle-asm-databases-with-kfed/

[satish]

Thanks a lot

asm list disks show nothing. might be headers are corrupted.

We want to recreate diskgroup and restore rman backup.
How can we know which disks are part of which diskgroups,

Thanks,
Satish

[ErmanArslansOracleBlog]

Use kfed read for each disk and check kfdhdb.grpname attribute(s)