Data Loss Possible When Using FastBack Server Deduplicated Repository
An IBM Tivoli Storage Manager (TSM) FastBack deduplicated repository can become corrupted when a memory allocation error occurs.
The TSM FastBack deduplication service process attempts to obtain the contiguous memory it needs via a memory allocation command. If that memory allocation command fails, FastBack still continues with its processing, causing the FastBack deduplicated repository to become corrupted. Once a local FastBack deduplicated repository has been corrupted, the repository must be rebuilt. APAR IC89143 was opened for this issue.
- FastBack 126.96.36.199 through 188.8.131.52
Conditions for the problem to occur:
Both of the following conditions must occur:
- An affected FastBack release is used with a FastBack server deduplicated repository
- The FastBack deduplication service does not have enough contiguous memory available
How can you tell if you are affected?
Any customer using a FastBack server deduplicated repository with an affected release could potentially be affected.
There is not a consistent set of error messages after the repository has been corrupted.
Use the local FastBack repository in standard mode (NTFS repository) and not as a deduplicated repository.
Recommended alternatives to deduplicated repositories:
- Use NTFS compression to reduce the repository usage of the FastBack server. When Disaster Recovery (DR) is not required, additional data reduction is achieved by setting the Cleanup level to “High”.
- Use other data reduction technologies such as hardware storage deduplication.
Beginning with the FastBack 184.108.40.206 interim fix, processing will stop immediately when memory cannot be obtained, preventing corruption of the deduplicated repository.
|FastBack release levels affected||First level with fix|
|220.127.116.11 through 18.104.22.168||22.214.171.124
available in Fix Central