This behavior has been reported in all Backup Exec versions starting with 12.0. These errors have been reported when running backup and utility jobs; such as erase jobs.
The reason these soft write errors are logged is caused by how LTO drive vendor counts errors differently. Backup Exec calculates soft-write errors using a certain T10 SCSI Specifications Log Page and Parameter in one of two ways.
1. Prior to April 2009 Device Driver Installer (DDI20090415) for Backup Exec 12.5
Backup Exec performed a calculation based off of the following T10 SCSI Specifications Log Page and Parameters:
0002h - Log Page 0x02, Parameter 2 Total (e.g., rewrites or rereads)
0003h - Log Page 0x02, Parameter 3 Total errors corrected
Backup Exec would treat the larger of these two numbers as the number of soft-write errors for all tape drives, and use that number to represent the number of soft-write errors.
NOTE: For versions 12.0 and older, this is still how these numbers are calculated for all tape drives.
For Backup Exec 12.5 users: Installing the latest Device Driver Installer (DDI) through Live Update will contain the patch which alters Backup Exec to log only the number from Log Page 0x02, Parameter 3.
Issues addressed by Hotfix 322898 (DDI20090415 Device Driver Installer)
2. Post April 2009 Device Driver Installer (DDI20090415) for Backup Exec 12.5
The reason there are still a high number for soft-write errors reported on Quantum LTO drives after applying the latest DDI is due to how Quantum LTO drives perform "Read After Write" checks. The following excerpt is from Quantum Hardware Technical Bulletin 8706:
Backup Exec collects LTO drive soft write errors using SCSI Log Page 0x02, Parameter 3 which is defined by T10 as "Total errors corrected. This is total write retries less total unrecoverable write errors."
Each LTO drive vendor counts errors differently. On the Quantum drives, Read After Write checks are carried out when each "Sub Dataset" (There are 128 "sub-datasets: in a dataset which contains approx. 1.6 Mbytes of user data) is written. Each sub data set can be re-written up to a maximum of 128 times with each rewrite incrementing the rewrite counter in Log Page 0x02, giving a maximum potential (but almost impossible to achieve as a hard write error would probably occur before this limit was reached) of 128 rewrites x 128 sub datasets = 16384 rewrites per dataset. This very detailed examination of data and the resultant potential for large numbers of rewrites can lead to a high number of soft errors being reported by a Quantum LTO drive.
The entire Quantum article is posted under http://www.quantum.com
under Service and Support
-> Knowledge Base
under Technical Bulletins Article 8706
NOTE: These errors in no way prevent data from being properly written to backup media. Symantec encourages users to enable the Verify option on backup jobs to ensure the integrity of the data being written to the media for all Backup Jobs, regardless of this condition.
See Related Articles Section for more information regarding Verify jobs.