In NetBackup version 7.5.0.x, DSSU (disk staging storage unit) cleanup may not function, triggering backup failures due to the disk becoming full.
When this issue occurs, the nbemm log will show the following scenario:
The SQL query is initiated using the virtual name of the server:
[ImageObject::computeWhereClause] sql = <( C.MediaServerKey = 1000002 )>, retval = <0>
[ImageObject::computeWhereClause] sql = <( C.MediaDescription = 'VIRTUAL_NAME' )>, retval = <0>
*** When reviewing a dump of the database, the fragment table shows the active node instead of the virtual name for the image fragments.
The sql query returns 0 records in fragment state 3 (ready to be deleted):
[ImageObject::computeWhereClause] sql = <( C.FragmentState = 3 )>, retval = <0>
In the case where a disk staging storage unit is attached to master in a clustered environment, when expmgr searches for expired back up images to delete, it considers only the active node for searching the images. The result is that query returns 0 fragments and no images are removed from the disk.
The formal resolution for this issue (Etrack 2799175) is included in the following release:
- NetBackup 7.5 Release Update 5 (22.214.171.124)
NetBackup 126.96.36.199 is now available - please access the Related Article below for more information.
Manually deleting the images can be done to remove the files on the server. If that is unacceptable, an Emergency Engineering Binary (EEB) replacement binary for nbemm is available from Symantec Technical Support. A support case should be opened that references this document and Etrack 2799175.
Clustered master servers that have the following configured:
- Are at NetBackup version 7.5 or above
- Have a disk staging storage unit configured