Forum Discussion

Andy_Cam's avatar
Andy_Cam
Level 3
8 years ago

SQL Backup Using Alternate Tapes

Hi, I have a MS-SQL-Server policy configured to backup 30 databases on one SQL server.

The backups run fine apart form using two tapes. The parent job kicks off and runs the first database backup on one tape. When this is complete, the second database backup starts but insists on using a different tape drive and a different tape. When this backup completes, the third database backup starts and uses the same drive and tape used for the first database backup.

The problems I have is the whole backup takes along time loading and unloading tapes and uses two tapes rather than one.

Multiplexing is set to 1 on the Storage Unit and the Schedule. All SQL Policies have the same problem.

Anyone any ideas?

7.6
windows server 2012
  • Andy_Cam's avatar
    Andy_Cam
    8 years ago

    Hi Michal,

    I found a solution to this problem after a lot of faffing around. The two drives that were being used for these backups were marked in NetBackup as 'Needs Cleaning'. I ran D:\Program Files\VERITAS\Volmgr\bin\tpclean -M on the two drives the fault disappeared.

    (The drives are in an IBM3500 library which manages it's own drive cleaning but for some reaseon NetBackup assumed they still needed tpclean running.)

  • Tape_Archived's avatar
    Tape_Archived
    Moderator
    8 years ago

    Well, if you want to use only one Tape at a time this can be achived by various ways. You can set "Maximum Concurrent write Drives" on the storage unit to 1. So it depends what you really want to acheive. 

    Regarding mount and unmount times, if its within a mintue or two i would not worry. Netbackup does not umount the Tape unless that Tape is really FULL or other tape type or other Retention Tape is required, it tries to manage efficiently to reduce more mounts and umounts.

    • Michal_Mikulik1's avatar
      Michal_Mikulik1
      Moderator
      8 years ago

      Hello,

      trying to resolve the same problem with a Sybase database. There are 20 dbs in the script and each one wants another tape then previous one, which is in summary about 30 minutes dismount/remount delay for all databases backup.

      Since many versions ago, there is a parameter Media Unmount Delay (default 3 mins) in NetBackup which is intended for these situation. So after certains types of backup (user initiated of db application backups) tape remains in a drive, waiting for another db backup job, thus eliminating dismounts/remounts.

      But it does not seem to work (it is a Master 8.0 but the same behaviour was with 7.7).

      I am about to review the logs and maybe open a call..

      Regards

      Michal

       

      • Andy_Cam's avatar
        Andy_Cam
        Level 3
        8 years ago

        Hi Michal,

        I found a solution to this problem after a lot of faffing around. The two drives that were being used for these backups were marked in NetBackup as 'Needs Cleaning'. I ran D:\Program Files\VERITAS\Volmgr\bin\tpclean -M on the two drives the fault disappeared.

        (The drives are in an IBM3500 library which manages it's own drive cleaning but for some reaseon NetBackup assumed they still needed tpclean running.)