NFS Data Backup

Background:

Due to the constrains in production storage area, the data which is accumulated at a high rate, is staged and distributed to SCS NFS storage area.   The data distribution is described in http://www.slac.stanford.edu/grp/cd/soft/unix/distnlctadata.html.
The document here describes an utility developed to provide a backup solution for such data archived in the NFS area using SCS Stacker 9714 DLT robot tape facility. The diagram below shows the components.

Backup and Restore:

There are two procedures: backup-nfs for backup, restore-nfs for restore. The procedures must be run on tapeserv1, a tape server machine, which is running Solaris and maintained by SCS taylor. The basic functions provided by backup procedure include
  • tape server verifing;
  • tape mounting via DLT robot;
  • tape rewinding via mt, magnetic tape control software;
  • data archiving to tape;
  • tape dismounting and ejecting via DLT robot.
  • The synopsis for backup procedure is
    backup-nfs nfs-data-path tape-name
    in which nfs-data-path is an absolute path where data is to be backed up ,  tape-name is a tape name you define.  For instance:

    The functions in restore procedure include:

  • tape server checking;
  • tape mounting via DLT robot;
  • tape rewinding via mt, magnetic tape control software;
  • data restoring;
  • tape dismounting and ejecting via DLT robot.
  • Type  "restore-nfs",  a simple user interface will walk through what data to restore and where to put.
    For instance:


     

    How to use them:

    Since the DLT Robot is owned by SCS and shared among a number of experimental groups in SLAC,  we are not allowed to leave tapes permanetly in the DLT Robot.  So, there is a need for a person to go to SCS farm room on the first floor to insert tape into the DLT or  take out tape when back/restore is done.  The steps for running  backup and restore are lised below:
  • go to SCS farm room (you need get a key from SCS Help Desk);
  • press execute button on DLT;
  • insert tape and close the cap;
  • log in tapeserv1 machine ( you can do so either in the farm room or from your office);
  • run "backup-nfs nfs-data-path tape-name" for backup or "restore-nfs" for restore;
  • tape should be dismounting and ejecting via DLT robot.
  • press execute button on DLT retrieve tape from the cap
  • set tape to write-protect mode, mark tape by tape-name
  • Tape will be dismounted and ejected automaticly if there is any problem. Should tape be stucked for some reasons, you can run procedure tapeout to back out tape. The procedures are in /afs/slac/g/cd/soft/scripts on AFS system.

    Remarks:

    SCS is about to release a new Mass Storage System that uses HPSS mstore facility. With mstore backup procedure
    can be completely automated and restore done in an efficient way. This work is in progress.

    References:

    [1] Kristi Luchini:  "Unix-based Data Storage Requirement"

    http://www.slac.stanford.edu/grp/cd/soft/pepii/slaconly/disk_storage/index.html

    [2] Jingchen Zhou:   "NLCTA Data Distribution"
    http://www.slac.stanford.edu/grp/cd/soft/unix/distnlctadata.html
    "Controls System UNIX Computing Environment and Data Management Facilities"
    http://www-group/cdsoft/presentations.html


    Contact: Jingchen Zhou (X4661, jingchen@slac). Last edited on 05/09/02