none
tape drive cleaning RRS feed

  • Question

  • I'm wondering if anyone can provide an explanation of how DPM determines if it needs a tape drive cleaning.  Is it a process of the number of hours a tape drive is writing data?

    I have numerous jobs that seem to fail periodically because DPM thinks it needs to clean the drive.  I've created a powershell scheduled task to automatically clean the drive once a month, but it seems like the clean isn't recorded, as after the scheduled task runs, it may or may not actually show the correct date for the last time run.  Even thought the drive was cleaned less than two weeks ago, again a failed job because DPM thinks it needs to clean the drive somewhere along the multi-TB backup to tape we do.  This drive in is LTO-4.  Another tape library with LTO-2 drives has never asked for a cleaning....ever.

    Monday, January 10, 2011 5:31 PM

Answers

  • In this case the library auto clean is happening under the hood and DPM is just not aware that drive is busy. When DPM is trying to mount a media into the drive, it would receive an error and backup fails. DPM tries to run a corrective action job of FI and retrigger of tape backup job immediately

    I don't see an alternate way to solve the problem other than mutually exclusive schedules. Is it possible to fine tune the schedule of libraries auto clean job?

     

     


    Thanks Praveen T [MSFT] This posting is provided "AS IS" with no warranties, and confers no rights.
    Friday, February 11, 2011 6:50 AM
    Moderator

All replies

  • Hi,

    The tape library / drive usually sets a flag when the drive requires cleaning. How the library / drive does this may vary from hardware vendor to the next, it may be time based or based on soft errors. Backup software typically queries the flag and if set will report the drive requires cleaning.   Libraries that support self cleaning will clean the drive automatically if you have a cleaner slot, but DPM needs an operator to inactivate the alert. 


    Regards, Mike J. [MSFT] This posting is provided "AS IS" with no warranties, and confers no rights.
    Tuesday, January 11, 2011 12:36 AM
    Moderator
  • I guess the real problem is how to alleviate DPM's worry about the drive.  The library supports self cleaning but DPM doesn't recognize that, so hence the schedule to try to sneak a cleaning in before the large backup jobs happen.   There are backup jobs that literally take days to go to tape, as DPM has to dump an entire replica to tape. The problem is halfway through the backup DPM may think it needs to clean the drive, and then this stops the backup.  Since we had forced a clean less than two weeks ago, I'm trying to determine if DPM is using hours, errors..or something else when it comes up with the great idea the drive needs cleaning.

    Is there a log or something that might capture this?

     

    thanks

    Thursday, January 13, 2011 7:54 PM
  • You can get the exact error details  from the MSDPM*.errlog in path

    c:\Program Files\Microsoft DPM\DPM\Temp (for DPM2010)

    c:\Program Files\Microsoft DPM\DPM (for DPM2007)

    Key word for search can be - "NeedsCleaning".

     

    By looking at your explanation i guess the alert is getting raised due to the IO errors from drive.(which did not happen on LTO2)

    so can you please test the LTO4 drive for IO errors outside of DPM backup/ your vendor tool may be?

     

     

    Thanks,

    Nikhil[MSFT]

    This posting is provided "AS IS" with no warranties, and confers no rights.

     

    Monday, January 17, 2011 6:50 AM
  • We ran into cleaning issues on our TL4000 libraries w/ LTO5 drives.  Dell consulted with IBM and determined that their LTO5 drives will require cleaning after 40 motion hours.  For us, this means a drive needs to be cleaned every 2 weeks or less (full backup once a week, so every 2 tape RPs).

    Our libraries are configured to auto-clean when needed, which isn't a problem.  The problem is that DPM will simply fail the job writing to the drive that is being cleaned instead of pausing the job, waiting for the clean to finish, and then continuing.  Simply pausing the job and waiting for one of the other tape drives to become available would be nice. 

    It isn't the end of the world, but when you are writing multiple 5TB jobs to 4 drives and it fails one of the jobs 10 hours in, it gets annoying. 

    We are moving to a proactive cleaning schedule instead of waiting for the drives to request cleaning.

    Saturday, January 29, 2011 3:52 AM
  • In this case the library auto clean is happening under the hood and DPM is just not aware that drive is busy. When DPM is trying to mount a media into the drive, it would receive an error and backup fails. DPM tries to run a corrective action job of FI and retrigger of tape backup job immediately

    I don't see an alternate way to solve the problem other than mutually exclusive schedules. Is it possible to fine tune the schedule of libraries auto clean job?

     

     


    Thanks Praveen T [MSFT] This posting is provided "AS IS" with no warranties, and confers no rights.
    Friday, February 11, 2011 6:50 AM
    Moderator