Run enough tapes through tape drives and eventually the accumulated dust and rubbed of tape particles will affect the drives' ability to read and write. The good news is that all modern tape drives have a read-behind-write design which allows them to detect this condition. When a head-cleaning is required these drives will send out an alert on their SCSI bus.
The response to this SCSI alert depends on how NetBackup and the hosting robot, where applicable, are configured. Read on to see what your options are:

• The SCSI alert, sometimes called TapeAlert, will be intercepted by versions of NetBackup since release 3.2 and an appropriate message will get logged into your system log files by the avrd daemon.
  In a robotic environment, NetBackup will look to see if the library has a cleaning tape of the appropriate media type for this drive. If present, and if the tape has any cleanings left, it will proceed to clean the drive. In the latest (4.5 and up) versions of NetBackup, this feature can be disabled by creating the "touch" file NO_TAPEALERT as:

  /usr/openv/volmgr/database/NO_TAPEALERT
  

  
  Note that this TapeALert mechanism will be in effect regardless of the cleaning frequency you may have set for this drive (see below).
• Robots have an alternate communications link to their drives, usually in the form of a serial cable, which the drives use to signal the robot they are in need of a head-cleaning. If the robot has been loaded with cleaning tapes and is configured to use them, it will go ahead and clean the drive. In some cases this causes problems if NetBackup simultaneously tries to use this drive. Typically this results in a DOWN drive.
• Finally, if you do not wish to rely on the drives detecting their own need for cleaning but force pre-emptive cleanings, NetBackup's tpclean utility can be used to set a periodic cleaning frequency. However, doing so is discouraged as excessive cleaning, done after all with a scouring pad of sorts, will accelerate tape drive head wear. Also, in a Shared Storage Option (SSO) environment, where each media server has its own accounting of how much time the drive has been used, the cleaning frequency setting doesn't make sense at all.

The big hole in all this is once again NetBackup's lack of device management. In a distributed environment with many media servers all across the enterprise, you're going to have to build your own tools to check up on the drives to make sure they are in fact being cleaned; to make sure your cleaning tapes have cleanings left on them; and forget about managing your stand-alone drives.

The NBU modules already come with some scripts to start you on your way. There is robot-status.pl to list the drives in a robot and their cleaning status. Or robot-snapshot.pl which will tell you how many cleanings are left on the the cleaning tapes in the robot. You can easily modify these scripts to suit your needs and of course, we at PBnJ will be more than happy to help you to do so.