There are a few rationales behind the change.
1. To provide a fix to allow proper handling of a cleaning tape by DLT tape
libraries (Burt #29835.)
2. In the case of a stand-alone drive, to provide a consistent drive
behavior across various tape drive types. AIT and Exabyte drives will
physically eject a tape when an unload command is issued. This change will
make the DLT drives respond in a similar way, even though it does not
physically eject the tape.
3. The change will require a user to manually reload the tape to enable the
DLT drive to read it. This prevents accidentally overwriting a tape user
may otherwise think has been unloaded. Again, this is no different from the
AIT and Exabyte drives.
Also, the change should only affect stand-alone DLT drives and not DLT
stackers. Once the stacker robotics unloads and loads a new tape into the
drive, a LOAD command will be issued to put the drive in ready mode.
In your particular case, to add incremental dumps, an alternative is to use
nrst0a instead urst0a. If you don't really care about the previous data
on the tape, you can even use rst0a.
C.P. Hu
FSR QA, Network Appliance
-----Original Message-----
From: Chris Thompson [mailto:cet1@cus.cam.ac.uk]
Sent: Monday, December 11, 2000 7:21 AM
To: toasters(a)mathworks.com
Subject: Relying on automatic load on DLT7000 drive
Maybe it's a mistake having put myself on the weekly newbugs mailing
list: examining the list always leads to alarm and despondency... :-)
This week it's 29835. That describes a problem with cleaning tapes
and issuing the SCSI LOAD command against a DLT7000, and has
| Solution or Fix:
| The filer will refrain from issuing the LOAD command to the DLT7000
| at all, since the device autoloads the tape. Available in 5.3.7R2,
| 6.0.2 and 6.1.
That sounds as if NetApp is about to remove a feature (accident?) which
I think I have been relying on. We have a (Tandberg) DLT7000 drive with
no stacker attached to an F740. It certainly "autoloads" when a tape is
inserted. However, after an unload, the same tape can be reloaded without
manual intervention by accessing it from the F740 - I presume that's
the result of the LOAD command at issue. I have used this for adding
incremental dumps at times when there's no human available to change
tapes (e.g. weekends).
Apart from that, there's a positive advantage in the idle state of the
drive being the unloaded one. If the tape is loaded, a power outage will
leave the tape directory marked invalid, and the tape controller will
need to scan the whole tape for filemarks before it can do anything
with it, which can take well over an hour. (Not that this has happened
to us on the F740: it's based on experience with DLTs in other contexts.)
Chris Thompson University of Cambridge Computing Service,
Email: cet1(a)ucs.cam.ac.uk New Museums Site, Cambridge CB2 3QG,
Phone: +44 1223 334715 United Kingdom.