Tape with error - MOVE DATA reports very high values

Hi everybody,

An LTO tape has developed errors. I issued a MOVE DATA to empty the tape as
much as possible. After that, I intended to to a RESTORE VOLUME. (The tape
is in a primary pool)

During the MOVE DATA, there were unreadable files and unreadable bytes.
This was expected, but what I didn't expect was the following:

The volume has a "Full" status, with an estimated capacity of 335,865.3 MB
(99,5 Pct Util before the MOVE DATA). After issuing the MOVE DATA command,
I regularly did a QUERY PROCESS command to see how far it was.

After some time, the Unreadable Bytes counter was *way* high, in the 900 GB
(no typo) range. Also, the Unreadable Files counter was well above 10,000.

I do not understand the Unreadable Bytes counter being that high. After the
MOVE DATA finished, there were files left on the tape, only much less than
reported by the Unreadable Files counter displayed by the QUERY PROCESS:
the command SELECT (*) FROM CONTENTS WHERE VOLUME_NAME='815ACP' shows that
only 114 files are left on the tape.

In the log, there were multiple (well over 150 for the files I sampled)
entries for the same unreadable file:

04/08/03 14:30:31     ANR1161W Move Data skipping damaged file on volume 815ACP:
                       Node GONZO, Type Backup, File space /oracle, File name
                       /8.1.7/8174/Disk1/stage/Patches/oracle.p2k.ott/8.1.7.4.0-
                       /1/DataFiles/ lib.0.0.jar.

This exact message occured 162 times, for example.


I suspect that each time this message is reported, the *entire* size of
this file gets added to the "Unreadable Bytes" counter, and that this file
has contributed 162 times to the "Unreadable Files" counter. I also suspect
that the reason for the multiple reportings of the same file is because the
files are "fragmented" on tape. (but 162 fragments of one file would be a
but much, wouldn't it? According to the CONTENTS table, the AGGREGATED
value for this file is 34/65 and the filesize is 42153627 bytes.)

Is my suspicion correct, and if so, is that behaviour correct?


We run TSM Server 5.1.6.3 on AIX 4.3.3 ML10.

Thanks,
--
Jurjen Oskam

PGP Key available at http://www.stupendous.org/