On Mon, 8 Jan 1996 10:11:00 PST Paul Esson writes:

>Subject: Recovering Damaged Files
>
>We recently upgraded to version 2 of ADSM for AIX and have spent the last
couple
>of weeks backing up out primary tape storage pool for disaster recovery
>purposes.
>
>All was well until we hit a problem with  two particular tape volumes
A00319
>and A00320 which  caused the backup of the storage pool to terminate, the
>messages in each case were as follows:-
>
>ANR 9999 D asrtrv.c(471)
>End reached prematurely on volume nnnnnn
>
>ANR 1218 E
>Excessive read errors on nnnnnn
>
>When I queried the volumes there were indeed significant read errors.  I
then ran
>audits on each of the volumes, fix=no, which confirmed that approximately
99%
>of the files on each tape were damaged.  Further audits of the two volumes
either
>side of the failing tapes, i.e. A00318 and A00321 also indicated extensive
numbers of
>damaged files.
>
>The drives in question are 3490E's, part of a C22 tape susbsystem.  There
>are no corresponding AIX hardware error reports
>
>I next tried an image copy of one of the volumes using the AIX tcopy
>utility. This copied 80 bytes of the first record on the tape and then
recognised an end
>of tape condition.  Running tcopy against each of the volumes which failed
the
>audit produced exactly the same results, end of tape being reached 1 record
(80
>bytes) into the tape.
>
>I am currently trying to skip past these end-of-tape markers to see if I
can copy any
>valid data off the tapes, though I am unsure as to whether ADSM could read
any
>such image copy? Don't know if the label will copy or is corrupt for a
start.
>
>Has any one experienced anything remotely similar? [ I have placed a call
>with IBM UK AIX support.]
>
>I now find myself in the situation where I cannot continue the tape pool
>backup until I resolve the issue of the damaged files on these volumes.  If
I run
>an audit volume with fix=yes I suspect all database references to the files
in these
>volumes will be deleted and that will be that.
>
>Can anyone suggest an alternative plan for recovering the files ?  Any
>comments would be much appreciated.
>
>Yours,
>Paul Esson
>SAIC UK
>c/o BP Exploration
>Aberdeen  UK
>+44 (0) 1224 834418
>essonpa AT xxeupo13.hou.xwh.bp DOT com

Paul,
I experienced a similar problem at a previous site.  The problem with your
tapes may not be physical, but logical.  That is, the data on the tape does
not match what ADSM thinks is on the tape.  I would strongly suggest that
you have your IBM support put you in touch with ADSM support out of Tuscon
before you run AUDIT on the tapes.  They will want to know what level or
your AIX 3490 tape drivers.  Apparently this type of error, has been
observed with older versions of the AIX 3490 drivers.

In our case, the errors were recieved because the block count for one file
did not match the block count that ADSM expected.  This happened on a file
that spanned two tape volumes.  Once I deleted the ADSM entry for this one
file, I was able to access the other files on both tapes.  I then moved the
files on these tape volumes to other tape volumes  using the MOVE DATA
command.

I hope this helps.

DanC
=============================================
Daniel G. Crouse
Software Information Services, Inc.
Email:  Dan.Crouse AT MAILPORT.DELTA-AIR DOT COM
Voice:  800-233-6838
=============================================