NDMP attached Device markes tapes as bad (4047730)

Title: NDMP attached Device has tapes appear as bad
Date: 07/2007
NV Version: 745
OS Version: all
Application version:
Plugin version: NDMP V65

Description:
This article talks about how NDMP attached devices can not switch block sizes when reading tapes and report the tapes as BAD.

Symptoms:
This problem can occur when a tape device has been swapped as the library is removed and added back again.

The Binary log shows an error but the error appears to be a read error on the drive.
Extract from binary log:
SCSI :2368 16 0 144026 CMD STATUS 'CHECK CONDITION' [2]
SCSI :2368 17 0 144026 CMD SENSE KEY 'MEDIUM ERROR ' [3] - ASC 00 ASCQ 00
SCSI :2368 18 0 144026 CMD SENSE INFO 'NO ADDITIONAL SENSE'
SCSI :2368 20 0 144026 CMD RETURN VALUE 2
SCSI :2368 14 0 144026 CMD MODE SELECT [ 15 00 00 00 0c 00 ]
SCSI :2368 16 0 144026 CMD STATUS 'GOOD' [0]
SCSI :2368 20 0 144026 CMD RETURN VALUE 0
TAPESCSI:2368 69 0 144026 Tape block size set to 0

If diagnostics are enabled on the tape drive by the filer you see the following in the messages file:
Mon Jul 23 15:40:26 BST [MSVUKLHDFL00001: tape.cmd.bCntDbg:error]: Tape block was longer than byte count on tape device 0d.125. Requested = 32768, Read = 262144
Mon Jul 23 15:40:26 BST [MSVUKLHDFL00001: tape.cmd.chkCondDbg:info]: Tape device 0d.125: Check Condition: SCSI Op Code Read(06) (CDB 0x08: 0x008000 bytes): no sense - No additional sense (0x0 - 0x0 0x0 0x0).