Now that I know the date they want to see the database from, I've tried
(twice) to amrestore the files in a safe location, but with no joy.
After setting the host/disk/path/date to restore, I get the following
from amrestore:

---
amrecover> extract

Extracting files using tape drive /dev/nst0 on host peterbilt.
The following tapes are needed: Daily03

Restoring files into directory /tmp/restore
Continue? [Y/n]: 

Load tape Daily03 now
Continue? [Y/n]: 
Connection to  failed
extract_list - child returned non-zero status: 1
---

and then it dies.  There is an extended pause before the "Connection to
failed" as it goes through the tape.

I did find the faq-o-matic's mention of extract_list in connection
with a different error and, based on that, took a look at
/tmp/amanda/amidxtaped.debug on the tapehost, where I found this:

---
amrestore:   0: skipping start of tape: date 20030626 label Daily03
...
amrestore:  13: restoring peterbilt.__wts_mapcon.20030626.0
Error 32 (Broken pipe) offset 98304+32768, wrote 134520109
amrestore: pipe reader has quit in middle of file.
amrestore: skipping ahead to start of next file, please wait...
amidxtaped: amrestore terminated normally with status: 2
Rewinding tape: done
amidxtaped: pid 14119 finish time Thu Jul 10 10:56:31 2003
---

So am I looking at an irrecoverable backup or is there something I can
do to troubleshoot this?