On Mon, Mar 14, 2005 at 10:37:11PM +0000, Bruce S. Skinner wrote:
> 
> My last dozen or so emails to amanda-users have gone into a black
> hole, I'll try this from another network under another subject.
> 
> 
> So let's try and walk (crawl?) before we run.
> 
> I've set things up with only one small disk to be backed up, set
> dumpcycle 0 and removed the holding disk and tape changer from the
> config.  It's still failing.  There is a signal 13 (PIPE) error in
> sendbackup.debug.
> 

This will likely not assist you, just for information sake.

Most amanda admins don't realize that when indexing is on
there are actually two tars that run.  One does the actual
creation of the archive.  The output of this command is
duplicated (think of the unix tee command) with one copy
going to the holding disk or tape drive as appropriate.
The second copy of the newly created archive is send to
another tar which reads through it and creates a "table
of contents" which is the index.

It has been my impression from the posted articles that
about 90% of the tar pipe errors are from this second
tar, the one creating the index from the archive, not
the tar creating the archive itself.  I'd be happy to
be told I'm all wrong, but it seems to be a fragile
part of amanda that manifests itself inconsistantly
and in such a way as to not be analyzable (sp?).
People who have the problem seem to fumble around
making changes and it goes away without them being
able to point to a specific reason for the repair.

-- 
Jon H. LaBadie                  jon AT jgcomp DOT com
 JG Computing
 4455 Province Line Road        (609) 252-0159
 Princeton, NJ  08540-4322      (609) 683-7220 (fax)