On Thu, 20 Feb 2003 at 2:47pm, rwong AT mrs.mni.mcgill DOT ca wrote

> Another observation I noticed while examining the backups from 3-4 days 
> ago.
> 
> It appears that the first tape does contain data.  The only spot that 
> doesn't have data is file 1.  The tape header is there, file n >= 2 appear 
> to all contain backups.  This would explain why the subsequent tapes may 

That's normal.  File 1 is the tape label.  Files 2+ are backups, the first 
32k of which is the header.

In looking over the script of unix tool use you sent, there are a lot of 
typos.  I'm not sure if that's just in the transfer or what, but the 
commands you want to run are:

mt fsf 1 (gets you to start of first backup image)
dd if=/dev/nst0 of=backup.image bs=32k skip=1

That will put the (tarball/dump image) in backup.image after skipping over 
the amanda header.  Subsequent runs of dd will do the same for each 
successive image.

> end up empty, because (according to amtoc) it attempts to write that huge 
> file to the 2nd tape to the file 1 marker position which it can't seem to 
> do (it as in amanda? the vxa-1 system? I dunno which...).

Every tape should have a label as the first file.

> If I attempt to access file markers greater than 1, I don't have any 
> problems and there are no syslog messages.  But if i attempt to access 
> file 1, it keels over and I get messages in syslog.  I guess this is why 
> amrestore is crapping out.  This is good in the sense that the backups are 
> occurring.  But the amanda is setup to dump the smallest filesystems first 
> and then the largest.  which means the level 0 dumps for 2-3 file systems 
> aren't happening.
> 
> Would 2.4.4p1 actually help?
> 
> I am going to try tarring to a blank tape manually and see what happens.

This will answer the question regarding 2.4.4b1.  Make sure to tar 
multiple times.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University