On Mon, Dec 12, 2005 at 03:18:24PM +0100, Paul Bijnens wrote:
> Montagni, Giovanni wrote:
> >I'm wondering if there's a command or a way to check if files backed up 
> >were written to tape cerrectly.
> >
> >The amverify command, as explained in the docs, "determines if the tape 
> >is readable, but does not do any internal consistency check on the image."
> 
> But just before it says: "and pipes the output to a restore program (if 
> available) with an option to create a catalogue of the backup.
> 
> For gnutar backups, it passes the image through "tar -t", producing a 
> table of contents, which would trigger quiet a lot of internal errors
> in the file.
> 
> >
> >How can i control if files are written correctly?
> 
> Doing a test restore from time to time.

Giovanni, one thing to consider is that between the time amanda
backs up a file and the time you attempt to verify the accuracy
of that backup, many files are likely to have changed, been
deleted, or added.  Unless you are using file system snapshots
to give you a static file system to back up, file by file veri-
fication is not very feasible.

> 
> Also, remember that, even if the backup is perfect, but you specified
> the wrong top level directory in the disklist, the backup is still 
> useless.  Some errrors can only be caught by doing a test restore.
> 
> Specifying the wrong top level directory is easy!!!  Having a very large
> disk, and splitting it up with DLE's like:
> 
> bigserver.example.com  /bigdisk/[a-m]*  comp-user-tar
> bigserver.example.com  /bigdisk/[n-z]*  comp-user-tar
> 
> gives you the false impression, that you covered the whole alfabet.
> But it omits files beginning with a number, with a dot, with an accented
> character, etc.  (Tip: add an entry that is the opposite of the previous
> instead:  "/bigdisk/[^a-m]*"  is OK.)

If the disk were split into multiple DLEs as your two examples
above, then this third pattern would capture [n-z]* again.
Perhaps a third pattern more like "/bigdisk/[^a-mn-z]*" would
be more appropriate.  I include separate ranges, [^a-mn-z],
rather than a single range "[^a-z]" as it is possible in some
character sets that the two are different (eg. an accented "m"
collates higher than "m" but lower than "n").

Another way of doing this, not better - just different, is to
also backup "/bigdisk" entirely with a separate DLE with a
unique exclude list to omit the things backed up by the other
DLEs.

-- 
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)