>
> I've found its best to do the pre and post stuff seperate from amanda,
> as when amanda is running, it has a few file locks and those files
> cannot be backed up by amanda so you lose the most vital info of the
> backup, what is it and where is it.  My method works without this
> limitation.  In my case the only 'pre' is a run of amcheck just so
> the rest of my script's are looking at the correct tape as obtained
> from the tapelist.  My post consists of tarballing up the config and
> indice containing dirs, and appending them to the end of the tape
> after making sure the tapetype is set small enough to leave room on
> the tape for them.   In this manner the exact config and a truely
> uptodate indice of all the tapes in the tapelist, is right on the end
> of the tape.  In my case thats about another 375 megs.  On disk its a
> <shrug>, on tape you'll need to compensate because they really are
> fixed length.  To recover, I play those last two files back with dd
> into tar, and the rest of it can be fairly automatic from there.
>
> At least thats the theory, practice occasionally differs as we all
> know so well. :-)

ok, i have to do some more work, i have to stop a lot of services, then i
have to run "getfacl --skip-base -r mypath | gzip > myacls.gz" to backup
the acls or i have to use star which backup the acl's. i would use gnutar
because it's easier to restore but my getfacl command needs 50 minutes to
run, thats to long. the post script only restart the services but to add
the database stuff at the end of the tape is a very good idea. please
could you send me your pre and post script? then i can see which files
you're appending at the end of the tape.

claus