>... I'd like to back up at least
>their mail, and probably web directories.
>...
>I've been trying to do this with exclude lists, but I haven't hit the
>solution yet. ...
Wow! I wouldn't have had the guts to try this with exclude lists.
They give me (real :-) headaches just trying to do "normal" things :-).
Maybe you could do it with inclusion instead of exclusion? Take a
look at:
ftp://gandalf.cc.purdue.edu/pub/amanda/gtar-wrapper.*
During the pre_estimate you could run "find" to gather a list of what
you want to back up, squirrel that away and then pass it to gtar for
both the estimate and the real run.
FYI, there is also an exclusion test script there:
ftp://gandalf.cc.purdue.edu/pub/amanda/gtartest-exclude
... that might make figuring out the exclude list easier if you want to
continue down that path.
>On that note, it is annoying that if I have an exclude
>file, say /etc/amanda/exclude.home, running amanda with that file
>produces different results than:
>
>tar cvf /dev/null /home -X /etc/amanda/exclude.home
>
>Why is this?
What's different? I assume different files are listed/backed up?
Who does more, Amanda or "cvf"? Is Amanda a subset of "cvf", or are
they all mixed up? Was Amanda doing a level 0? Are you sure you used
the same tar that Amanda does (look at a /tmp/amanda/amandad*debug file
for the tar path Amanda uses on that client).
Amanda runs GNU tar in "incremental backup" mode. Take a look at
a /tmp/amanda/sendbackup*debug file on the client to see the extra
arguments involved. I would expect the lists to be the same if you do
a full dump with Amanda (in which case the listed increment file will
be empty) and with "cvf", but otherwise all bets are off as it would
also depend on what files had been modified.
>--Ray
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
