On 2017-11-07 13:36, Ned Danieley wrote:
On Tue, Nov 07, 2017 at 01:29:34PM -0500, Austin S. Hemmelgarn wrote:
OK, so you're talking about functionally permanent archiving instead
of keeping old stuff around for a fixed multiple of the dump cycle.
If that's the case, you may be better off pulling the dumps off the
tapes using amfetchdump, and then uploading them for there. That
use case could in theory be handled better with some extra code in
Amanda, but I don't know how well the lack of deletion would be
handled on Amanda's side.
yeah, I need to upload monthly full dumps to dropbox and keep them forever.
the monthly dumps are to vtapes, and I thought it would be neat if I could
then just transfer the vtapes to dropbox using amvault.
Strictly speaking, amvault doesn't transfer vtapes, it retapes the dumps
on the vtapes to a new location. While this sounds like a somewhat
pointless distinction, it's actually pretty significant because it means
you can use a different type of tapes for your secondary storage, with
almost every single tapetype option different (which is extremely useful
for multiple reasons). That's actually part of the reason that it's a
preferred alternative to mirroring tapes with the Amanda's RAIT device.
The issue here though is the 'keep it forever' bit. If Amanda is given
an automated tape changer (a library of vtapes is an automated changer),
it assumes it can reuse the tapes as it sees fit. I think there's a
config option that lets you change that, but once you do that, you need
to keep adding tapes (or vtapes) to the library, which can get out of
hand really quickly (especially if you don't plan ahead when deciding on
how things will get labeled).
One option for this though, if you can afford to use something other
than Dropbox, would be to use the Amazon S3 support to store your data
in Amazon Glacier storage (which is insanely cheap at about 0.07 USD per
TB of storage), and enable versioning (so that wen a 'tape' gets
overwritten, the old version gets kept around) and keep old versions
forever. If you're interested in doing this, I can write up
instructions for how to get things set up with Amazon to do this (We
actually do something very similar for off-site backups where I work,
just without Glacier or versioning (but those are easy to set up)).
