On 2018-01-30 17:18, ghe wrote:
On 01/30/2018 12:29 PM, [email protected] wrote:
I feel like I've asked this before, but I can't find any emails.
I can't believe this isn't an FAQ. Or rather, there is an FAQ, but the
answer is (a) very sparse and (b) doesn't really answer the question.
I had a machine. That machine was getting regular backups. The machine
died. I have replaced it with a new machine. So having had this
emergency, I now want to keep, in perpetuity, my last full backup of
the now-dead machine.
How big was the dead disk? Do you have space to store the whole thing?
Did amanda do a level 0 of the whole dead disk to 17? If not, there are
very likely pieces of that disk on several of your virtual tapes.
amrestore deals with all that.
The backup in question is on (virtual) tape number 17. So let's say
I take the approparite files that are in my /storage/amanda/vtapes/slot17
directory and copy them somewhere safe. Six months go by, my real
slot17 gets reused, and I take those old files and copy them into slot44.
What is my next step? How do I get those backups back into my amanda
index so that I can amrecover from them? Is that what amreindex does?
Is that what amrestore does?
What I'd do is recover the last files amanda backed up from that disk,
using amrestore. I'd restore to a disk, consider that the perpetual
backup, and not try to get that old disk data anywhere in amanmda's
database -- amanda is very much oriented to reusing things in a cycle,
and trying to get her to change her ways can be difficult.
amrestore's a pleasant piece of software to use. You just tell it the
date you want to restore, the disk, the files, and some other things (I
use it infrequently, and I have to read the man page every time).
amrestore figures out which tapes you need, and restores the data.
Then you can do what you want with them -- burn to optical, buy a new
disk, whatever.
I would suggest the same approach myself. In fact, that's pretty much
what we do where I work. Whenever we permanently decomission a system,
it gets pulled from the backup rotation, and we image the disk and store
the disk image in archival storage that's separate from the storage we
use for regular backups. Our procedure is similar for a failed disk we
don't plan to replace, except instead of imaging it as-is, we rebuild it
from backups and then image it (the imaging procedure was the norm
before we switched to amanda, so it's just kind of stuck around).
