zfs q regarding backup strategy

[Dr. Nikolaus Klepp]

Anno domini 2021 Sun, 3 Oct 19:45:22 +0100
 tech-lists scripsit:
> On Sun, Oct 03, 2021 at 02:46:47AM -0700, David Christensen wrote:
> >On 10/2/21 16:21, tech-lists wrote:
> >
> >> OK. if I think I understand correctly then if we need to keep everything
> >> then I'll need to keep every incremental snapshot.
> >
> >
> >ZFS is novel and complex.  It requires different thinking than
> >traditional partitions/ slices, volumes, filesystems, etc..  
> 
> Yep I hear you. Someone else mentioned bacula which keeps track of files
> in a database. Maybe thats the path I'll need to go down.
> 
> I'm looking for a recent (say from 2020 on) book about ZFS and OpenZFS
> on FreeBSD with some worked examples. The Mastery books which I think
> you mentioned earlier, the latest I can find is from 2016 and a lot has
> happened to zfs in five years. It's a problem with all technology books
> where the technology changes rapidly.
> 
> I'm familiar with zfs for my own use. It's saved data from broken disks
> several times. Just unfamiliar in this particilar use case:
> 
> [1] offsite backup over ssh or tls
> 
> [2] *all* changes to live site must be roll-backable [?] from backup,
> for all time. If a file disappears from the live site I should be able
> to restore it from backup.
> 
> [3] backup must be searchable. Doesn't have to be searchable all the
> time, it just has to be if something needs to be restored.
> 
> Basically if the system gets set up, runs for a couple of years say.
> Then I get asked to restore the site into the state it was a year and a
> half ago. Then three months on from that restore point, restore it somewhere 
> else like another test system. I need to be able to do that reliably.
> 
> But yeah, in the abscence of a book I really need to do testing.

That's waht zfs allows you to do - well, if your disk is big enough to hold all snapshots :)

Nik

-- 
Please do not email me anything that you are not comfortable also sharing with the NSA, CIA ...