Prev: String of nulls in /var/adm/mesages
Next: ldom host cannot see ldom primary
From: Ian Collins on 4 Jan 2010 21:44
David Combs wrote:

> If the other posts (ie warnings) in this thread are
> "zfs: that's the way it works -- like it or not!", about
> NOT being able to via a SINGLE top-level "level-0" zfs
> snapshot (to later save to tape), and get *everything*,
> INCLUDING NESTED ZFS'S AND *THEIR* SUBTREES, etc,
>
> that far and away the SAFEST and MOST FOOLPROOF way to
> go is to have only ONE zfs in the pool --

What would you do with such a snapshot? You keep mentioning "save to
tape", but how would you do that with a snapshot?

> so I *can* via a *single* snapshot, get it ALL.

That's what recursive snapshots are for. That is the way zfs is designed.

> (I didn't even suspect that nested zfs's caused the multitude
> of complications of snappshotting, as detailed in the very
> surprising but excellently informative posts earlier on
> this same thread, until I saw those very posts.
>
> Simply NO SUCH DOCUMENTATION (and implicit warnings) anywhere
> else I've seen. (Why not? Sure seems to be **ESSENTIAL** info.)

It is all clearly documented in the zfs man page.

--
Ian Collins
From: Gary R. Schmidt on 5 Jan 2010 09:23
David Combs wrote:
[SNIP]
'Tis not difficult, assuming the the top-level file system is called
"filesystem":

# zfs snapshot -r filesystem(a)YYYYMMDD
# zfs send -R filesystem(a)YYYYMMDD > /dev/rmt0
# zfs destroy -r filesystem(a)YYYYMMDD

Approximately as difficult as falling off a log - and all straight from
the zfs man page, (except I send the stream to a file on a USB disk).

Cheers,
Gary B-)
 | 
Pages: 1
Prev: String of nulls in /var/adm/mesages
Next: ldom host cannot see ldom primary