From: David Combs on 26 Dec 2009 19:48 Suppose you're using zfs, and down under you create another one, and stuff (files, dirs, etc) beneath it. Heck, maybe even yet another zfs down within THAT one, it also having stuff under it. So, time to make a snapshot. (1) Suppose I snapshot (just) the topmost one. (I know what I'd *like* to happen -- but what really does happen? That is, does EVERYTHING get looked at and properly processed?) (2) For the purpose of writing the COMPLETE snapshot (well, to coin a term, let's call lit a "level 0" snapshot) that I can write/save onto a (long) tape (and take off-sitei). I can do that "level 0" one also via doing solely the TOP one? Thanks, David
From: Zfs.. on 26 Dec 2009 20:20 On Dec 27, 12:48 am, dkco...(a)panix.com (David Combs) wrote: > Suppose you're using zfs, and down under you create > another one, and stuff (files, dirs, etc) beneath it. > > Heck, maybe even yet another zfs down within THAT one, > it also having stuff under it. > > So, time to make a snapshot. > > (1) Suppose I snapshot (just) the topmost one. > > (I know what I'd *like* to happen -- but what > really does happen? > > That is, does EVERYTHING get looked at and properly processed?) No the snapshot is per dataset so if you snapshot the top most vdev then you get a snapshot of whats in that dataset not anything underneath. Each dataset is independent of the pool, although all the resources of the pool are available to all the datasets all the time, this is controlled via dataset quotas/reservations etc. When you create a pool. you can mount this top level vdev and then all you would need is one snapshot, however the beauty of zfs is that a lot of the time you need to have control over a particular dataset so the creation of mutiple hirearchial datasets is very appealing. > > (2) For the purpose of writing the COMPLETE snapshot (well, to > coin a term, let's call lit a "level 0" snapshot) that I > can write/save onto a (long) tape (and take off-sitei). > > I can do that "level 0" one also via doing solely the TOP one? No you need to snapshot each dataset and use send/recv to write it off host.
From: Mark Musante on 26 Dec 2009 22:56 You can do a recursive snapshot with "zfs snapshot -r <dataset>" and that will snapshot <dataset> and every dataset underneath it. --- frmsrcurl: http://compgroups.net/comp.unix.solaris/ZFS-A-nested-zfs-down-within-a-higher-up-zfs-SNAPSHOT-the-top-one-bot-one-done
From: Zfs.. on 27 Dec 2009 09:48 On Dec 27, 3:56 am, Mark Musante <u...(a)compgroups.net/> wrote: > You can do a recursive snapshot with "zfs snapshot -r <dataset>" and that will snapshot <dataset> and every dataset underneath it. > > --- > frmsrcurl:http://compgroups.net/comp.unix.solaris/ZFS-A-nested-zfs-down-within-... Will this give you just one snapshot though or multiple snapshots ? I dont have access to a Solaris box at the minute to try this out !
From: Mark Musante on 27 Dec 2009 10:33 It will be one snapshot per dataset, but you can use send/recv recursively as well, so you can treat them as a single snapshot, and generate a single stream to write to tape. --- frmsrcurl: http://compgroups.net/comp.unix.solaris/ZFS-A-nested-zfs-down-within-a-higher-up-zfs-SNAPSHOT-the-top-one-bot-one-done
|
Next
|
Last
Pages: 1 2 Prev: USB Drive: Can't open /dev/rdsk/c2t0d0s2: I/O error Next: problem booting from dvd |