new jail utility is available. announcement.
Aiza
aiza21 at comclark.com
Wed Jul 21 11:37:45 UTC 2010
Valentin Bud wrote:
> On Wed, Jul 21, 2010 at 12:52 PM, Aiza <aiza21 at comclark.com> wrote:
>
>> Not yet, when I have a spare box I might, although I quite like using
>>> zfs for jails as you can limit the disk usage dynamically per zfs
>>> filesystem and I didnt see any support there yet, even basic support
>>> like there is with ezjail would be nice.
>>>
>>>
>> Zfs was left out because its over kill. Sparse image jails gives the same
>> protection at a 10th of the overhead.
>>
>>
> Hello community,
>
> ZFS shouldn't be left out. Besides limiting the disk usage dynamically per
> zfs FS
> you have another big advantage - snapshots. Suppose you want to upgrade
> ports
> is a jail and something goes kaboom you just revert to the previous working
> snapshot.
> I agree you can copy the image back and forth but zfs snapshots are faster
> and not
> that space consuming.
>
> The layout that I plan to use is the following:
>
> storage/jails
> |>storage/jails/group1
> | |
> |
> |>storage/jails/group1/jail1
> |
> |>storage/jails/group1/jail2
> |
> |>storage/jails/group2
> | |> ...
> |
>
> Group can be any kind of characteristic you want to take into account
> regarding
> those jails (eg. group1 - mail servers, group2 - web servers, groupX -
> companyY, etc.).
> You can also go with more levels of depth but for me it's enough.
>
> This way if your server doesn't handle all the jails you have running,
> simply
> buy new hardware, install FBSD (or just copy the ZFS root container over to
> the new
> system) and migrate the jails over.
>
> I am waiting for network stack virtualization to come out and dreaming about
> live jails
> migration in the future of FBSD :).
>
> I would like you to reconsider ZFS support and thanks for qjail :).
>
> a great day,
> v
What you are doing behind the jail system back using zfs, qjail does
with the -z zone option right up front. And the archive and restore of
qjail jails is less than 3 seconds right now. How much faster does it
need to be?
More information about the freebsd-questions
mailing list