A seperate ZFS question

[Trond Endrestøl]

On Fri, 28 Apr 2017 10:28-0400, Ultima wrote:

> -n is your friend. Never used zpool add without it to verify changes are as
> you intend. Not using it first is similar to jumping off a plane without
> verifying your parachute is on properly.

Not to mention using -v.

Also, the OP can use:

   zfs snap -r zrootnew at transfer

   zfs send -RLev zrootnew at transfer | {ssh,nc,something-something}
or
   zfs send -RLev zrootnew at transfer > /some/file/somewhere

for creating a replicated stream preserving both data and metadata.

Use these commands for restoring the missing data:

   {ssh,nc,something-something} | zfs recv -Fduv newpool
or
   zfs recv -Fduv newpool < /some/file/somewhere

If all drives are connected to the same system, the replication can be 
simplified to:

zfs send -RLev zrootnew at transfer | zfs recv -Fduv newpool

Remember to remove the recursive @transfer snapshot afterwards.

To preview what will happen:

zfs destroy -nRv newpool at transfer

To go ahead and make the plunge:

zfs destroy -Rv newpool at transfer


Why are we top posting all of a sudden?

> On Fri, Apr 28, 2017 at 9:52 AM, Trond Endrestøl <
> Trond.Endrestol at fagskolen.gjovik.no> wrote:
> 
> > On Fri, 28 Apr 2017 14:14+0100, Paul Macdonald wrote:
> >
> > >
> > > I fat fingered adding a cache drive and was unable to remove the SSD
> > cache
> > > drive (ada2)
> > >
> > > I added a 2nd SSD (ada3) for safety as i didn;t want to risk the pool at
> > all
> > > and exporting wasn;t feasible in timescale.
> > >
> > > Can someone explain :
> > >
> > >             what i've actually managed to set up here (!),
> > >
> > >             the ramifications of forceably offlining/removing ada2
> > >
> > >             and if i should worry when ada2 gets full?
> > >
> > >                  capacity     operations    bandwidth
> > > pool          alloc   free   read  write   read  write
> > > ------------  -----  -----  -----  -----  -----  -----
> > > zrootnew       692G   460G     15    174   351K  2.40M
> > >   mirror       683G   237G     14    138   328K  1.43M
> > >     gpt/zfs0      -      -      6     50   155K  1.44M
> > >     ada1p3        -      -      7     50   176K  1.44M
> > >   ada2        9.17G   223G      0     36  22.9K   989K
> > > cache             -      -      -      -      -      -
> > >   ada3        39.0G   194G     23      5   168K   264K
> >
> > First, you have gpt/zfs0 and ada1p3 in a mirrored configuration. Next,
> > you have ada2 striped (RAID 0) to the first pair of mirrors. This is a
> > very weird "RAID 1 + 0" combo. Guess what will happen when you lose
> > ada2!
> >
> > Maybe you should describe what you wanted to happen in the first
> > place.
> >
> > If you really wanted a three-way mirror, you have just learned the
> > hard way about the difference between zpool add and zpool attach.
> >
> > The latter is for adding additional members to existing mirrors, while
> > the former is for adding raidz{1,2,3} groups to a pool, and the add
> > command got you into this mess.
> >
> > "zpool attach zrootnew ada1p3 ada2" is the right command if you wanted
> > to create a three-way mirror in the first place.
> >
> > I suggest you make good backups of this pool, destroy it, recreate it
> > properly, and restore the missing data.

-- 
+-------------------------------+------------------------------------+
| Vennlig hilsen,               | Best regards,                      |
| Trond Endrestøl,              | Trond Endrestøl,                   |
| IT-ansvarlig,                 | System administrator,              |
| Fagskolen Innlandet,          | Gjøvik Technical College, Norway,  |
| tlf. mob.   952 62 567,       | Cellular...: +47 952 62 567,       |
| sentralbord 61 14 54 00.      | Switchboard: +47 61 14 54 00.      |
+-------------------------------+------------------------------------+