zfs discuss - Jan 2008 - mounting a copy of a zfs pool /file system while orginal is still active

Hello SUN gurus I do not know if this is supported, I have a created a zpool
consisting of the SAN resources and created a zfs file system.  Using third part
software I have taken snapshots of all luns in the zfs pool.  My question is in
a recovery situation is there a way for me to mount the snapshots and import the
pool while the original is still active.  Right now all I am able to do is
export the original than I can  import the snapshots and access the zpool and
filesystem.  Im looking at a no downtime solution can this be done ?
 
 
This message posted from opensolaris.org

On 1/31/08, Hector De Jesus <hec1979 at gmail.com>
wrote:
>
> Hello SUN gurus I do not know if this is supported, I have a created a
> zpool consisting of the SAN resources and created a zfs file system.  Using
> third part software I have taken snapshots of all luns in the zfs pool.  My
> question is in a recovery situation is there a way for me to mount the
> snapshots and import the pool while the original is still active.  Right
now
> all I am able to do is export the original than I can  import the snapshots
> and access the zpool and filesystem.  Im looking at a no downtime solution
> can this be done ?
>
>
> This message posted from opensolaris.org
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
>


Interesting thought.  I guess what you''d have to have is an import flag
that
allows you to "import as".  So you could:

zpool import --import-as yourpool.backup yourpool

I definitely see why you''d want to do it.  I haven''t a clue if
you can :)

--Tim
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20080131/9eb1d86a/attachment.html>

Nope, doesn''t work.

Try presenting one of those lun snapshots to your host, run cfgadm -al, 
then run zpool import.


#zpool import
no pools available to import


It would make my life so much simpler if you could do something like 
this: zpool import --import-as yourpool.backup yourpool







Tim wrote:
> On 1/31/08, *Hector De Jesus* <hec1979 at gmail.com 
> <mailto:hec1979 at gmail.com>> wrote:
>
>     Hello SUN gurus I do not know if this is supported, I have a
>     created a zpool consisting of the SAN resources and created a zfs
>     file system.  Using third part software I have taken snapshots of
>     all luns in the zfs pool.  My question is in a recovery situation
>     is there a way for me to mount the snapshots and import the pool
>     while the original is still active.  Right now all I am able to do
>     is export the original than I can  import the snapshots and access
>     the zpool and filesystem.  Im looking at a no downtime solution
>     can this be done ?
>
>
>     This message posted from opensolaris.org <http://opensolaris.org>
>     _______________________________________________
>     zfs-discuss mailing list
>     zfs-discuss at opensolaris.org <mailto:zfs-discuss at
opensolaris.org>
>     http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
>
>
>
>
> Interesting thought.  I guess what you''d have to have is an import
> flag that allows you to "import as".  So you could:
>
> zpool import --import-as yourpool.backup yourpool
>
> I definitely see why you''d want to do it.  I haven''t a
clue if you can :)
>
> --Tim
> ------------------------------------------------------------------------
>
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
>   
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20080131/5e1eee97/attachment.html>

Dave Lowenstein wrote:
> Nope, doesn''t work.
> 
> Try presenting one of those lun snapshots to your host, run cfgadm -al, 
> then run zpool import.
> 
> 
> #zpool import
> no pools available to import
Does format(1M) see the luns ?  If format(1M) can''t see them it is 
unlikely that ZFS will either.
> It would make my life so much simpler if you could do something like 
> this: zpool import --import-as yourpool.backup yourpool
      zpool import [-o mntopts] [ -o property=value] ... [-d dir |
      -c cachefile] [-D] [-f] [-R root] pool | id [newpool]

          Imports a specific pool. A pool can be identified by its
          name or the numeric identifier. If newpool is specified,
          the pool is imported using the name newpool.  Otherwise,
          it is imported with the same name as its exported name.


# zpool import foopool barpool



-- 
Darren J Moffat

Try it, it doesn''t work.

Format sees both but you can''t import a clone of pool "u001"
if pool
"u001" is already imported, even by giving it a new name.



Darren J Moffat wrote:
> Dave Lowenstein wrote:
>> Nope, doesn''t work.
>>
>> Try presenting one of those lun snapshots to your host, run cfgadm 
>> -al, then run zpool import.
>>
>>
>> #zpool import
>> no pools available to import
>
> Does format(1M) see the luns ?  If format(1M) can''t see them it is
> unlikely that ZFS will either.
>
>> It would make my life so much simpler if you could do something like 
>> this: zpool import --import-as yourpool.backup yourpool
>
>      zpool import [-o mntopts] [ -o property=value] ... [-d dir |
>      -c cachefile] [-D] [-f] [-R root] pool | id [newpool]
>
>          Imports a specific pool. A pool can be identified by its
>          name or the numeric identifier. If newpool is specified,
>          the pool is imported using the name newpool.  Otherwise,
>          it is imported with the same name as its exported name.
>
>
> # zpool import foopool barpool
>
>
>

Darren J Moffat wrote:
> Dave Lowenstein wrote:
>> Nope, doesn''t work.
>>
>> Try presenting one of those lun snapshots to your host, run cfgadm - 
>> al,
>> then run zpool import.
>>
>>
>> #zpool import
>> no pools available to import
>
> Does format(1M) see the luns ?  If format(1M) can''t see them it is
> unlikely that ZFS will either.
>
>> It would make my life so much simpler if you could do something like
>> this: zpool import --import-as yourpool.backup yourpool
>
>      zpool import [-o mntopts] [ -o property=value] ... [-d dir |
>      -c cachefile] [-D] [-f] [-R root] pool | id [newpool]
>
>          Imports a specific pool. A pool can be identified by its
>          name or the numeric identifier. If newpool is specified,
>          the pool is imported using the name newpool.  Otherwise,
>          it is imported with the same name as its exported name.
Given that the pool is snapshot of one or more vdevs in an existing  
ZFS storage pool, not only is the "name" identical, so it is the  
"numeric identifier". If can be determined that when using "zpool
import ....", duplicates are suppressed, even if those duplicates are  
entirely separate vdevs containing block-based snapshots, physical  
copies, remote mirrors or iSCSI Targets.

The steps to reproduce this behavior on a single node, using files and  
stand Solaris utilities is as follows:

# mkfile 500m /var/tmp/pool_file
# zpool create pool /var/tmp/pool_file
# zpool status pool
   pool: pool
  state: ONLINE
  scrub: none requested
config:

         NAME                  STATE    READ WRITE CKSUM
         pool                  ONLINE      0     0     0
           /var/tmp/pool_file  ONLINE      0     0     0

errors: No known data errors

# zpool export pool
# dd if=/var/tmp/pool_file of=/var/tmp/pool_snapshot
   { wait, wait, wait, ... more on this later ...}
1024000+0 records in
1024000+0 records out
# zpool import -d /var/tmp
   pool: pool
     id: 14424098069460077054
  state: ONLINE
action: The pool can be imported using its name or numeric identified
config:

         pool                  ONLINE
           /var/tmp/pool_file  ONLINE

Question: What happened to the other ZFS storage pool call  
pool_snapshot?

Answer: Its presence is suppressed by zpool import. If one was to  
rename /var/tmp/pool_file to some other directory, the /var/tmp/ 
pool_snapshot will now appear.

# mv /var/tmp/pool_file /var/pool_file
# zpool import -d /var/tmp
   pool: pool
     id: 14424098069460077054
  state: ONLINE
action: The pool can be imported using its name or numeric identified
config:

         pool                      ONLINE
           /var/tmp/pool_snapshot  ONLINE      0     0     0

At this point, if one was to go ahead with the import of pool, (which  
would work) then rename /var/pool_file back to /var/tmp/pool_file, its  
presence would now be suppressed. Conversely, if the rename was done  
first, then a zpool import was attempted, again only one storage pool  
would exists at any given time.

Clearly there is some explicit suppressing of duplicate storage pools  
going on here. Browsing the ZFS code looking for answer, the logic  
surrounding zfs_inuse(), seem to cause this behavior, expected or not.

	http://cvs.opensolaris.org/source/search?q=vdev_inuse&project=%2Fonnv

========
As mentioned earlier, the  {wait, wait, wait, ...} can be eliminated  
by using Availability Suite Point-in-Time Copy, by itself, or in  
combination with Availability Suite Remote Copy or iSCSI Target, all  
of which are present in OpenSolaris today, and all are much fast then  
the dd utility.

As one that supports both Availability Suite and iSCSI Target, not  
suppressing duplicate pool names and pool identifiers, in combination  
with a rename on import, "zpool import -new <name> ...", would
provide
a means to support various copies, or nearly identical copies of a ZFS  
storage pool on the same Solaris host.

While browsing the ZFS source code, I noticed that "usr/src/cmd/ztest/ 
ztest.c", includes ztest_spa_rename(), a ZFS test which renames a ZFS  
storage pool to a different name, tests the pool under its new name,  
and then renames it back. I wonder why this functionality was not  
exposed as part of zpool support?

- Jim
>
> # zpool import foopool barpool
>
>
>
> -- 
> Darren J Moffat
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Jim Dunham
Storage Platform Software Group
Sun Microsystems, Inc.
work: 781.442.4042
cell:   603-724-3972

http://blogs.sun.com/avs
http://www.opensolaris.org/os/project/avs/
http://www.opensolaris.org/os/project/iscsitgt/
http://www.opensolaris.org/os/community/storage/

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20080204/fe425b7a/attachment.html>

>
> While browsing the ZFS source code, I noticed that "usr/src/cmd/ 
> ztest/ztest.c", includes ztest_spa_rename(), a ZFS test which  
> renames a ZFS storage pool to a different name, tests the pool  
> under its new name, and then renames it back. I wonder why this  
> functionality was not exposed as part of zpool support?
>
See 6280547 want to rename pools.

Just hasn''t been hight on the priority list.

eric

Hello Dave,

Monday, February 4, 2008, 7:15:54 PM, you wrote:

DL> Try it, it doesn''t work.

DL> Format sees both but you can''t import a clone of pool
"u001" if pool
DL> "u001" is already imported, even by giving it a new name.


I guess because you got exact copy you also got the same zfs labels
which is confusing zfs.

-- 
Best regards,
 Robert                            mailto:milek at task.gda.pl
                                       http://milek.blogspot.com