At this point, the only thing would be to use 11.1 to create a new pool at 
151's version (-o version=) and top level dataset (-O version=).   Recreate the 
file system hierarchy and do something like an rsync.  I don't think there is 
anything more elegant, I'm afraid.  

That's what I did yesterday :)


Sent from my iPhone

On Dec 13, 2012, at 12:54 PM, Jan Owoc <> wrote:

> On Thu, Dec 13, 2012 at 11:44 AM, Bob Netherton <> 
> wrote:
>> On Dec 13, 2012, at 10:47 AM, Jan Owoc <> wrote:
>>> Yes, that is correct. The last version of Solaris with source code
>>> used zpool version 28. This is the last version that is readable by
>>> non-Solaris operating systems FreeBSD, GNU/Linux, but also
>>> OpenIndiana. The filesystem, "zfs", is technically at the same
>>> version, but you can't access it if you can't access the pool :-).
>> That is a touch misleading.  This has always been the case since S10u2.  You 
>> have to create the pool AND the file systems at the oldest versions you want 
>> to support.
>> I maintain a table of pool and version numbers on my blog ( 
>> com/bobn) for this very purpose.   I got lazy the other day and made this 
>> mistake between 11ga and 11.1.
>> Watch the ZFS send approach because you might be sending a newer file system 
>> version than is supported.  Yes, I've done that too :)
> Bob, you are correct. There is now a new version of "zfs" in Solaris
> 11.1. I assume it's incompatible with the previous version:
> Any suggestions how to help OP read his data on anything but Solaris
> 11.1 or migrate it "back" a version?
> Jan
zfs-discuss mailing list

Reply via email to