By User12611829-Oracle on Jan 07, 2012
ZFS pool and file system functionality may be added with a Solaris release. These new capabilities are identified in the ZFS zpool and file system version numbers. To find out what versions you are running, and what capabilities they provide, use the corresponding upgrade -v commands. Yes, it is a bit disconcerting at first, using an upgrade command, not to upgrade, but to determine which features exist.
Here is an example of each output, for your reference.
# zpool upgrade -v This system is currently running ZFS pool version 31. The following versions are supported: VER DESCRIPTION --- -------------------------------------------------------- 1 Initial ZFS version 2 Ditto blocks (replicated metadata) 3 Hot spares and double parity RAID-Z 4 zpool history 5 Compression using the gzip algorithm 6 bootfs pool property 7 Separate intent log devices 8 Delegated administration 9 refquota and refreservation properties 10 Cache devices 11 Improved scrub performance 12 Snapshot properties 13 snapused property 14 passthrough-x aclinherit 15 user/group space accounting 16 stmf property support 17 Triple-parity RAID-Z 18 Snapshot user holds 19 Log device removal 20 Compression using zle (zero-length encoding) 21 Deduplication 22 Received properties 23 Slim ZIL 24 System attributes 25 Improved scrub stats 26 Improved snapshot deletion performance 27 Improved snapshot creation performance 28 Multiple vdev replacements 29 RAID-Z/mirror hybrid allocator 30 Encryption 31 Improved 'zfs list' performance For more information on a particular version, including supported releases, see the ZFS Administration Guide. # zfs upgrade -v The following filesystem versions are supported: VER DESCRIPTION --- -------------------------------------------------------- 1 Initial ZFS filesystem version 2 Enhanced directory entries 3 Case insensitive and File system unique identifier (FUID) 4 userquota, groupquota properties 5 System attributes For more information on a particular version, including supported releases, see the ZFS Administration Guide.In this particular example, the kernel supports up to zpool version 31 and ZFS version 5.
Where you can run into trouble with this is when you create a pool or file system and then fall back to a boot environment that is older and doesn't support those particular versions. The survival tip is keep your zpool and vfs versions at a level that is compatible with the oldest boot environment that you will ever fall back to. A corollary to this is that you can upgrade your pools and file systems when you have deleted the last boot environment that supports that particular version.
Your first question is probably, "what versions of ZFS go with the particular Solaris releases ?" Here is a table of Solaris releases since 10/08 (u6) and their corresponding zpool and zfs version numbers.
|Solaris Release||ZPOOL Version||ZFS Version|
|Solaris 10 10/08 (u6)||10||3|
|Solaris 10 5/09 (u7)||10||3|
|Solaris 10 10/09 (u8)||15||4|
|Solaris 10 9/10 (u9)||22||4|
|Solaris 10 8/11 (u10)||29||5|
|Solaris 11 11/11 (ga)||33||5|
Note that these versions are for the release as well as if you have patched a system to that same level. In other words, a Solaris 10 10/08 system with the latest recommended patch cluster installed might be at the 8/11 (u10) level. You can always use zpool upgrade -v and zfs upgrade -v to make sure.
Now you are wondering how you create a pool or file system at a version different than the default for your Solaris release. Fortunately, ZFS is flexible enough to allow us to do exactly that. Here is an example.
# zpool create testpool testdisk # zpool get version testpool NAME PROPERTY VALUE SOURCE testpool version 31 default # zfs get version testpool NAME PROPERTY VALUE SOURCE testpool version 5 -This pool and associated top level file system can only be accessed on a Solaris 11 system. Let's destroy it and start again, this time making it possible to access it on a Solaris 10 10/09 system (zpool version 15, zfs version 4). We can use the -o version= and -O version= when the pool is created to accomplish this.
# zpool destroy testpool # zpool create -o version=15 -O version=4 testpool testdisk # zfs create testpool/data # zpool get version testpool NAME PROPERTY VALUE SOURCE testpool version 15 local # zfs get -r version testpool NAME PROPERTY VALUE SOURCE testpool version 4 - testpool/data version 4 -In this example, we created the pool explicitly at version 15, and using -O to pass zfs file system creation options to the top level dataset, we set that to version 4. To make things easier, new file systems created in this pool will be at version 4, inheriting that from the parent, unless overridden by -o version= at the time the file system is created.
The last remaining task is to look at how you might upgrade a pool and file system when you have removed an old boot environment. We will go back to our previous example where we have a version 15 pool and 4 dataset. We have removed the Solaris 10 10/09 boot environment and now the oldest is Solaris 10 8/11 (u10). That supports version 29 pools and version 5 file systems. We will use zpool/zfs upgrade -V to set the specific versions to 29 and 5 respectively.
# zpool upgrade -V 29 testpool This system is currently running ZFS pool version 31. Successfully upgraded 'testpool' from version 15 to version 29 # zpool get version testpool NAME PROPERTY VALUE SOURCE testpool version 29 local # zfs upgrade -V 5 testpool 1 filesystems upgraded # zfs get -r version testpool testpool version 5 - testpool/data version 4 -That didn't go quite as expected, or did it ? The pool was upgraded as expected, as was the top level dataset. But testpool/data is still at version 4. It initially inherited that version from the parent when it was created. When using zfs upgrade, only the datasets listed are upgraded. If we wanted the entire pool of file systems to be upgraded, we should have used -r for recursive.
# zfs upgrade -V 5 -r testpool 1 filesystems upgraded 1 filesystems already at this version # zfs get -r version testpool NAME PROPERTY VALUE SOURCE testpool version 5 - testpool/data version 5 -Now, that's more like it.
For review, the tip is to keep your shared ZFS datasets and pools are the lowest versions supported by the oldest boot environments you plan to use. You can always use upgrade -v to see what versions are available for use, and by using -o version= and -O version, you can create new pools and datasets that are accessible by older boot environments. This last bit can also come in handy if you are moving pools around systems that might be at different versions.
Thanks again to Craig and John for this great tip.