Re: [zfs-discuss] backwards/forward compatibility
On Thu, 29 Apr 2010, Ian Collins wrote: You can create pools and filesystems with older versions if you want them to be backwards compatible. I have done this when I was sending data to a backup server running an older Solaris version. From the zpool manual page, it seems that it should be possible to create an older pool version using newer software. Of course, it is also necessary to make sure that any created filesystems are a sufficiently low version. Bob -- Bob Friesenhahn bfrie...@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/ GraphicsMagick Maintainer,http://www.GraphicsMagick.org/ ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] backwards/forward compatibility
On 04/29/10 10:21 AM, devsk wrote: I had a pool which I created using zfs-fuse, which is using March code base (exact version, I don't know; if someone can tell me the command to find the zpool format version, I would be grateful). Try [zfs|zpool] upgrade. These commands will tell you the current supported version and the version of each pool/filesystem. If the command makes you nervous, run it as a normal user! I exported it and now I tried to import it in OpenSolaris, which is running Feb bits because it says 134 Feb 2010 in uname output (2009-06 updated with pkg image-update). It fails to import it because of incompatible version. b134 will be using pool version >= 22. Backward compatibility is a given and I know Solaris will never compromise on that and Sun has been industry leader in that. What about forward compatibility? I would think zfs pool format would be stable by now. No, newer pool support newer features. Imagine the arse ache you would experience importing a dedup poll to a system then didn't support it! So, what happened here? I know I am going to get shouted at for saying this but this tells me that ZFS is still not on-disk stable in the same vein as BTRFS is not on-disk stable. But it gets better. I can mount and use BTRFS FSs which were created after 2.6.32 (which was a long ago) in a 2.6.32 kernel i.e. I can boot my older kernel if I need to and still access the newer FS and files in it, without breaking anything. Of course, the newer kernels work with FSs created in older kernels. You can create pools and filesystems with older versions if you want them to be backwards compatible. I have done this when I was sending data to a backup server running an older Solaris version. May be I am doing something wrong. May be it is just about using '-f' flag and things will work out and nothing will break. Is it? I look fwd to guidance from the community on this. Post back the output of the upgrade commands an the errors you get when importing the pool. -- Ian. ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] backwards/forward compatibility
I had a pool which I created using zfs-fuse, which is using March code base (exact version, I don't know; if someone can tell me the command to find the zpool format version, I would be grateful). I exported it and now I tried to import it in OpenSolaris, which is running Feb bits because it says 134 Feb 2010 in uname output (2009-06 updated with pkg image-update). It fails to import it because of incompatible version. Backward compatibility is a given and I know Solaris will never compromise on that and Sun has been industry leader in that. What about forward compatibility? I would think zfs pool format would be stable by now. So, what happened here? I know I am going to get shouted at for saying this but this tells me that ZFS is still not on-disk stable in the same vein as BTRFS is not on-disk stable. But it gets better. I can mount and use BTRFS FSs which were created after 2.6.32 (which was a long ago) in a 2.6.32 kernel i.e. I can boot my older kernel if I need to and still access the newer FS and files in it, without breaking anything. Of course, the newer kernels work with FSs created in older kernels. May be I am doing something wrong. May be it is just about using '-f' flag and things will work out and nothing will break. Is it? I look fwd to guidance from the community on this. -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss