Long ago, the potential for using ZFS in Unraid was debated.  Whatever became of that idea?

 

 

personally I think BTRFS have better potential.

My post isn't an endorsement for ZFS per-se, although I do support migration to whatever file system best protects data integrity.  I was just curious if the idea [of migration] had been scuttled.

not sure, but I did hear that a BTRFS implementation was also planed. maybe ver. 6.0 ?!

but if that happened the hardware requirements would need to be updated

and system would have to be bump to  64 bit.

 

ZFS will not allow for mixed size HDD  config. as current system does.

 

 

ZFS will not allow for mixed size HDD  config.

 

Well, that's bad news.  All my drives are 2Tb, but I'd hate to think I had to simultaneously swap them ALL out to migrate to larger disks.

 

ZFS will not allow for mixed size HDD  config.

 

Well, that's bad news.  All my drives are 2Tb, but I'd hate to think I had to simultaneously swap them ALL out to migrate to larger disks.

 

Plus all drives in a ZFS array need to spin up no matter how small the read is... I'm really not a fan of that...

I think Tom was looking at btrfs as a solution for the cache drive.

Well you would not need unraid with zfs, as most of the functiobality that unraid provides is built in to the zfs it self. But it have some drawbacks as well. No mixed size drives. No drive spin management, etc. Very high system specs, aspecially memory.

 

Sent from my SGH-T889 using Tapatalk

 

 

ZFS is an excellent solution to a different problem that unRAID solves.

 

You could argue that ZFS around parity drives is a good idea, and i probably is, but there are plans for better unraid parity anyway.

ZFS will not allow for mixed size HDD  config. as current system does.

 

Truth is that ZFS, unlike currently used reiserfs, is capable of handling a raid config itself.

*But* you can perfectly create a pool out of a single disk.

 

In unRAID, the raid feature is implemented by the (modified) mdadm driver.

This is blocklevel and is lying underneath the filesystem.

Why shouldn't it be possible to use ZFS on each disk individually? You'll "just" end up with a pool for every drive.

Which would require some modifications to emhttp, I think.

 

BTW: you can run snapraid (which is offline parity) for a list of individual ZFS disks/pools, see: http://forums.servethehome.com/solaris-nexenta-openindiana-napp/1299-zfs-raid-snapraid-media-server-napp-box.html

The biggest advantage of ZFS to my mind is the self-healing properties, protecting against bit rot etc. That is ideal for a hard disk archive. But the downsides have already been mentioned: no mixed drives, multiple drives per pool means all drives spin up together, etc. The merits of ReiserFS have been discussed in ancient threads on here, but the primary reason it's still in use is its robust journaling. BTRFS is most likely the path forward, though there has been no mention of it beyond the cache drives.

The thing with the zfs biggest advantage is to use it you need a very robust hardware setup, especially RAM. The feature is very memory hungry. Btrfs on the oher hand have self healing functions too it is part of the CoW nature of the fs and it is more flexible and less resource hungry.  Win win IMHO.

Not that I am promoting one over the other, but I do prefere one over the other. I just wish Linux where easier to use as I still can not find a suitable distro. For my self.

Sent from my SGH-T889 using Tapatalk