ZFS for Linux 0.6.9

For additional information about this project, please visit the project page.

Available downloads

Ubuntu PPA

For Ubuntu Lucid

zfs-fuse-0.6.9.tar.bz2

For all platforms (1.3 MB)

Maintenance Snapshot

For all platforms (1.3 MiB)

Ubuntu PPA

For Ubuntu Maverick

Release Notes

The testing branch has graduated into our brand-new release, which promises to be a lot more stable than the 0.6.n family has been to date.

How do I start

There are fresh packages for Gentoo and Ubuntu. RHEL (in the Epel-5 repo) and Fedora 13 have recent beta versions

If you cannot find a suitable package, you can compile your own version, see the homepage or the FAQ and specifically 

Q. What Branch?

A. Maintenance Snapshot

What is in it?

Relative to 0.6.0 testing has (among others): 

• pool version 23
• this includes full Dedup support. The major upstream annoyances have been patched by SUN; Still beware of the Dedup FAQ
• non blocking zfs send/recv (enable local piping e.g.)
• really this is due to an all-new multithreaded dispatch of ioctls
• zpool split
• to boldly split your infinitives and ... mirrors ! No more detach and force import, just 'zpool split pool newpool'
• sharenfs support (with nfs-kernel-server) 
• see links below for details on configuration
• fix for 'dataset busy' on export
• kstat interface
• libumem merged from libumem upstream source
• PowerPC support! (tested on Debian PPC)
• Autoreplace (Hot Spares!) support (replaced Solaris'es SMF features with a configurable script)
• you can configure a script in /etc/zfs/zfs_pool_alert to handle vdev failures. If your system has perl, you can use the contributed script
• Configurable stack size limits (by popular demand)
• Dynamically raise kernel mmaps limit for daemon proces
• Experimental waf build support (SCons is still the main mechanism) 
• Many many upstream features and fixes merged from onnv-gate 
• Primitive autoexpand support
• It doesn't really use the autoexpand property like in Solaris. "zpool online" can be used to trigger autoexpand. According to Emmanuel a simple mount should even do the trick. Beware of resizing physical volumes/partitions that back your pool's vdevs though, in case you might need to reattach devices of smaller size (in case of mirrors/raidz this won't work if the existing vdevs are larger).
• A bash-completion script in contrib (deploy in /etc/bash_completion.d/)

For more information on the sharenfs handling, see NFS auto-share (sharenfs) support landed in unstable and the initial group thread