Why SUSE Likes The Btrfs File-System
Phoronix: Why SUSE Likes The Btrfs File-System
While many assumed Fedora would be the first tier-one Linux distribution shipping with Btrfs by default, it looks like openSUSE may end up being the one. OpenSUSE has been looking at switching to Btrfs for their next release (post-13.1) and already in its current state feel Btrfs is safe for users -- nearly one year after SUSE Enterprise felt Btrfs is production-ready...
Would be really nice to see these snapshotting capabilities of Btrfs come to Debian (apt, apt-get, synaptic, aptitude) too...
We've been running btrfs on multiple systems (desktops and server) in multiple configurations (single, raid0, raid1). After 15 months, we haven't encountered any issues.
I think where people get into trouble with btrfs is that if their system experiences a problem (e.g., power outage), their first thought is to run btrfsck. However, btrfsck is the last thing you should try. The first thing is to just let btrfs resolve any issues on its own; much of the fsck functionality occurs at mount. If that doesn't work, mount with the '-o recovery' option. That lets btrfs go back in time, trying previous filesystem tree roots until it finds one with integrity. You may lose the last few seconds of changes, but your data will be intact.
In fact, btrfs saved me when an external drive was silently corrupting data (hardware fail). Since btrfs checksums data blocks, it reported the issue. SMART indicated no problem with the drive, but in fact it was writing trash. I previously had EXT4 on the drive, and it was silent. The backups I was writing to it were worthless, but until I converted to btrfs I didn't know there was an issue.
15 months eh? You can't be using it for anything serious then. I tried it a year ago, and it was still ridiculous slow on certain operations, some of them triggered on common stuff like apt-get. The issue only got resolved recently, until then btrfs was effectively USELESS on any Debian based linux distributions.
Originally Posted by macemoneta
I stopped using it after I got a corrupted sector and btrfs correctly reported it. Their fsck could also report it. BUT NOTHING COULD FIX IT!!!! I ended up losing more than half of the data on the drive because it was corrupted and there exists no tools to fix errors on btrfs.
If you're using back-level software that a distribution considers 'stable', you can't expect to have the fixes that the developers have already applied long ago. On Fedora, I run the Rawhide no-debug kernel (currently 3.12.0-0.rc5.git3.2, per: http://fedoraproject.org/wiki/RawhideKernelNodebug), and the latest btrfs-progs from Koji (http://koji.fedoraproject.org/koji/p...packageID=6398). As I said, no problems whatsoever.
Originally Posted by carewolf
Fedora and Arch its been working fine for more than 2years here. Don't complain that you had older versions of the tools, kernel, and overall filesystem at your disposal. If you're gonna be testing 'next-gen' stuff you want to be using RECENT software which probably means running a distro that tracks RECENT releases like Fedora or Arch or Gentoo or Debian Unstable. Anything else and all your complaining and problems comes down to "Hey I was using an old version of the software and found bugs!!! Didn't test to see if the new releases fixed them... YOUR GUYS' PAST SUCKS!"
Originally Posted by carewolf
While I am glad that Btrfs is finally going to get some end-user testing, I will continue to be sitting pretty on ext4 for now.
I guess there's no surprise that the ZFS advocates are banding together under OpenZFS and also try to push ZFS onto Linux through efforts like ZFSOnLinux.
BTRFS is arriving now and it has the huge advantage of being shipped and maintained directly in the Linux kernel which means it will likely quickly become the de facto default file system on Linux once it has been tested in the wild (like here with SUSE), which in turn means there will be a lot more developer resources allocated towards BTRFS as all the companies which rely on Linux will want to get the best performance/features out of it.
Meanwhile the open source ZFS implementation which due to licence means can't be included in the kernel has been cut off from 'upstream' enhancements from Oracle as they no longer open source them, leaving the open ZFS effort without any real commercial backing from what I can see.
This likely means that it will struggle against BTRFS feature-wise in the long run, and since BTRFS is shipped with the kernel I can't see ZFS getting any real foothold on Linux once BTRFS is shipped as default with most distros.
So they need to act now and try to make it as painless as possible to use ZFS on Linux, which I think is what they are trying to do with ZFSOnLinux.
I don't think it will work though. The advantage of being able to be shipped in the kernel is just too big, so I believe BTRFS will become the standard on Linux and ZFS will remain largely confined to the BSD's with little to no Linux presence to speak of.
Feel free to do that. I made an ebuild of Snapper for Gentoo, for instance.
Originally Posted by uid313
So does BTRFS has some kind of fsck, yet?