Bcachefs Merged Into the Linux 6.7 Kernel (phoronix.com) 39
The new open-source, copy-on-write file system known as Bcachefs has been successfully merged into the Linux 6.7 kernel. "Given the past struggles to get Bcachefs mainlined, I certainly didn't expect to see Linus Torvalds act so soon on merging it," writes Phoronix's Michael Larabel. "But after it spent all of the 6.6 cycle within Linux-Next, overnight Linus Torvalds did in fact land this new file-system developed by Kent Overstreet."
From a Slashdot story published on Friday August 21, 2015: Bcachefs is a new open-source file-system derived from the bcache Linux kernel block layer cache. Bcachefs was announced by Kent Overstreet, the lead Bcache author. Bcachefs hopes to provide performance like XFS/EXT4 while having features similar to Btrfs and ZFS. The bachefs on-disk format hasn't yet been finalized and the code isn't yet ready for the Linux kernel. That said, initial performance results are okay and "It probably won't eat your data -- but no promises." Features so far for Bcachefs are support for multiple devices, built-in caching/tiering, CRC32C checksumming, and Zlib transparent compression. Support for snapshots is to be worked on.
From a Slashdot story published on Friday August 21, 2015: Bcachefs is a new open-source file-system derived from the bcache Linux kernel block layer cache. Bcachefs was announced by Kent Overstreet, the lead Bcache author. Bcachefs hopes to provide performance like XFS/EXT4 while having features similar to Btrfs and ZFS. The bachefs on-disk format hasn't yet been finalized and the code isn't yet ready for the Linux kernel. That said, initial performance results are okay and "It probably won't eat your data -- but no promises." Features so far for Bcachefs are support for multiple devices, built-in caching/tiering, CRC32C checksumming, and Zlib transparent compression. Support for snapshots is to be worked on.
well alright then (Score:2)
Re: well alright then (Score:3)
I hate to just be crass and dismissive, but it doesn't even have snapshots. I will never again use a filesystem that doesn't support snapshots for anything other than a toy. I've used ZFS since it was merged to FreeBSD and still have my first raidz2 pool in operation with 100% uptime and no data loss.. Why would I ever switch??
Re: (Score:1, Flamebait)
Because a smelly hippie does not approve of the license.
Re: (Score:2)
Its not the smelly hippie you need to worry about. Its the lawyers.
Small details may seem trifling and pedantic, but that indiference won't help you when a judge declares your computer unlawful because some indignant patent troll decided to make your life hell
Licenses *matter*.
Re: well alright then (Score:4, Informative)
Have you read the CDDL licence? It comes with an explicit patent grant. It's freely embeddable in other projects of any other licence you care to use, from BSD to proprietary. The legal risk posed by ZFS is negligible, because the CDDL explicitly granted you all of the rights to use and distribute; Oracle, or any other rights holders, aren't going to be able to sue when their own licence terms state what they allowed you to do in black and white.
The "incompatibility" with the GPL is solely due to the way the GPL is written. The CDDL is absolutely fine being embedded in projects under other licences, and OpenZFS is used in plenty of other systems without any problems whatsoever.
Re: well alright then (Score:4, Informative)
bcachefs provides btrfs style writeable snapshots, at subvolume granularity. [bcachefs.org]
Bcachefs was designed to support snapshots. There are some missing pieces yet and not every aspect of the out-of-tree file system is included in the kernel merge.
Re: (Score:3)
bcachefs doesn't have any scrub capabilities either. This is important for data assurance. Hopefully this gets added soon.
Re: (Score:2)
Well, I think you're being excessively dismissive, because IIRC it's not even ready for production use yet. It sounds like they're working on an approach to snapshooting that's a little different, and if that works out great. Of course it might not work out, but it's not a problem for you because there's no way you should even consider using bleeding edge stuff like this on production systems. That doesn't somehow detract from the value of this project.
Reaslistically, even if this thing offered greater pe
Re: (Score:2)
I hate to just be crass and dismissive, but it doesn't even have snapshots. I will never again use a filesystem that doesn't support snapshots for anything other than a toy. I've used ZFS since it was merged to FreeBSD and still have my first raidz2 pool in operation with 100% uptime and no data loss.. Why would I ever switch??
Cute. Sh1tt1ng a brick in public over a lack of a feature in a NEW filesystem that is still under active development.
FWIW ... nothing ever gets merged into the Linux kernel as 100 percent feature complete; it is always a work in progress.
Re: (Score:2)
Cute. Sh1tt1ng a brick in public over a lack of a feature in a NEW filesystem that is still under active development.
FWIW ... nothing ever gets merged into the Linux kernel as 100 percent feature complete; it is always a work in progress.
Which makes me wonder why it is newsworthy that it is included. Yeah the GP was a bit crass but the reality is we're having this discussion because it was posted on a news site, so it is a value added discussion to point out that it is not feature complete, especially compared to the filesystems TFS is comparing them to.
Re: (Score:2)
Re:well alright then (Score:5, Interesting)
Release early, release often. This is good and necessary.
Something to be aware of in this: Linux file system development has been stagnant recently. Linux has lost a lot of the file system development horsepower it use to have. This is mostly due to the cloud. Fewer people care about Linux file systems because the cloud operators build their own storage platforms that deliver redundancy, integrity, snapshots, etc., and in such an environment the file system used by host OS doesn't matter much. So fewer organizations are paying staff to spend time on new kernel file system code.
Perhaps this can inspire talented people to work on Linux file systems again.
Re: (Score:3)
The last major filesystem work was when Meta spent a ton of man-hours fixing btrfs bugs, bringing that FS some stability, up to and including the raid5/raid6 write hole (non)-issue.
The big thing is if Red Hat can toss this filesystem, or a modern filesystem with FS level snapshots, checksumming, and other items, as opposed to having to do a stack of stuff:
1: dm-integrity on each drive, so md-raid gets a hard error if it gets a bad checksum.
2: md-raid for the heavy lifting.
3: kmod-kvdo for compression/ded
Re: (Score:2)
While you gave a good list, most people will not even use half of those features... but anyway you have more modern alternatives:
1 and 2 - btrfs support raid 1 without any problem
3- btrfs support dedupe
4- btrfs directly do no t support ssd cache, you will have to add bcache to your btrfs setup
5- btrfs also do not directly support encryption, you need to add luks below the btrfs
6- btrfs support data checksum, snapshots and shrink
yes ,btrfs isn't perfect, encryption is the main missing piece of the puzzle mi
Re: (Score:2)
I think he described Stratis, which seems to somehow wrap all these tools into some kind of system.
Re: (Score:1)
add drbd and you also get online replication!
Re: (Score:2)
I see no problem with a device mapper stack as part of a solution as long as tooling hides it. It's not like more complex ZFS configurations are trivial to maintain either, if the complexity is not hidden by a good toolset it's only a small improvement.
The tools for creating and maintaining device mapper stacks for redundancy and snapshotting are absent or trash, but that's not device mapper's fault.
Re: (Score:2)
This is something I wish Larry would do... offer ZFS with a GPL 2.0 license, so OpenZFS (a downstream of actual ZFS) could be included in the mainstream Linux kernel.
Re: (Score:2)
> OpenZFS (a downstream of actual ZFS)
It's upstream now for at least four platforms.
Oracle ZFS is basically forked away forever.
Jonathan Swartz made licensing promises and reneged. Trying to upstage Steve Jobs caused problems as well.
Re: (Score:2)
Obviously still a lot of work to go but linux will finally get a filesystem with ZFS like features with a friendly license.
Will we? I mean we already "have" BTRFS but it still isn't considered fully stable 14 years after it was added to the kernel.
Did Kent learn to work with people? (Score:2, Troll)
Re: (Score:2)
he did learn, and there are more people helping, so things went actually smooth for the amount of changes that this required... if only it was submitted to the -next first (that it ended to do later on) , it would have been even smoother
The name (Score:3)
I was trying to figure out what a "Bca - chef" was until I saw it was a filesystem and the B-cache-fs moniker made sense.
Re: (Score:2)
How do you pronounce it?
Re: (Score:3, Informative)
Bitch eff ess
Re: (Score:2)
Re: (Score:2)
bee cash ef es
Re: The name (Score:2)
Ah I see, I was going with buh-cach-é fs
Re: (Score:2)
I read it as beach-f-s
Surprising (Score:2)
Merge all the things! (Score:2)
Linux will a fat OS if it continues like this.
I should provide my FlappyBird commandline games as well for having it included into the kernel.
Re:Merge all the things! (Score:4, Funny)
Linux will a fat OS if it continues like this.
I should provide my FlappyBird commandline games as well for having it included into the kernel.
Into the kernel? No.
Into the systemd? Hell yeah!
Intriguing (Score:2)
With OpenZFS now being developed on Linux, it looks like there will be some serious competition in the filesystem sector. (Is OpenZFS still run as a userland service, or is it a Kernel module now?)
The filesystem work won't help Linux in the desktop arena, but it will help it in the server arena, as Windows Server's filesystem is increasingly primitive and slow relative to the competition.
Sounds good (Score:2)
I think I'll use it in 2033.