Ubuntu Developers Seem To Be Really Pursuing ZFS Root Partition Support On The Desktop

[Guest]

"Waging war" is a very absurd and hyperbolic comment. Why would you want to "wage war" on a top-notch open-source filesystem; arguably the best filesystem which is available for Linux today.

ZFS can cure cancer for all I care, the reason why it doesn't belong in Linux is the license, and that by itself is a good enough reason. I would support the idea of adding an intentional panic when the module is loaded in the mainline kernel.

[Guest]

I'm glad Linux is moving forward and getting a terrific file system out to every day users. Good job! Data integrity for everyone! .. and it's about time.

What about supporting development of state of art file systems, such as HAMMER2? ZFS is already out there, but I haven't seen people calling to start work on a port to Linux.

[Guest]

ZFS won't be re-licensed anytime soon.

Which is fine by my terms, just one more reason to hate Oracle.

[jpg44]

Its kind of odd people seem to be so interested in ZFS when Linux has its own GPL filesystem called btrfs which works very well. Why doesnt Canonical start acting like they are not insane and help improve btrfs? You probably cant legally or morally do ZFS in a distro because of the licensing. I am not fond of BSD licenses because companies should give back their improvements rather than take take take.

[starshipeleven]

All they have to do is put it in a non-free repo (even tho it does not belong there, it's more permissively licensed than Linux itself.) and they are good. The end user can decide to install whatever they want.

That's still a work around though. I personally don't care. In either case I'm not the one that risks getting sued by Oracle or any other random gold digger that contributed to the kernel. It's the distro that ships it which does risk this.

This *is* a good thing too.. It's light years ahead of NTFS.. isn't actual proprietary software the enemy? Or are we going to say your open source isn't good enough for my open source and wait for Microsoft to make something better?

Meh, the only reason I'm not switching is that there isn't yet a single goddamn distro of those that advertise ZFS that managed to make a foolproof ZFS kernel module update procedure yet without having to use a LTS kernel. Serously it's 3 packages that need to be recompiled in a specific order, and there must be a way to keep the kernel locked to the last compatible version with the ZFS package. What is so fucking hard about that.

When I have some time I'll probably try with OpenSUSE Tumbleweed and hope its superior package manager can deal with this arduous task.

ZoL also offered to replace all of Oracles code in a massive rewrite and dual-license it GPL-CDDL. Kernel dev's went silent on that offer so... it isn't a problem with the ZoL team..

So, why didn't ZoL actually go through with this project? Until they change the license so they can be included in the kernel then yes it is still their own problem.

[gmturner]

OK, just had an outburst of skepticism and tried to read the CDDL, expecting to post here some flame-bait about how everyone was complaining about too much freedom...

But, no, that license is absolutely hugging incomprehensible. Personally, I'm pretty sure I can't afford enough lawyers to run software under that license.

[starshipeleven]

OK, just had an outburst of skepticism and tried to read the CDDL, expecting to post here some flame-bait about how everyone was complaining about too much freedom...

But, no, that license is absolutely hugging incomprehensible. Personally, I'm pretty sure I can't afford enough lawyers to run software under that license.

[skeevy420]

That's still a work around though. I personally don't care. In either case I'm not the one that risks getting sued by Oracle or any other random gold digger that contributed to the kernel. It's the distro that ships it which does risk this.

Meh, the only reason I'm not switching is that there isn't yet a single goddamn distro of those that advertise ZFS that managed to make a foolproof ZFS kernel module update procedure yet without having to use a LTS kernel. Serously it's 3 packages that need to be recompiled in a specific order, and there must be a way to keep the kernel locked to the last compatible version with the ZFS package. What is so fucking hard about that.

When I have some time I'll probably try with OpenSUSE Tumbleweed and hope its superior package manager can deal with this arduous task.

So, why didn't ZoL actually go through with this project? Until they change the license so they can be included in the kernel then yes it is still their own problem.

The "project" is just a discussion for that might be a month or two old. I can understand the ZoL people not wanting to do major rewrites if it'll end up being the same situation that it currently is. It makes sense not to pull the trigger if they don't know if the Linux kernel maintainers will allow it or not.

FWIW, as of 0.8 it's just one module that is fully up-to-date with Linux 5.0rc and the zfs-utils package. We're down to two packages that all package managers can handle the dependency tracking of and there aren't anymore DKMS build order worries due to the SPL merge. It's pretty safe as long as one sticks to ZoL supported kernels...basically don't run bleeding edge git kernels without a backup kernel in place...which is something that should be done when running a bleeding edge kernel anyways...

ZFS 0.8 is the foolproof version you've been waiting for.

IMHO, the kernel locking part isn't an issue if you use a distribution that supports ZFS like Manjaro since it's the job of a distribution maintainer to make sure things like that are all compatible with each other. If the distribution is well maintained, that shouldn't happen.

[Chugworth]

Its kind of odd people seem to be so interested in ZFS when Linux has its own GPL filesystem called btrfs which works very well. Why doesnt Canonical start acting like they are not insane and help improve btrfs? You probably cant legally or morally do ZFS in a distro because of the licensing. I am not fond of BSD licenses because companies should give back their improvements rather than take take take.

Btrfs works up to a point. The RAID5/6 data loss bug has been left unfixed for quite some time now, which doesn't look good at all. And in my own testing I've seen issues when trying to rebuild mirrored drives if the drive is nearly full of data. When contemplating a data storage upgrade, I set up a virtual machine with several virtual drives and did some comparison testing between Btrfs and ZFS, and I ended up much more confident in ZFS.

And of course, if Btrfs is Linux's answer to ZFS, then it sure doesn't help for Red Hat to drop it completely.

[starshipeleven]

FWIW, as of 0.8 it's just one module that is fully up-to-date with Linux 5.0rc and the zfs-utils package. We're down to two packages that all package managers can handle the dependency tracking of and there aren't anymore DKMS build order worries due to the SPL merge.

Ah that's good news finally.

I'm really getting annoyed by how Btrfs upstream is still very far from dealing with basic shit like "a drive disappears from the array" in RAID1 and "actually dealing with a degraded array without doing a silent mess until reboot, then freaking out on boot and requiring a ton of hand-holding to fix the mess".

IMHO, the kernel locking part isn't an issue if you use a distribution that supports ZFS like Manjaro since it's the job of a distribution maintainer to make sure things like that are all compatible with each other. If the distribution is well maintained, that shouldn't happen.

I'd still prefer much more OpenSUSE over that so I think I'll try to see how stable OpenSUSE Tumbleweed can be on a ZFS array first.