Announcement

Collapse
No announcement yet.

Ubuntu 23.10 Restores ZFS File-System Support In Its Installer

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by Luke_Wolf View Post

    As much as I usually disagree with timofonic, no no, he's got a point.

    Unless Oracle decides someday that it's suddenly worth relicensing to something that is compatible with the GPL which can thus be merged into Linux, or FreeBSD finally manages to take off in a big way it is 100% a dead end. "Oh but it's compatible between Linux, FreeBSD, and OpenIndiana" really isn't a selling point. If it was compatible with Windows it might be, but it's not. It's very very far from being a "cross platform universal format" and before you start trying to point to this https://github.com/openzfsonwindows/openzfs it's not even close to being in a stable state and only has a couple people working on it.
    You can literally **click, click, click** and install OpenZFS 2.2.0-rc4 on Windows right now. Yeah, it might not be the most stable, but its been under development for six years and the developer has been actively contributing to OpenZFS proper for three years. You can also use OpenZFS on Windows with WSL2. It's a little more complicated, but it uses the Linux driver and works quite well. OSX has a package for 2.1.6 and source tags for 2.1.7.

    A lot of that work for both Windows and OSX is done by lundman, a person working for an ISP in Japan that's put in 6 years of work in porting ZFS to more platforms. Back in the day there was a certain project from the Linus feller, a single developer, that ported a kernel to a new platform and it wasn't the most stable of things...then other people helped out and now we're 4 days away from its 32nd Anniversary. That old, buggy kernel from a single developer now powers the most powerful computing devices in the world and has hundreds of contributors. Bcachefs has like one or two people working on it. Only has a couple of people working on it is the Marvel Origin Story of nearly every Open Source Project.

    I don't think anyone cares about the OpenZFS licensing outside of Linux/GPL people trying to make points that have much ado about nothing outside of philosophical debates. The licenses conflicting don't stop it from being a module, it doesn't stop allowing end-users to bake it into the kernel themselves, and it doesn't stop distributions from hosting the modules in a separate package.

    Comment


    • #12
      Originally posted by Luke_Wolf View Post

      As much as I usually disagree with timofonic, no no, he's got a point.

      Unless Oracle decides someday that it's suddenly worth relicensing to something that is compatible with the GPL which can thus be merged into Linux, or FreeBSD finally manages to take off in a big way it is 100% a dead end. "Oh but it's compatible between Linux, FreeBSD, and OpenIndiana" really isn't a selling point. If it was compatible with Windows it might be, but it's not. It's very very far from being a "cross platform universal format" and before you start trying to point to this https://github.com/openzfsonwindows/openzfs it's not even close to being in a stable state and only has a couple people working on it.
      I don't see how it not being GPL even matters. Who cares about this? ZFS has the most complete featureset of any filesystem I've seen.

      Comment


      • #13
        Originally posted by skeevy420 View Post

        Back in the day there was a certain project from the Linus feller, a single developer, that ported a kernel to a new platform and it wasn't the most stable of things...then other people helped out and now we're 4 days away from its 32nd Anniversary. That old, buggy kernel from a single developer now powers the most powerful computing devices in the world and has hundreds of contributors. Bcachefs has like one or two people working on it. Only has a couple of people working on it is the Marvel Origin Story of nearly every Open Source Project.

        I don't think anyone cares about the OpenZFS licensing outside of Linux/GPL people trying to make points that have much ado about nothing outside of philosophical debates. The licenses conflicting don't stop it from being a module, it doesn't stop allowing end-users to bake it into the kernel themselves, and it doesn't stop distributions from hosting the modules in a separate package.
        You accidentally made a point. Bcachefs might be the Linux of filesystems too in the near future

        OpenZFS licensing is an issue. But the reasoning is pragmatic too. The shim layer and not being integrated adds severe overhead too, for example. ZFS is very RAM hungry too.

        Anyway, I'm happy ZFS exists. It motivated others to be innovative and it's an interesting solution right now.

        BSDs are a niche in comparison to Linux.

        Windows and macOS ports are not part of OpenZFS main tree, that's an issue. I wouldn't use those ports in a daily production or stable desktop system, those are just nice gimmicks to me.

        Comment


        • #14
          Originally posted by Veto View Post
          LUKS + BTRFS instead please
          Like no? I spent 6 hours yesterday debugging an Oracle Linux install that said it was only 90% full with df -h on BTRFS / but there was no space to write anything to disk. Whenever I tried to grow the underlying disk in the cloud platform at work it would lock up. After three people we finally got that server fixed! In the future it is LVM and XFS on RHEL, not of this Oracle Linux BS with BTRFS!

          Comment


          • #15
            Originally posted by timofonic View Post
            OpenZFS licensing is an issue. But the reasoning is pragmatic too. The shim layer and not being integrated adds severe overhead too, for example. ZFS is very RAM hungry too.
            What? Storage is the slowest component. ARC using RAM you aren't doing anything with is good and intentional. You can also tune the ARC limit to whatever you want.

            Comment


            • #16
              Originally posted by timofonic View Post
              You accidentally made a point. Bcachefs might be the Linux of filesystems too in the near future
              you guys have way too high expectations on Bcachefs.

              Comment


              • #17
                Originally posted by kylew77 View Post
                Like no? I spent 6 hours yesterday debugging an Oracle Linux install that said it was only 90% full with df -h on BTRFS / but there was no space to write anything to disk. Whenever I tried to grow the underlying disk in the cloud platform at work it would lock up. After three people we finally got that server fixed! In the future it is LVM and XFS on RHEL, not of this Oracle Linux BS with BTRFS!
                It's basic knowledge that df -h doesn't report accurate readings on Btrfs. You cannot seriously claim to be a system admin running Btrfs disks if you don't know that. Don't blame the fs when you have zero idea of what you are doing yourself.

                Comment


                • #18
                  Originally posted by darcagn View Post
                  This is great news. Ubuntu's inclusion of ZFS makes it a wonderful liveCD to install Gentoo root-on-ZFS.
                  Debian's fine as well, dkms exists.
                  ## VGA ##
                  AMD: X1950XTX, HD3870, HD5870
                  Intel: GMA45, HD3000 (Core i5 2500K)

                  Comment


                  • #19
                    Originally posted by darcagn View Post
                    This is great news. Ubuntu's inclusion of ZFS makes it a wonderful liveCD to install Gentoo root-on-ZFS.
                    Is it a paid installer DLC ?

                    Comment


                    • #20
                      Originally posted by cynic View Post

                      you guys have way too high expectations on Bcachefs.
                      Why not?

                      You never should underestimate the stubbornness of a software engineer.

                      Linux achieved success with just the creation of an humble UNIX kernel clone.

                      Also, Kent worked at storage divisions of big corpos such as Google. I'm sure the so much crap saw in their infrastructures motivated him to start the insane task of making a new and modern filesystem for Linux.

                      Anyway, sorry for the offtopic...

                      Comment

                      Working...
                      X