Announcement

Collapse
No announcement yet.

Linus Torvalds Doesn't Recommend Using ZFS On Linux

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

  • Originally posted by drjohnnyfever View Post
    Another point I'd like to make is that the CDDL issue is as much a problem with ZFS as it is with Linus not permitting the kernel to be licensed under the terms of the GPLv2 or the CDDL. Yes Linux has a lot more inertia than ZFS does on that front but right now Canonical are shipping distros with ZFS available. There are no real technical issues. Linus wants to be a dick about the CDDL but he's OK with vendors linking proprietary drivers into the kernel, doesn't give a second thought to it. Its all hooey.
    Originally posted by k1e0x View Post

    This is very true.. it doesn't matter what Linus says we have ZFS for Linux right now.. all the technical problems are resolved with it being out of tree. I think that out of tree really is a blessing to be honest..

    Not sure why ppl want it in tree anymore.

    I'm using a supported filesystem and I need to upgrade my kernel.
    apt-get upgrade

    I'm using a evil out of tree <insert negative comments here> ZFS file system and I need ot upgrade my kernel.
    apt-get upgrade
    False. The very nature of being out of tree means that there will be breakage and that no updates or improvements come for free; out of tree kernel code is always trailing behind technically and the out of tree maintainers have a significantly higher burden to maintain compatibility and reliability. No one is saying you can't use it, you are probably using Ubuntu, so it is in some limited sense "supported", as you claim, but only by Canonical; from the point of view of the Linux kernel, it is quite plainly unsupported 3rd party software.

    FUD all you want about the alleged unreliability of BTRFS, but being in tree means it has the eyes of the kernel world on it and is updated with evolving kernel interfaces and tested regularly. Whichever ZFS implementation you prefer surely has a much smaller audience for audit and maintenance. It is their choice (ultimately Oracles, surely, as patent and copywrite owner on most or much of the code and ideas) to not license it appropriately for consideration in mainline Linux. Don't blame Linus for behaving reasonably and following clear and responsible standards by refusing to change the rules for a known bad player who could remedy the situation themselves.

    Linus is not "being a dick", he is expecting Oracle to play by the same rules everyone else does. He dislikes all out of kernel code, whether it is ZFS or a proprietary graphics driver - you do remember the infamous middle finger to NVIDIA, right? Not sure he's as "OK" with vendors linking proprietary drivers in as you think, he just isn't in a position to prevent them from exploiting a legal loophole. As the thread this article referred to illustrates, people whine about it when their precious out of tree kernel blob fails them in production, but how is Linus supposed to be responsible for the shortcomings of uncooperative third parties? He certainly has much more important matters to worry about within the actual Linux kernel. In the case of Ubuntu, since they encourage or support ZFS, it would fall to them to not push a kernel update for ZFS users that breaks their ZFS module until they have verified the module works with the new kernel.

    Comment


    • Originally posted by jpnc View Post
      False. The very nature of being out of tree means that there will be breakage and that no updates or improvements come for free; out of tree kernel code is always trailing behind technically and the out of tree maintainers have a significantly higher burden to maintain compatibility and reliability. No one is saying you can't use it, you are probably using Ubuntu, so it is in some limited sense "supported", as you claim, but only by Canonical; from the point of view of the Linux kernel, it is quite plainly unsupported 3rd party software.
      By ZFS not being in-tree, it simply falls to the distribution maintainers to provide a complete, supported package. The end user running Ubuntu doesn't need to have any concern over whether kernel updates will break ZFS, which makes it a very attractive distribution for storage servers.

      Comment


      • Originally posted by k1e0x View Post
        Linus clearly doesn't understand the problem.
        Perhaps, but likely you don't either. Why? Well, why is Reiser 4/5 not going to be in the kernel?

        Right or wrong, for whatever reason, the kernel (folks) don't like incorporating filesystems without a big corporate sponsor. The big corporate sponsor of ZFS (flavor be damned) is S'nOracle.

        Comment


        • Originally posted by cjcox View Post

          Perhaps, but likely you don't either. Why? Well, why is Reiser 4/5 not going to be in the kernel?

          Right or wrong, for whatever reason, the kernel (folks) don't like incorporating filesystems without a big corporate sponsor. The big corporate sponsor of ZFS (flavor be damned) is S'nOracle.
          It has zero to do with a big corporate sponsor, it's all about the license. As long as you cannot put ZFS under GPLv2, you simply cannot add it to the vanilla Linux kernel source code and still abide by the license.

          Comment


          • Originally posted by muncrief View Post
            All I want is something that can detect bit rot on single drives, without causing some other kind of calamity. I keep everything backed up every 15 minutes, and if I could simply detect bit rot I'd never lose data.

            But there simply doesn't appear to be any supported and stable file system on Linux capable of performing that simple task. It's really a shame.
            I'm pretty sure this is the main reason why there's still a market for BSD. Maybe you should have a look. I've been using ZFS for my personal fileserver for a decade now (running on FreeNAS, a fork of OpenBSD) and it's been incredibly stable, even on dirt cheap consumer hardware. I've never lost data.

            Comment


            • Originally posted by F.Ultra View Post

              It has zero to do with a big corporate sponsor, it's all about the license. As long as you cannot put ZFS under GPLv2, you simply cannot add it to the vanilla Linux kernel source code and still abide by the license.
              I probably should have been more specific. They are "the" owner, and thus, not only do that have to "sponsor", but also have to sign off on change of license.

              Comment


              • I don't give a shat about data safety. I could loose 90% of all the files in my /home and not even notice it.

                Really important stuffs are copied on a couple of computers and another couple pendrives.

                Fuck ZFS, XFS and Btrfs.
                Performance is King.

                Comment


                • Originally posted by oiaohm View Post
                  Why are you pretending that the spirit of the law means contracts and licenses can mean something entirely different than what is written? If that was the case, it would've been much less important to read the fine print.

                  I don't know why CDDL would be incompatible with GPL, but the reverse is true, at least in the sense that you can't necessarily put GPL code into a CDDL project because it might consist of proprietary software. But I don't know why CDDL software couldn't be mixed with GPL, which is the case here. License compatibility doesn't have to be a two way street and even if licenses are incompatible in and of themselves, that doesn't necessarily mean that the software is only compatible with one of the licenses.

                  Comment


                  • LOLOLOLOL "Performance is king." Yeah... because *nobody* cares when their banking records go away, or phones stop working, or the Internet goes down, or...
                    Sorry, dude. I mean, performance matters: a lot. But by-and-large, if you're not performant enough, that means you need to scale up. But *data integrity* is job 1 in the vast majority of cases. (There are exceptions such as transactional, and ephemeral data, /tmp being a great example.) But the data that people care about, people *care* about.

                    Comment


                    • Originally posted by cape View Post
                      i don't give a shat about data safety. I could loose 90% of all the files in my /home and not even notice it.

                      Really important stuffs are copied on a couple of computers and another couple pendrives.

                      Fuck zfs, xfs and btrfs.
                      Performance is king.
                      ext2 FTW

                      Comment

                      Working...
                      X