Announcement

Collapse
No announcement yet.

There's Talk Again About Btrfs For Fedora

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

  • #51
    Originally posted by xeekei View Post
    No, I'm telling you that you missed the point. Mason needs to convince the distros, while Sun only needed to convince themselves. It's a lot easier to convince yourself than other people.

    IF (you do know about hypothetical scenarios, I hope?) Chris Mason had his own distro that was big and respected, he probably would've used Btrfs on it already. Putting the development speed effectively on par with ZFS, or at least close to it. Now stop playing internet bully, no one likes those.
    I doubt that would have much of an effect. The original ZFS developers did not have to deal with people breaking kernel APIs every 3 months. The btrfs developers do. Also, the original ZFS developers had dtrace to assist them while the btrfs developers do not.
    Last edited by ryao; 01-18-2013, 01:31 AM.

    Comment


    • #52
      Originally posted by ryao View Post
      I doubt that would have much of an effect. The original ZFS developers did not have to deal with people breaking kernel APIs every 3 months. The btrfs developers do. Also, the original ZFS developers had dtrace to assist them while the btrfs developers do not.
      Which doesn't matter because btrfs is in the tree and not a 3rd party external blob. Breaking APIs is a benefit for his project, not a negative.

      As for dtrace - that's seriously your argument? How about, Sun poured in a bunch more resources and manpower than btrfs has had? That's a much better argument.

      Comment


      • #53
        On the topic of general Btrfs, I'm very happy for the fact that I'll be building what amounts to a NAS server, which means that it will be able to use Btrfs internally and connect to clients via NFS. Right now for my storage needs I need to use NTFS, as for the occasional times that I am forced to use Windows (mostly game-rated things), it reads nothing else but NTFS, FATs and UDF. Also with an addon it can read specifically-formatted EXT3. But it does have NFS share support - thus I'll effectively get to use Btrfs from Windows. Fun times!

        Also, I just ran btrfsck today to check if everything is still going smoothly on my openSUSE installation. And it is, no errors. I did see some errors in the past, but those related to bad RAM and false positives.

        And speaking of openSUSE and Btrfs, I find it too bad that Snapper is not yet ported to some of the other distros. For one, it's not in Portage. And I'll need it for my NAS server, running Gentoo. I don't suppose there will be any issues manually compiling it, though. And perhaps I could look at how adding things to Portage works, while I'm at it.

        Originally posted by jwilliams View Post
        Which is irrelevant to my comment that btrfs still does not have a conservative distro that uses btrfs as the default filesystem.

        It took 4 or 5 years for ZFS to become the default filesystem in Solaris.

        btrfs development is going on 6 years now, and no conservative linux distro trusts btrfs enough to make it the default filesystem.
        That makes btrfs slower in development than ZFS. And I see no sign that things are speeding up for btrfs. I wonder if btrfs will ever be good enough to use as the default filesystem for a conservative linux distro.
        Except that Solaris is not a Linux distribution. Now, do you know of any Linux distributions (conservative ones, at that) that use ZFS as the default?

        Comment


        • #54
          Its worth having a look at some of the data corruption reports on the mailing list, eg
          http://thread.gmane.org/gmane.comp.f...ms.btrfs/22243
          * User decides to try out BTRFS, finds that files get corrupted
          * Devs suggest some test cases to narrow down issue.
          * User finds that tests case give corruption on ext4 as well
          * Turns out to be hardware problem

          Moral: there are hardware issues that you may not notice for a long time with ext4. btrfs will spot them straight away as i checksums all your data.

          Comment


          • #55
            Originally posted by xeekei View Post
            Okay, you are still not getting it. The difference is that in Btrfs case, these "widely trusted, large distros" didn't invent it. While in ZFS case, it did. I'm talking about uneven playing fields. I'm trying to get you to stop bashing the development time.
            I absolutely saw what you wrote, and it is complete nonsense. None of the widely used, trusted linux distros will make btrfs the default filesystem because btrfs is not trustworthy.

            So you say, if only we had a distro made by the same untrustworthy people that develop btrfs, then btrfs could be the default filesystem for that untrustworthy distro. It's magic! The untrustworthy btrfs combined with an untrustworthy distro somehow makes btrfs trustworthy? Two untrustworthies make trustworthy?

            With logic like that, you should be a politician.

            Comment


            • #56
              Originally posted by GreatEmerald View Post
              Except that Solaris is not a Linux distribution. Now, do you know of any Linux distributions (conservative ones, at that) that use ZFS as the default?
              Solaris is a widely trusted OS. Whether it is linux is irrelevant.

              If the ZFS license was compatible with the GPL, there probably would be linux distros using ZFS as default. Assuming that ZFS on linux was as stable as on Solaris. Which it is not, yet.

              Comment


              • #57
                Originally posted by jwilliams View Post
                So you say, if only we had a distro made by the same untrustworthy people that develop btrfs, then btrfs could be the default filesystem for that untrustworthy distro. It's magic! The untrustworthy btrfs combined with an untrustworthy distro somehow makes btrfs trustworthy? Two untrustworthies make trustworthy?
                Nope. What they are saying is that due to the fact that ZFS was developed by the same company that developed Solaris, they are clearly biased towards it. Thus the fact that it's used by default on Solaris does not mean that ZFS is trustworthy.

                Originally posted by jwilliams View Post
                Solaris is a widely trusted OS. Whether it is linux is irrelevant.

                If the ZFS license was compatible with the GPL, there probably would be linux distros using ZFS as default. Assuming that ZFS on linux was as stable as on Solaris. Which it is not, yet.
                Right. In that case, you ought to be asking "are there any widely trusted OSs that run Brtfs by default?" instead. And since we have established that Solaris developers could be biased, you need to answer the question "are there any widely trusted OSs beside those developed by Sun/Oracle that run ZFS by default?"

                Comment


                • #58
                  Originally posted by GreatEmerald View Post
                  What they are saying is that due to the fact that ZFS was developed by the same company that developed Solaris, they are clearly biased towards it. Thus the fact that it's used by default on Solaris does not mean that ZFS is trustworthy.
                  Nonsense. Solaris is a widely-trusted OS, and whether or not the developers might be biased is irrelevant.

                  The fact that ZFS is used as default in a widely-trusted OS does mean ZFS is trustworthy.

                  The fact that btrfs is not used as default in any widely-trusted OS means that btrfs is not trustworthy.

                  This is not rocket science. But feel free to keep grasping at straws and posting nonsense.
                  Last edited by jwilliams; 01-18-2013, 11:41 AM.

                  Comment


                  • #59
                    Originally posted by jwilliams View Post
                    Nonsense. Solaris is a widely-trusted OS, and whether or not the developers might be biased is irrelevant.

                    The fact that ZFS is used as default in a widely-trusted OS does mean ZFS is trustworthy.

                    The fact that btrfs is not used as default in any widely-trusted OS means that btrfs is not trustworthy.

                    This is not rocket science. But feel free to keep grasping at straws and posting nonsense.
                    You sir have won the most ridiculous poster of the week award.

                    The truth is that no filesystem is magically 100% trusted or untrusted.

                    If Fedora moved the default btrfs tomorrow, would you say it was completely untrusted today and completely trusted tomorrow? Of course not.

                    The truth is that it's 99.9% trusted right now, and developers are just waiting a while until it gets 99.99% trusted before they deploy it more widespread.

                    Perhaps Sun had a different trust cutoff that they required. Or perhaps the FS it was replacing at the time just wasn't as trusted as EXT4 is by current linux distros (since the replacement doesn't need to be 100% trustworthy, just equivalent to whatever it is replacing), which is probably more likely. Or perhaps btrfs isn't as solid as ZFS was - the point is that you can't take a simple statement like you are and try to spin it into proof of something that it's not. Real life is messy, not black and white.

                    Comment


                    • #60
                      Originally posted by smitty3268 View Post
                      You sir have won the most ridiculous poster of the week award.

                      The truth is that no filesystem is magically 100% trusted or untrusted.

                      If Fedora moved the default btrfs tomorrow, would you say it was completely untrusted today and completely trusted tomorrow? Of course not.

                      The truth is that it's 99.9% trusted right now, and developers are just waiting a while until it gets 99.99% trusted before they deploy it more widespread.

                      Perhaps Sun had a different trust cutoff that they required. Or perhaps the FS it was replacing at the time just wasn't as trusted as EXT4 is by current linux distros (since the replacement doesn't need to be 100% trustworthy, just equivalent to whatever it is replacing), which is probably more likely. Or perhaps btrfs isn't as solid as ZFS was - the point is that you can't take a simple statement like you are and try to spin it into proof of something that it's not. Real life is messy, not black and white.
                      Mac OS 5 used HFS, why don't we all use that, it is trustworthy, right? A major OS had it as default! We don't need new file systems with features and crap! Why aren't we all using fat12 still?

                      Comment

                      Working...
                      X