Announcement

Collapse
No announcement yet.

Reiser4 Brought To The Linux 5.0 Kernel

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

  • #71
    Originally posted by F.Ultra View Post

    What you have missed here is that it's certain configurations of btrfs that is marked as unstable (especially raid 5/6), for other uses such as raid1,0 and 1+0 it has been stable for years now.
    You are missing both the point and the argument - I am talking about none of the unstable features mentioned, Im talking about a plain CoW setup which does not involve even subvolumes or raid at all. Statistically (Google it) there is no other filesystem that are in the mainline tree with more issues than btrfs. Besides the unstable nature of it, it also has a unique set of limitations that does not exist else where in the mainlined alternatives nor reiser4, among others the fact that using btrfs + mdadm for as an alternative to the support that IS marked as unstable in btrfs' own feature set like the ones you mention, or the lack of swap-file support and even in the most traditional setups with no special features enabled there is a significant amount of bug reports and file corruption to the point where people have created threads asking if ANYONE has used btrfs without issues. That is alarming and is the point - Reiser4 by nature does not actually allow for much of these issues to exist because of the vastly different design it's built on compared to a mainlined FS which further proves my point in that it is in fact political reasons that is hindering the inclusion of reiser4 into the mainline linux kernel, and not a matter of code quality nor standards it does not mest, to be a part of the kernel. This is indeed a very solid FS for any kind of workloads and has been continiously been under active development since around the time ext4 was created. Most features ZFS support on Linux is also available in reiser4, yet ZFS is not mainlined because of the license being incompatible but that is not a problem for reiser4. The fact that its not mainlined does not mean its not a strong contender to any of the other FS available as with ZFS - Many people use ZFS and many people use reiser4 despite having to do a few extra steps to enable it. Anyway, for those using either Arch, Manjaro or other derivates (or gentoo) and are interested in trying it out, I just have just recently created a custom linux-ck based kernel with reiser4 support for both arch and gentoo (will land in AUR asap) and I also intend to start working on a livecd to make it easier as well for those who are not comfortable patching or configuring kernels on their own. Just a heads up for those who are not offended by the reiser name.

    Comment


    • #72
      Originally posted by petterk View Post
      Wow - Im surprised of how ignorant this community really is.

      "The once promising file system [...]"

      Once? Its one of the more modern approaches to innovative features and design written entirely from scratch.

      I just want to add to this that ext4, which everyone think of as a modern filesystem, does not even have feature parity in 2019 with XFS which was designed in 1993 - go figure.

      I also think it is really sad that reiserfs4 has not received the consideration it deserves. (On pure technical merit)

      Also nobody seems to think that it is a big deal that ZFS performance drops like a rock when it gets past 80% space utilization.
      I have a lot of respect for ZFS and used it a lot; both with ZOL and zfs-fuse back then. It is rock solid, but this 80% limit problem has been a thorn in my side for years.


      Comment


      • #73
        Reiser was framed, probably by those nefarious Btrfs developers. #FreeHans

        Comment


        • #74
          Originally posted by F.Ultra View Post

          What you have missed here is that it's certain configurations of btrfs that is marked as unstable (especially raid 5/6), for other uses such as raid1,0 and 1+0 it has been stable for years now.
          Tell that to my 16TB of unscrubbable, unfixable, unfsck-able RAID10 btrfs formatted drive pool.

          I had to find some drives laying around to cp from the 16TB and found that I lost hundreds of gigabytes of data.

          I saved my butt by keeping a separate backup of critical data.

          I could have lost wedding videos, including my own, -forever- and people like you keep pushing crap like btrfs.

          Comment


          • #75
            @profoundWHALE - Tell that to my 16TB of unscrubbable, unfixable, unfsck-able RAID10 btrfs formatted drive pool.

            I had to find some drives laying around to cp from the 16TB and found that I lost hundreds of gigabytes of data.
            Back up your data! How many billion times do people have to be told that raid and btrfs are not substitutes for doing your backups?

            Comment


            • #76
              Originally posted by profoundWHALE View Post

              Tell that to my 16TB of unscrubbable, unfixable, unfsck-able RAID10 btrfs formatted drive pool.

              I had to find some drives laying around to cp from the 16TB and found that I lost hundreds of gigabytes of data.

              I saved my butt by keeping a separate backup of critical data.

              I could have lost wedding videos, including my own, -forever- and people like you keep pushing crap like btrfs.
              If you think that similar things have never happened on ext4, xfs or reiserfs then my cat have a bridge to sell you. If it's anecdotes you want then I can tell you that I have plenty of storage servers at work with 110T volumes (24x9.1TiB drives in btrfs raid1 on each server) running on btrfs without any problems.

              Comment


              • #77
                Originally posted by F.Ultra View Post

                If you think that similar things have never happened on ext4, xfs or reiserfs then my cat have a bridge to sell you. If it's anecdotes you want then I can tell you that I have plenty of storage servers at work with 110T volumes (24x9.1TiB drives in btrfs raid1 on each server) running on btrfs without any problems.
                Well, that kind of defeats the purpose of the argument that is used by some comments here in that mainlined filesystems is more stable than the ones which are not in the kernel.

                With that said - this is almost exclusively a BTRFS-issue what profoundWHALE is describing as it happens to *A LOT* more people using btrfs, than for example ESPECIALLY ext4, JFS and even reiserfs, not to mention (because of it's design through inline checksumming and atomic operations) - Reiser4.

                Comment


                • #78
                  Originally posted by andyprough View Post

                  Back up your data! How many billion times do people have to be told that raid and btrfs are not substitutes for doing your backups?
                  ...meanwhile it is rarely nescessary to do backups of data being on Reiser4 filesystems because of the nature of the filesystem. Mirror/failover-subvolumes (replicas - raid1 is not), inline checksumming (sort of like the checksumming done by XFS), atomic operations, allocate on flush and a great fsck-utility. Btrfs even claims some of it's built in raid-features are STABLE - yet they have admitted the fact that using madm + btrfs for raid (in almost every scenario) is going to cause some sort of issues. That is a unique and critical design mistake. Not once have I ever heard of a filesystem not being able to support mdadm or even in some cases lvm.

                  Comment


                  • #79
                    Originally posted by Raka555 View Post

                    I just want to add to this that ext4, which everyone think of as a modern filesystem, does not even have feature parity in 2019 with XFS which was designed in 1993 - go figure.

                    I also think it is really sad that reiserfs4 has not received the consideration it deserves. (On pure technical merit)

                    Also nobody seems to think that it is a big deal that ZFS performance drops like a rock when it gets past 80% space utilization.
                    I have a lot of respect for ZFS and used it a lot; both with ZOL and zfs-fuse back then. It is rock solid, but this 80% limit problem has been a thorn in my side for years.

                    I totally agree with your arguments on both ext4 - it's foundation is by far modern and carries over (imo) way to many elements from its older predecessor ext3, but it is a somewhat safe choice given the different journaling-methods it provides, making it atleast very much possible to avoid loss of data or corruption. ZFS is a great file system for some situations, like it's integrity focus which has been there from the start of the development as well as (at least for the operating systems it originally was meant for) native encryption, subvolumes (I believe the first of it's kind to provide such a thing), "self-healing" algorithms for preventing bit-rot and corruption but also it's various features which are quite similar to what reiser4 also has added through the years of development like transparent compression, mirror/failover support and so on. However, it is far from optimal for an average desktop user - it uses surprisingly much memory (you can limit it - with a penalty in performance), fixed 4k block allocation which is the most common in almost every FS I can think of right now, Reiser4 being one I know which does not require it and therefore stores data more efficiently. But yeah. It's not, in any kind, a revolution in filesystems. A great alternative being even older, is JFS. A properly optimized JFS partition is very snappy and responsive, some claim lower CPU usage as well, and is one of the more mature file systems noone seems to care about any more. Besides reiser4, ZFS (because memory nor disk usage is a concern for me atm) or JFS (yeah) would be my choice on SSDs (not very optimal for HDDs because of the high level of fragmentations it often produces).

                    Comment


                    • #80
                      Originally posted by petterk View Post

                      ...meanwhile it is rarely nescessary to do backups of data being on Reiser4 filesystems because ....
                      Because, apparently Reiser4 users are idiots? There is no technical substitute for backups. And no excuse for not backing up. Reiser4 will not have any solution for you in the case of catastrophic hardware failure, flooding, fire, or massive electrical surge. Thinking you are safe because "Mirror/failover-subvolumes" is just sheer foolishness.

                      Comment

                      Working...
                      X