Announcement

Collapse
No announcement yet.

XFS File-System Picks Up New Features With Linux 5.1 Kernel

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

  • #11
    Originally posted by Chugworth View Post
    The problem is, if I wanted a plain, simple filesystem, XFS would not be my first choice. And if I wanted a feature-filled filesystem, XFS would certainly not be my first choice. The features they add to it are just going to feel like they've been tacked-on.
    XFS make much more sense on lvm2 instead. And since lvm2 can use mdraid* transparently in background to archieve redundancy (or in some cases - performance) it makes quite a good candidate for btrfs or zfs replacement.

    *) Not sure if it's considered stable yet... I think Red Hat is working (hard) on it.

    Comment


    • #12
      Originally posted by Zucca View Post

      XFS make much more sense on lvm2 instead. And since lvm2 can use mdraid* transparently in background to archieve redundancy (or in some cases - performance) it makes quite a good candidate for btrfs or zfs replacement.

      *) Not sure if it's considered stable yet... I think Red Hat is working (hard) on it.
      something something Stratis something something

      Comment


      • #13
        Originally posted by mcilloni View Post
        I'd really like to see compression and encryption in simpler filesystems like XFS and ext4. Being able to LZ4/LZO everything transparently is fantastic and one of the main reasons I usually only deploy ZFS/Btrfs nowadays.
        Ditto. I don't mind a possible performance hit with ZFS when it means I can both encrypt everything and gzip-9 my man pages and text documents, lz4 my executables and libraries, etc all without thinking about it.

        I'd love it if XFS picked up most of the features I use ZFS for. XFS was my go-to for years and it's still what I use when I don't need ZFS or, ugh, exfat for USB drives.

        I doubt that XFS will ever be like ZFS -- a single solution that handles everything. Don't need LUKS, don't need LVM, don't need a HW raid controller....all you need is Z, Z, Z is all you need....eventually, hopefully BTRFS will be there....

        I might be a retarded ZFS advocate, but that doesn't meant I don't realize that us Linux users need a Linux file system that competes feature for feature with ZFS because, IMHO, OpenZFS + SomeBSD = Why use Linux for File Servers?

        Comment


        • #14
          We operate a 77 TB hardware RAID6 with BBU that we run ext4 on. We are happy with the speed of the array. Data is typically appended and seldomly changed. For computations, we copy data to a temporary storage (an NVME RAID0). Most important for us is data integrity.

          Are there any compelling arguments to switch to XFS? It is said to be best choice for big partitions, but is that still a factor considering ext4 improvements over ext3?

          Comment


          • #15
            Originally posted by ypnos View Post
            Are there any compelling arguments to switch to XFS?
            Not for your usecase. I mean really, you could run anything on a hardware RAID and it would be the same as the card has its own RAM cache.

            What makes or breaks the performance of a hardware RAID is the card. As long as you are using a stable filesystem on top of that it's all fine.

            Comment


            • #16
              Originally posted by jacob View Post

              I feel the same way, but XFS still has a lot going for it. It has some advanced features that ext4 doesn't offer, yet at the same time it's extremely mature and reliable, more so than Btrfs or ZFS. That makes it a good compromise. It also offers great performance in random-writes scenarios, where CoW filesystems typically suck, such as VM hosting or various database workloads. So there is definitely a place for it.
              2 questions:

              1 Does XFS perform good on SSD's? I know Michael has benchmarks, but what about the real world performance on SSD's?
              2 Is there a way to convert EXT4 to XFS without losing data (like there is for EXT4 to Btrfs conversion)?

              Comment


              • #17
                Originally posted by Vistaus View Post
                2 Is there a way to convert EXT4 to XFS without losing data (like there is for EXT4 to Btrfs conversion)?
                This works for any filesystem:

                -make a backup, just in case
                -boot a livecd or another system,
                -rsync your root partition's full contents somewhere else,
                -reformat,
                -rsync the stuff back,
                -chroot into the new system,
                -edit fstab so root is using the right UUID/label/whatever,
                -run the commands to update your kernel/initrd/whatever image
                -run the commands to update/reinstall GRUB
                -exit from the chroot
                -boot the system
                -????
                -profit

                Comment


                • #18
                  Originally posted by skeevy420 View Post
                  I might be a retarded ZFS advocate, but that doesn't meant I don't realize that us Linux users need a Linux file system that competes feature for feature with ZFS because, IMHO, OpenZFS + SomeBSD = Why use Linux for File Servers?
                  Linux has a filesystem that competes feature for feature with ZFS: it's called ZFS.

                  As for why use Linux for a file server, well, maybe not. But file servers are an infinitesimal percentage of the market, and of the Linux market in particular.

                  Comment


                  • #19
                    Originally posted by ypnos View Post
                    We operate a 77 TB hardware RAID6 with BBU that we run ext4 on. We are happy with the speed of the array. Data is typically appended and seldomly changed. For computations, we copy data to a temporary storage (an NVME RAID0). Most important for us is data integrity.

                    Are there any compelling arguments to switch to XFS? It is said to be best choice for big partitions, but is that still a factor considering ext4 improvements over ext3?
                    Holy hell just hope you never have to run a fsck on that thing. How long did it take to format? I can't imagine; maybe you have expanded it over time and didn't notice. If you want data integrity then ext would not be high on my list. The last time I ran fsck on a much smaller ext filesystem it corrupted itself and destroyed the whole thing. I've had ext systems end up dead even shutting down normally. I hate that filesystem.

                    Happy XFS user for over 20 years in my production systems. Even for small simple stuff I always go XFS because it has been so reliable for me. Nothing else comes close.

                    Comment


                    • #20
                      Originally posted by Vistaus View Post
                      1 Does XFS perform good on SSD's? I know Michael has benchmarks, but what about the real world performance on SSD's?
                      Just like with any other type of media, it depends on your use case. XFS will perform very well for some workloads and not so well for others. Besides, does it perform well in terms of what? Sequential write performace? Random write performace? Read performace? File access per number of files? Internal fragmentation? Large or small files? Etc etc etc.

                      Originally posted by Vistaus View Post
                      2 Is there a way to convert EXT4 to XFS without losing data (like there is for EXT4 to Btrfs conversion)?
                      Not that I know of. Even the Ext4 to Btrfs conversion is a proof of concept more than anything else; in practice it's certainly not recommended to go that route.

                      Comment

                      Working...
                      X