Announcement

Collapse
No announcement yet.

Linux 6.11 To Introduce Block Atomic Writes - Including NVMe & SCSI Support

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

  • Linux 6.11 To Introduce Block Atomic Writes - Including NVMe & SCSI Support

    Phoronix: Linux 6.11 To Introduce Block Atomic Writes - Including NVMe & SCSI Support

    The Linux kernel's block subsystem for storage is poised to introduce support for atomic writes with the upcoming Linux 6.11 merge window...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Happy to see new innovations in XFS

    Been using it for countless amount of years throughout a lot of power outages

    Zero issues thus far and performance is incredible (like copying 100gb directory within the same drive being instantaneous)
    Last edited by Kjell; 07 July 2024, 08:43 AM.

    Comment


    • #3
      so much for silverblue i guess?

      Comment


      • #4
        Originally posted by ol3geezer View Post
        so much for silverblue i guess?
        How is fedora silverblue related to this?

        Comment


        • #5
          Originally posted by Kjell View Post
          Happy to see new innovations in XFS

          Been using it for countless amount of years throughout a lot of power outages

          Zero issues thus far and performance is incredible (like copying 100gb directory within the same drive being instantaneous)
          Did you try F2FS as comparison?

          I never used XFS other than testing, I just didn't see the point. Maybe I'm missing out?

          Comment


          • #6
            Originally posted by Weasel View Post
            Did you try F2FS as comparison?
            Yes for a couple of months

            I live in the country side where bad weather frequently kills the electricity

            Hence why I encountered fsck errors on weekly basis, well at times even after proper shutdown

            This wasn't sustainable. I moved to XFS and never had such issues again

            Quick benchmark revealed no performance regressions at the time and motivated me to take the leap

            Comment


            • #7
              Originally posted by Kjell View Post
              Been using it for countless amount of years throughout a lot of power outages
              Oh, it didn't always handle those. I can definitely tell you that!

              Comment


              • #8
                Originally posted by Weasel View Post
                I never used XFS other than testing, I just didn't see the point. Maybe I'm missing out?
                It scales well and is reasonably fragmentation-resistant, so long as you continually reserve some spare capacity (I think ~5% is enough).

                Comment


                • #9
                  Originally posted by Kjell View Post
                  performance is incredible (like copying 100gb directory within the same drive being instantaneous)
                  That is reflinks.

                  Similar to hardlink, but you get CoW behaviour, so when one copy diverges it writes new blocks, while sharing any others in common with the other data you copied from (which likewise can diverge, the same inodes are shared via reflink but neither file owns them).

                  It is bound to the scope of a mount AFAIK (well filesystem partition), not the drive/disk though. IIRC, one caveat is you can't easily copy/sync to another XFS disk (like a USB backup drive), at least rsync doesn't support reflink copies last I checked... So if you had quite a lot of reflinks in use, a 1TB disk might actually need quite a bit larger size to make a copy to a new disk/partition.
                  Last edited by polarathene; 07 July 2024, 07:38 PM. Reason: Added caveat

                  Comment


                  • #10
                    Originally posted by coder View Post
                    Oh, it didn't always handle those. I can definitely tell you that!
                    I can imagine

                    Just yesterday I lost power (for 2 hours ughh) whilst working on a large unsaved VSCodium document.. Nothing was lost.

                    Really appreciate Codium's cache and all the work behind XFS 🙏

                    Surge protected power strip has worked nicely thus far, in sense of no damaged hardware. Next step is a uninterruptible power supply

                    Comment

                    Working...
                    X