Announcement

Collapse
No announcement yet.

There's Talk Again About Btrfs For Fedora

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

  • #76
    Originally posted by AdamW View Post
    btrfs isn't really just about performance, it's about capabilities. btrfs is hugely different from a simple partition format like ext4; btrfs incorporates volume management and redundancy and all sorts of other features that are usually layered on top of simple formats with tools like LVM and mdraid. The capabilities btrfs brings to the table are really useful for distributions, which is why there's always a desire to make it default, but the tools and performance may well need to catch up before this is plausible.
    Btrfs snapshots alone would really simplify the murky waters that are Linux system restore utilities right now. A decent UI on that, scheduled snapshotting, and easy restore would kick the crap out of other options.

    Tangentially, I just had a reaffirming interaction with btrfs. My main machine lost power, btrfs had a superblock go bad and would segfault on boot, from my recovery disk brfsck --repair fixed it easy. Once that tool becomes the mainline fsck.btrfs it would have recovered no problem. Promising!

    Comment


    • #77
      Originally posted by ryao View Post
      1. btrfs' uses hashes as well, but it has 32-bit hashes on 32-bit userlands and 64-bit hashes on 64-bit userlands. As long as there are no collisions, you can verify the integrity of data, but the probability of a collision is quite high, especially in the 32-bit case.
      2. btrfs has no ditto blocks, so if metadata is corrupted, there is a strong possibility that it will be unable to recover.
      I have been informed by the btrfs developers that these two points are wrong. The checksums are CRC32C on all platforms, which make them weaker than I thought. Also, btrfs does have ditto blocks and uses them by default on everything but what it detects to be a SSD. This is better than not having them at all as I had previously been led to believe.

      Comment


      • #78
        Originally posted by ryao View Post
        I have been informed by the btrfs developers that these two points are wrong. The checksums are CRC32C on all platforms, which make them weaker than I thought. Also, btrfs does have ditto blocks and uses them by default on everything but what it detects to be a SSD. This is better than not having them at all as I had previously been led to believe.
        CRC32C is not bad for data checking. If there is an error there is only 1 in 4 billion chance that BTRFS does not notice. If you get enough disk errors that there is a chance of one getting through that then you have bigger things to worry about. There is no point in requiring a cryptographic level hash, because if someone is about to tamper with the data on your disk they can just as easily tamper with the checksums.

        A glance at the ECC RAM page on wikipedia seems to say that you get 8bits of checksum for each 64bit word. So by my maths 1 in 256 RAM errors would go undetected.

        Comment

        Working...
        X