4 ms·

Linux block devices: hints for debugging and new developments

• rwmj 5y ago
  Disappointed this doesn't talk about nbdkit, the easiest way to spin up block devices for testing: https://libguestfs.org/nbdkit-loop.1.html https://libguestfs.org/nbdkit-loop.1.html eg. Make a 7 exabyte disk: $ sudo modprobe nbd $ nbdkit -U /tmp/sock memory 7E $ sudo nbd-client -unix /tmp/sock /dev/nbd0 $ sudo mkfs.btrfs /dev/nbd0 $ sudo mount /dev/nbd0 /tmp/mnt $ df -h /tmp/mnt Filesystem Size Used Avail Use% Mounted on /dev/nbd0 7.0E 3.4M 7.0E 1% /tmp/mnt I even did a talk about it: https://archive.fosdem.org/2019/schedule/event/nbdkit/ https://archive.fosdem.org/2019/schedule/event/nbdkit/
  • project2501a 5y ago
    hot damn, i wanna see mkfs work on a 7 exabyte disk...
    • rwmj 5y ago
      mkfs.btrfs is fine - I didn't make that up, that's exactly the output when I ran those commands on my laptop just now. Other filesystems don't handle this situation very well or at all. Anyway the real power of nbdkit is not that you can create huge disks, but the plugin/filters system, so you can write highly customized block devices for specific use cases, and do things like error injection and test compression or deduplication strategies.
  • globalc 5y ago
    Ah, valid point, mea culpa. I have not used nbd, but I'm aware of it. Was very impressed to see even a NetBSD implementation. There is just so much around in that area. In my defense, the storage related PM and colleagues who reviewed the article did not bring up nbd either.