I currently have a storage server with the following config.

Multiple raid6 volumes (mdadm) -> aggregated into a lvm volume group -> lvm volumes -> encrypted with luks1 -> (no partitioning) xfs file systems mounted and used by the os

I have the following criteria: I want to keep software raid (mdadm) with multiple raid sets, xfs, and lvm. I don’t mind using 2fa, but I don’t want to just store my secret keys on a dongle attached to my PC because that seems to defeat the point of encryption at rest.

My questions:

  1. Is there a better way to encrypt my data at rest?

  2. Is there a better layer at which to apply the encryption?

I’m mostly unhappy with luks1 over a whole lvm volume and looking for alternatives.

Thank you everyone for these great responses! I’ll be looking into these ideas :)

  • constantokra@lemmy.one
    link
    fedilink
    English
    arrow-up
    7
    ·
    9 months ago

    Encrypt the boot drive, and use dropbear ssh in initramfs to be able to unlock it over ssh during boot. Then set up your data drives however you want, and use a key file on your boot drive to unlock them, once you’ve unlocked it. All drives are encrypted when your machine is off, and you only need one password you can enter remotely to unlock the whole thing.

    Here’s a good resource on how to do the initramfs part https://www.arminpech.de/2019/12/23/debian-unlock-luks-root-partition-remotely-by-ssh-using-dropbear/

    Also, when you update the kernel you have to rebuild the initramfs with sudo update-initramfs -k all -u, or it won’t be able to boot to the new kernel.

    I’ve found it to be a super reliable setup.

    • ShortN0te@lemmy.ml
      link
      fedilink
      English
      arrow-up
      3
      ·
      edit-2
      9 months ago

      Have not looked through the setup steps of that link, but using FDE with luks and remote ssh unlock for years and have not had any problems.

      Also, when you update the kernel you have to rebuild the initramfs with sudo update-initramfs -k all -u, or it won’t be able to boot to the new kernel.

      Shouldn’t that be automatically handled by apt? I dont remember that i have setup a hook for that and i never rebuild my initramfs manually.

      • constantokra@lemmy.one
        link
        fedilink
        English
        arrow-up
        1
        ·
        9 months ago

        I was a bit surprised at it as well, but it doesn’t for me running Debian headless. If I reboot after a kernel update it’ll try to boot into the new kernel and fail waiting for the initramfs, but it’ll boot just fine into the previous kernel. Once I update the initramfs it works fine.

        If you know what resources you used to set it up, I’d be curious to take a look and see if I missed something.

        • ShortN0te@lemmy.ml
          link
          fedilink
          English
          arrow-up
          2
          ·
          9 months ago

          Steps are basically not more then this (Can not find the original blog i followed but this is the small write up i have made years ago)

          • install dropbear
          • update config to your liking
          • copy public ssh keys over
          • run update-initramfs -u (has to be rerun on config change)
          • done (for the server part)

          For some reason i install busybox too in the personal write up. But i do not think it is necessary.

          • constantokra@lemmy.one
            link
            fedilink
            English
            arrow-up
            1
            ·
            9 months ago

            That’s basically the same as my writeup from when I did it. Except I also had a -k all on update-initramfs. Not sure about the switches, so I’ll look into them. Thanks.