from Hacker News

Make Your Own Read-Only Device with NetBSD

by jaypatelani on 9/10/24, 7:30 AM with 12 comments

• by evanjrowley on 9/10/24, 7:35 PM

  This is an awesome guide and I appreciate it including ZFS.

  One thing I'm curious about is the deletion of this file:

    var/db/man.db
  

  I don't have first-hand experience with NetBSD but their online documentation seems pretty great. Does that also translate to the manpages? How could man.db be preserved for a read-only system?

  I'm toying with the idea of using NetBSD as a server for some fundamental network services. My goal is a scenario where I get Bus Factor'd and whoever takes over says:

    The previous guy left, but we should be able to run with this for now. At first we were worried, but it turned out to be just a self-contained Unix system. The thing hasn't been rebooted in 20 years, but the documentation's complete enough for ChatGPZ to understand it. Keep in mind this doesn't support Nuralink, so we'll have to dig out the old keyboard. Don't forget the USB-D adaptor.
  

  * In an alternative future, NetBSD will be ported to Nuralink.

• by nonrandomstring on 9/10/24, 9:24 AM

  Good article imho. Read-only with "ramdisk" (volatile memory filesystems) is a much overlooked defence. I am amazed at the number of IoT devices I've picked up in second hand stores that are loaded with state, secrets and PII. I got a GPS device with a large company's entire delivery address book, and a VoIP phone containing another company's entire internal voice network with names, office numbers, and personnel notes.

  Put this stuff into a capacitor backed volatile disk with the main OS running from non-volatile. Set the current drain so that it stays for at most 7-14 days after power is removed.

• by tmountain on 9/10/24, 10:03 AM

  I’m curious what the primary use case is for NetBSD these days. Is it mostly for embedded devices (routers, etc)?