I'm no good at typical prepper activities like constructing a shelter or stockpiling food. But I can at least play at being a digital prepper by, say, setting up a mirror of the Debian archive. These days, that's doable for individuals with a high-speed (e.g. fiber) Internet connection and a few hundred dollars to spend on storage. The trick, of course, is to make sure that the project isn't wasted, since I'm consuming substantial bandwidth from another mirror at least on the initial pull.
And my Debian mirror is set up. https://mirrors.mwcampbell.us/debian/ (Also available over HTTP without TLS, as is conventional for Debian mirrors.) CD/DVD images here: https://mirrors.mwcampbell.us/debian-cd/ Both archives are also available via anonymous rsync, also at mirrors.mwcampbell.us, so you can set up your own mirror using mine as detailed here: https://www.debian.org/mirror/ftpmirror
Yeah, I only have a 1 gigabit connection. The project was probably more a nerdy diversion than something actually useful.
I still have almost 2 TB left on this server, and am not planning to store nearly that much personal data, so I could also mirror or archive more stuff. Not planning to do any more Linux distros, since Debian is already probably the most comprehensive.
BTW, this whole activity was inspired by this article: https://thomashunter.name/posts/2025-03-23-post-apocalyptic-computing and the Lobsters comment thread, particularly this comment: https://lobste.rs/s/avwtt3/post_apocalyptic_computing#c_wtnoyy though that commenter pointed out that in a real post-apocalyptic scenario, the people still using computers would probably mostly be running Windows on whatever hardware they could scrounge.
So again, probably more fun than actually useful, but the process of setting up a Debian mirror *was* interesting.
I could mirror a Windows package repository, like the one for Chocolatey, Scoop, or WinGet, if there's a straightforward way to do that.
I think my next mirroring project will be the rustup and crates.io repositories, via https://github.com/panamax-rs/panamax
I don't actually know yet how much disk space that will take, but at least I'm only mirroring the latest toolchains, and only the most popular platforms, not all the tier 3's.
I figure I should call attention to the specs of this server. It's a Quartz64 single-board computer with a quad-core ARM Cortex-A55 processor (about the lowest-end arm64 processor you can get), 4 GB of RAM, 64 GB eMMC for the OS partition, and a 4 TB USB 3.0 SSD (already forgot which model). The latter SSD is running ZFS. I already established experimentally that it can push out roughly 1 Gbps serving static files.
Big tech is all about hyperscale. This is hyposcale. Let's see what it can do.
@matt and it's running NVDA.zip :)
@TheQuinbox Well yeah, that puts very little load on it.