On low end CPUs you can max out the CPU before maxing out network—if you want to get fancy, you can use rsync over an unencrypted remote shell like rsh, but I would only do this if the computers were directly connected to each other by one Ethernet cable.

If you’re running it via docker compose it’s trivial to upgrade, and there are no breaking changes. Pull, down, up, you’re done.

Frigate is pretty good, too. I’ve only been running it for a few months but I’m very happy with it.

Though, technically that leaves you more at risk of ransomeware or something that overwrites your data.

I rsync as well, but use snapshotting on the remote drives. So, a bad rsync would suck but shouldn’t really result in data loss. Ransomware on my local+remote server would of course be very bad…

I do something similar — I have a raspberry pi and a HD, with daily rsync and snapshots (monthly retained indefinitely, weekly retained for a month, daily retained for a week). It’s at family’s house, connected to my home via WireGuard via a VPS. Tailscale (or anything really) would also work here.

It’s a great setup! Just have some watchdog reboot if it can’t talk to home (a simple cronjob with ping -c1 home.lan || reboot or similar).

Even our “slow” 35Mbps upload speed is way more than enough for incremental rsyncs of my Immich library. The initial sync was done in person, though.

I got one from goHardDrive on eBay (link). It was cheap enough, looks flawless, and knock on wood has been working fine.

Googling around, the brand gets…mixed reviews. My use case is such that of this drive fails it’s not a big deal.

I’ve honestly never understood people who feel the need to “replace” Spotify. … Spotify has never made sense for my use-case.

I don’t know how to say this, but…you have extremely uncommon use-cases:

…during those times, my phone is either fully turned off (so I’ll use an MP3 player), or it’s in Airplane Mode.

Many people listen to music on stereos and don’t necessarily want a device plugged in, so

I just download the music I like to my device and listen to it via VLC.

either doesn’t work or is substantially less convenient than e.g. casting from a phone.

Not hating on your setup at all, but it’s very niche, in my experience.

https://en.m.wikipedia.org/wiki/ACID

Atomicity (something happens in its entirety or not at all), consistency (database is always in a valid state — if the database has constraints, they will always be honored), isolation (transactions don’t step on each other), durability (complete transaction is complete even if there’s a power failure).

Not a database expert, my parenthetical explanations may need work.

It’s interesting that, with Python, the reference implementation is the implementation — yeah there’s Jython but really, Python means both the language and a particular interpreter.

Many compiled languages aren’t this way at all — C compilers come from Intel, Microsoft, GNU, LLVM, among others. And even some scripting languages have this diversity — there are multiple JavaScript implementations, for example, and JS is…weird, yes, but afaik can be faster than Python in many cases.

I don’t know what my point is exactly, but Python a) is sloooow, and b) doesn’t really have competition of interpreters. Which is interesting, at least, to me.

Did the developer use any version control though? SCCS has been around since the early 70s, RCS and CVS since the 80s. The tools definitely existed.

Also, it was a single dev, which makes SCM significantly simpler!

ZigBee router thing:

I’ve been happy with the SMLIGHT SLZB-06M. You can easily flash firmware, and it has PoE which was important for me. I believe it also supports Thread, but I haven’t tried this yet (and I’m not sure if it supports it at the same time as Zigbee).

Zigbee smart plugs from Third Reality have been pretty solid in my experience, and they report power usage.

For circuit breaker level monitoring, I have an Emporia Vue2. I have it running esphome, completely local — unfortunately this requires some simple soldering and flashing, so it’s not turnkey. But it’s been rock solid ever since flashing it. (Process is well documented online.)

I’ve had decent luck with cheap wifi Matter bulbs, but provisioning them is finicky, and sometimes they just crap out and need to be power cycled; Zigbee bulbs (e.g., Ikea) have generally been reliable, though sometimes I’ve had difficulty pairing them initially. After power cycling a Matter WiFi bulb, it takes a while for it to respond to Home Assistant; Zigbee bulbs generally respond as soon as you power them on.

I have a wired smart light switch from TP-Link/Kasa (KS205), and it’s been completely hassle free (and totally local — Matter over wifi). The Kasa smart switch dongles I have work flawlessly but need proprietary pairing, and I’m afraid to update firmware in case they lose local support.

Good luck! Fun adventure :)

I think a lot of companies view their free plan as recruiting/advertising — if you use TailScale personally and have a great experience then you’ll bring in business by advocating for it at work.

Of course it could go either way, and I don’t rely on TailScale (it’s my “backup” VPN to my home network)… we’ll see, I guess.

…are Turing Complete, so what you can do with them is exactly equal.

But they’re only equal in the Turing complete sense, which (iirc) says nothing about performance or timing.

Hopefully you can publish in an open-access journal — if not it would be great if you could share an arXiv preprint :)

I hope I’m wrong! I’d definitely consider buying some — hopefully you can report back with results. If they’re slower than advertised but have the actual capacity that’d still be awesome!

This looks like it might be it:

https://www.tomshardware.com/pc-components/ssds/buyer-beware-fake-samsung-1080-pro-4tb-ssd-promising-unbelievable-158-gbs-speeds-for-dollar43-is-too-good-to-be-true

The drive doesn’t provide 4TB of storage either, considering the single NAND chip. That means if you were to attempt to write that much data to the SSD, at some point it would either fail or start overwriting existing data.

I’d say it gets a little different with command line utilities — maybe “utility” is the appropriate term here, but I’d call something like grep a program, not an application (again — “utility” also works).

To be sure, grep is extremely powerful, but its scope is limited.

Physics is like sex: sure, it may give some practical results, but that’s not why we do it.

— Richard P. Feynman

I think the same is true for a lot of folks and self hosting. Sure, having data in our own hands is great, and yes avoiding vendor lock-in is nice. But at the end of the day, it’s nice to have computers seem “fun” again.

At least, that’s my perspective.

Whatever you decide for your laptop, I’m a proponent of a barebones off-site setup if you’re trying for 3-2-1 backup or similar.

I use a raspberry pi 3 with a single HD (ZFS) retaining some number of daily/weekly/monthly snapshots. Daily rsync, everything over WireGuard+VPS (TailScale would work too).

Others mentioned virtualization — I have had issues with COW filesystems (btrfs), as COW does not always play nicely with VM drives (extreme fragmentation and very poor performance).