If you already have a public facing server for them to connect to then sure.

Set up Tailscale and an SSH key for remote tech support

I have 113k images going back two decades. The screenshot above doesn’t include RAW files, with those included I’m around 2 terabytes of total storage.

• Immich is in fact a photo album, and a damn good one at that.
• Immich keeps google’s grubby paws off my photos. I don’t need or want anyone datamining every precious memory I have in order to modify my behavior to their benefit.
• Immich shares photos between my wife and my phones.
• Immich ensures that if I lose my phone, my photos aren’t lost.
• Immich lets me easily re edit and re-export RAW files without creating duplicates or losing metadata
• Immich lets me conveniently share photos with friends and family without requiring them to have an account anywhere.

Mostly I self-host things when I want data synchronized between multiple devices, or I don’t want to lose it in the event I lose the device it was created on.

Also, like, phone screens are tiny and typing on them is terrible? Why would you want to do everything on your phone?

If encryption is enabled, don’t worry about it. Otherwise ‘dd if=/dev/zero of=definitely-the-bad-drive-do-not-fuck-this-up bs=4M’

Agreed, that should be many tens of pages not one. Also the mobile layout isn’t very good. I think it’s important to remember that normies use their phones for almost everything.

Got a 3 year old kid with another on the way. I just need it to be reliable so the kid can watch Sesame Street and the lights keep working.

I definitely see your point, but the difference is that it’s one thing to learn. Once you know docker, you can deploy and manage anything.

Cons of containers are slightly worse disk and memory consumption.

Pros:

• ease of installation
• declarative, consistent configuration
• security some degree of sandboxing
• dependency management is solved

Stick with the containers

Oh nice! I’m stoked about the background sidecar operations, that will speed things up considerably

but there is a reason i just explained it to you

Ok but is there room for the idea that your intuitions are incorrect? Plenty of things in the world are counter-intuitive. ‘docker-compose up -d’ works the same whether it’s one container or fifty.

Computer resources are measured in bits and clock cycles, not the number of containers and volumes. It’s entirely possible (even likely) that an all-in-one container will be more resource-heavy than the same services split across multiple containers. Logging from an all-in-one will be a jumbled mess, troubleshooting issues or making changes will be annoying, it’s worse in every way except the length of output from ‘docker ps’

I can see why editing config files is annoying, but why exactly are two services and volumes in a docker-compose file any more difficult to manage than one?

I disagree with pretty much all of this, you are trading maintainability and security for easy setup. Providing a docker-compose file accomplishes the same thing without the sacrifice

• separate volumes for configuration, data, and cache because I might want to put them in different places and use different backup strategies. Config and db on SSD, large data on spinning rust, for example.
• separate container for the database because the official database images are guaranteed to be better maintained than whatever every random project includes in their image
• separate networks because putting your reverse proxy on a different network from your database is just prudent

If you need honest-to-god office, then yeah you’ll need a windows installation. Either a VM or a second drive is best.

You can use windows indefinitely without activating, you’ll just have the watermark and default desktop background.

Seems fine, but you’re sorta hitting two fields at once. Application development (coding) is a different skill set from devops/deployment (docker). I’d stay pretty surface level on docker and the CLI for now and focus on building your app. You’ll know when you need to go off and learn those things.

Your best option by far is to overwrite windows completely. For most software development Linux is way better anyway.

I haven’t done this recently enough to guide you on the details, but step zero is to decide whether you are certain you want to dual boot or not. It adds a lot of complexity and brittleness that is best avoided if at all possible.

• Try to find Linux compatible replacements for the software you need.
• if that doesn’t exist, see if you can run it on Linux with wine.
• If that isn’t possible, consider running windows inside a virtual machine on Linux.
• If you do want honest, bare-metal windows then using two different physical drives will be easier and more reliable. Ideally your laptop has room for two drives, otherwise you can dangle a USB SSD (not a flash drive). Windows won’t install to a USB drive but Linux doesn’t care.

Yeah, I tried it but that experience isn’t as good as a native app. No swipe gestures, and an extremely basic UI

Miniflux has served me very well for years, combined with a few different apps. Reeder on iOS, I can’t remember what I used on android but there were plenty of options

Software & Services:

• Restic client
• Restic REST server
• https://github.com/rbuchberger/res-man + systemd timers or cron to configure & run restic nightly
• healthchecks.io for monitoring
• ntfy.sh for notifications

Destinations:

• Local raspberry pi with external hdd, running restic REST server
• RAID 1 NAS at parents’ house, connected via tailscale, also running restic REST

I’ve been meaning to set up a drive rotation for the local backup so I always have one offline in case of ransomware, but I haven’t gotten to it.

Edit: For the backup set I back up pretty much everything. I’m not paying per gig, though.