I read long ago you had to get malware on the air gapped machine first to begin with, and then it’s only accessible within a few meters. Also it can’t be accessed through walls. That was years ago though, maybe it’s changed now.

If it’s the same then after installing docker, creating a vaultwarden user, adding said user to docker group, and creating your vaultwarden directories, all that’s left is to curl the install script and answer the questions it asks.

I use bitwarden and the setup was fairly standard with the helper script. I use my own isolated proxy for all my services so that was already built. I haven’t used vaultwarden but if anyone that has used both can tell me the differences I could maybe help out.

I would say that is not the best way to keep/restore backups as you are missing the integrity checking features of a true backup system. But honestly what really matters is how important the data is to you.

I did something similar when migrating to 8. Consumer SSDs suck with proxmox so I bought 2 enterprise SSDs on Ebay before the migration and decided to do everything at once. I didn’t have all the moving parts you did though. If you have an issue, you will more than likely not be able to pop back in the old SSDs and expect everything to work as normal. I’m not sure what you’re using to create backups but if you’re not already I would recommend PBS. This way if there is an issue, restoring your VMs is trivial. As long as that PBS is up and running correctly (makes sure to restore a backup before making any changes to make sure it works as intended) it should be ok. I have 2 PBS’s. One on and off site.

PBS will keep the correct IPs of your VMs so reconnecting NFS shares shouldn’t be an issue either.

I’ve ran jitsi for 4 years now. You can keep your personal variables in an environment file that doesn’t really change and pull down a new compose file whenever you want to update. Ever since the switch to docker from native install it has made things much easier to maintain. I’m using a lxc with debian 12. 4 cores and 4gb ram. The only reason I’ve allocated that many resources is because we use it to record a podcast with anywhere from 4 to 10 people on the server at a time. As far as bitrate, resolution, etc, that’s all handled within your env file. You’d have to look at the docs to see what’s available for you to choose from.

Before you buy anything, put some of the same content that buffers on a USB stick or powered drive and play it directly from the pi4. Also connect via ethernet to your router from another PC and check your dl speed from the NFS share.

Check out !sbcs@lemux.minnix.dev

CPU is only one factor regarding specs, a small one at that. What kind of t/s performance are you getting with a standard 13B model?

What are your laptop specs?

Ollama without a GPU is pretty useless unless you’re using with Apple silicon. I’d just get rid of it until you get a GPU.

Consequences will never be the same

I guess I don’t understand. You followed the docker installation directions correctly and it didn’t work or you modified the directions in a way that you prefer and it didn’t work?

I have it installed for a few years now. I started with the AIO but moved to the separate container install after AIO was deprecated. I imagine the install process is too complex for portainer. https://docs.funkwhale.audio/stable/administrator/installation/docker.html

I did steps 1-4 and skipped the rest because I already have a proxy server running. Don’t remember anything related to snapd though. Mine is running in a Debian 11 VM on proxmox instead of an LXC, but the process should be the same. Also they have a matrix channel for help https://matrix.to/#/#funkwhale-support:matrix.org

From what I remember it was relatively painless to install, but upgrading can be a chore, especially this last upgrade. My main interest in FW was the federation aspect as far as finding new music. If you don’t care about federation, maybe a simpler option would work better for you.

At the very least you need to install a webserver and you need a proxy of some kind. If you truly want old school you can just create html pages hosted from the root of your webserver (although there are now easier modern ways to do this, you might learn more the classic way rather than using a CMS).

You will want a reverse proxy to lie between your webserver and the internet that handles SSL. Let’s Encrypt is a good option to generate a cert so that you only expose port 443 on your router to the internet and your webserver. You’ll have to open port 80 to generate the cert but can close it again once generated. Then you will have https.

That’s the basics. The how-to’s are easy to find online.

I’m not sure how soon you need this, but if you can wait sipeed has a $20 kvm with ATX control that should be out soon https://lunar.computer/news/sipeed-announces-new-20-risc-v-kvm-device/

There’s an interesting book I read recently related to this called The Anxious Generation: how the Great Rewiring of Childhood is Causing an Epidemic of Mental Illness. I’d recommend it.

As a counterpoint, EFF put out this article today: The Surgeon General’s Fear-Mongering, Unconstitutional Effort to Label Social Media

This is the part of the new compose file that I don’t understand:

PICTRS__OLD_REPO__PATH=/mnt/sled-repo PICTRS__REPO__TYPE=postgres PICTRS__REPO__URL=postgres://user:password@host:5432/db

When I try to add it, pictrs complains about not being able to connect to postgres, even though the credentials are correct and I’ve already run the postgres migration script. Did you not do this part when you upgraded?

Not every room or space will be hosted by someone self-hosting their server. I find it kind of appalling that this would be the solution. It’s certainly not what I’ve heard from people working on projects around moderation.

This is somewhat the goal, but without the hyperbole. Homeservers will be hosted by individuals or orgs, but will contain many rooms and spaces each. In 2020 there were over 20,000 homeservers. That number has without a doubt grown exponentially. The concept is the same with Mastodon. If you are worried about moderation on the matrix.org homeserver there are many lists of public homeservers across the web. Many people research homeservers to see if they are a good fit for them just as they do Mastodon instances.

This blog post gives a good idea of where matrix is heading. Notice their mission of decentralization.

Since agglomeration around a single instance is against the goals of Matrix and its Foundation, users need to have a way out and incentives to move.

We are committed both to making Matrix more accessible, and to doing the work to decenter the Matrix.org homeserver.

Matrix.org is meant to be an entry point, not a stopping point. As to your concern regarding built-in tools, matrix is just a spec, an open source federated communication protocol based on HTTP. The community builds tools. Matrix is not discord. In fact it is in opposition to the discord philosophy of centralization, data mining, advertising, and AI training.

It sounds like you are more concerned with the matrix.org homeserver than matrix itself. Matrix.org homeserver will eventually go away for personal use, this is the plan for the future. Matrix has always kept this homeserver open as proof of concept, but has not planned to keep it open forever as the goal is the widespread adoption of the protocol and for people and orgs to host their own servers and build tools using matrix.

The bullet points you listed are all currently able to be realized on any self-hosted homeserver.