Speculating is great for troubleshooting. Every time someone speculates a possible cause, it’s possible to devise a way to test it. It’s called hypothesising. Each tested hypothesis, regardless of the actual results, helps to further the understanding of the problem.
I’ve been using glauth + Authelia for a couple years with no issues and almost zero maintenance.
Yes, absolutely. Ideally there would be an automated check that runs periodically and alerts if things don’t work as expected.
Monitoring if the backup task succeeded is important but that’s tue easy part of ensuring it works.
A backup is only working if it can be restored. If you don’t test that you can restore it in case of disaster, you don’t really know if it’s working.
Ah got it. I didn’t know there was a free tier!
How do you use ChatGPT anonymously? It requires a valid login linked to a payment method. It doesn’t get any less anonymous than that.
I don’t mind the order of path, arguments and options, but what the hell is the deal with long arguments with a single dash? i.e. -name instead of —-name
I don’t think I’ve ever come across a DNS provider that blocks wildcards.
I’ve been using wildcard DNS and certificates to accompany them both at home and professional in large scale services (think hundreds to thousands of applications) for many years without an issue.
The problem described in that forum is real (and in fact is pretty much how the recent attack on Fritz!Box users works) but in practice I’ve never seen it being an issue in a service VM or container. A very easy way to avoid it completely is to just not declare your host domain the same as the one in DNS.
If they’re all resolving to the same IP and using a reverse proxy for name-based routing, there’s no need for multiple A records. A single wildcard should suffice.
Not sure if this is helpful in any way, but it might give you some clue.
100./8 addresses are reserved for CG-NAT.
This is probably the IPv4 address your modem/router is receiving from the ISP.
I might pick it back up some day but at the moment I have other projects going on at the moment.
I’m still using Proxmox myself but unfortunately it’s all fairly manually configured.
I started writing a Terraform provider for Proxmox a while ago.
Unfortunately, the API is a massive mess and the documentation is not very helpful either. It was a nightmare and I eventually gave up.
K3s is k8s
lol at the downvote. K3s is k8s. The very first 2 words in its website are Lightweight Kubernetes. https://k3s-io.github.io/
On macOS I’ve been using Ollama. It’s very easy to setup, can run as a service and expose an API.
You can talk to it directly from the CLI (ollama run ) or via applications and plugins (like https://continue.dev ) that consume the API.
It can run on Linux but I haven’t personally tried it.
It’s no surprise Apple uses CUPS. They wrote it, after all.
Edit: TIL Apple didn’t write CUPS themselves but they bought the company that did it pretty early in the game. Here’s a LWN article from the time, exposing some of the worries that came with the news of the acquisition: https://lwn.net/Articles/242020/
It’s pretty easy with Ollama. Install it, then ollama run mistral-7b (or another model, there’s a few available ootb). https://ollama.ai/
Another option is Llamafile. https://github.com/Mozilla-Ocho/llamafile
since they only distribute it via .deb
.deb are the Debian package format. Ubuntu is actually a Debian derivative, among others, which is why they use the same format. Debian lists a few of those derivatives in their docs: https://www.debian.org/derivatives/
Here’s my Debian setup for gaming: https://lemmy.world/post/9543661
I run it on my router which has the CG-NAT IP address.
Whilst you’re right that it could clash, it’s very unlikely (a 1 in 4194302 chance), I imagine Tailscale would detect the clash and change IPs though I could be wrong as it never happened to me (and probably never will - though in all fairness it will eventually happen to someone).
Been using Tailscale behind CG-NAT for years. It works wonderfully and very rarely needs to route through the DERP infrastructure - it’s almost always a P2P connection.
Ah NFS… It’s so good when it works! When it doesn’t though, figuring out why is like trying to navigate someone else’s house in pitch dark.