Openwisp is an orchestration platform, but it is very overwhelming to the home/homelab user and not suitable for someone expecting the Unifi “single pane of glass”. It works best when most devices are the same model, otherwise you’re just making templates for many diferent device types.

I only run two instances, both run nginx and static HTML sites (plus all the stupid mandatory bits like fail2ban, python for ansible, certbot, etc. They are very low usage and get no seo or anything so they are really, really low usage.

I’ve never been warned about resources so far, and it’s been 3 years. I intensionally don’t run any high-bandwidth stuff like a matrix server or file sync for that reason.

I just lock it right down with keys and firewall entries for SSH. Logs are pretty quiet, except for llm scraping, but they are rate-limited, so they go away quickly.

Be aware that Oracle presents image “shapes” as the os images for use,which include oracle, Ubuntu, and a few others. These do have oracle metrics gathering and agents installed to help with migration between data centre zones, so it’s conceivable that they can read what’s on the os. I don’t have any PII on there except public keys and my email address.

The install script ceased being supported 3 years ago on Alpine. I think it was over handling of non-systemctl systems, but I can’t remember.

I used to run pihole on alpine containers and one say the install just wouldn’t update the FTL server. I switched to another DNS solution at that point.

Best suggestion is to run the docker container, as another user suggested here.

Just to add to your comment:

As much as I hate oracle, I run their free-tier vps in a Canadian datacenter and it never required my cc. I think it’s geographic location-dependent.

If the caldav and cardav specs were more practical and less insane, we would probably have a lot more choice in software to these ends.

Yeah, I also just hooked up tbird to my radicale instance. It’s a bit overkill, but it does work to edit calendar items.

You may need to go catch up on this. The “dev” in this case caused more issues than they solved.

One can’t be missed if one didn’t contribute to anything in the first place.

Be aware that the Signal Foundation still runs servers for the signal service. If a state actor compromises them, e2ee is no longer guaranteed.

Anubis is an elegant solution to the ai bot scraper issue, I just wish the solution to everything wasn’t just spending compute everywhere. In a world where we need to rethink our energy consumption and generation, even on clients, this is a stupid use of computing power.

You are mistaken. Frigate ingests rtsp and direct streams using HLS, which only accepts h264 or h265. The vast majority of cameras encode in h264. H264 is a trivial decode operation on modern hardware within the last decade.

I am currently looking at CPU usage in my frigate container with 5 1080p RTSP streams and it is hovering at between 4% and 8%. Without any quicksync configured, just CPU and coral.

I don’t know what your deal is, or why you think you know more than folks who have been doing this for years, but stop, please. You don’t know what you’re talking about.

The quicksync built-in to the n100 PR n305 is more than adequate to encode/decode media. If that’s all you’re using it for, you don’t need an Intel arc.