The acl, norm, xattr, dnodesize options are all to make ZFS behave like a Linux filesystem and are pretty standard for using ZFS on Linux. They aren’t default because ZFS’ defaults are for Unix.

I do:

sudo zpool create \
   -o ashift=12 -O acltype=posixacl -O compression=lz4 \
   -O dnodesize=auto -O normalization=formD -O relatime=on \
   -O xattr=sa \
   mypool \
   raidz2 \
   wwn-0x5000cca284c06395 \
   wwn-0x5000cca295e115f3 \
   wwn-0x5000cca2a1ef9c90 \
   wwn-0x5000cca295c03910 \
   wwn-0x5000cca29dd216b0

I’m then going to optimize recordsize depending on the workload in datasets. E.g. Immich db might use 8K or 16K recordsize while the library dataset where the files are might be larger so that search is faster. Etc.

Secuarr?

Install Ollama on a machine with fast CPU or GPU and enough RAM. I currently use Qwen3 that takes 8GB RAM. Runs on an NVIDIA GPU. Running it on CPU is also fast enough. There’s a 4GB version which is also decent for device control. Add Ollama integration in Home Assistant. Connect it to the Ollama on the other machine. Add Ollama as conversation agent to the Home Assistant’s voice assistant. Expose HA devices to be controllable. That’s about it on high level.

HA with local LLM on Ollama. Can imtegrate the Android app as the default phone assistant. I don’t think it can use a wake word on the phone though. I invoke it by holding the power button, like a walkie.

Using Home Assistant with Qwen locally. It functions better than any version of Google Home I’ve had. Understands me without having to think about how I say things. Can ask it for one or multiple things at the same time. Can even make it so that it pretends to be Santa Claus while responding. My wife was ecstatic when she heard the Ho-ho-ho while asking to turn the coffee machine on on Christmas.

You gotta hook it to a local LLM. Then it’s boss.

I haven’t tried funnel but it works using an internal Talscale IP/host and port. E.g. http://the-immich-host:1234/ if the-immich-host is a Tailscale machine.

Works outside. I’m setting a standard DNS record on a standard DNS provider to an internal TS IP. The record works everywhere but the IP is only accessible when TS is on. Whether I’m on the local net or outside.

That’s one way to look at it. I used to look at paid VC-funded services like that. I no longer do as I’ve observed services I paid good money for get more expensive much faster than inflation and decrease in quality and features at the same time. It’s one reason I self-host many services I used to pay third parties for. I now look to alternatives from the get go and derisk existing dependencies. To be clear - profitability isn’t merely the only problem. The ownership and its profit growth strategy (and expectations) are. Those are not the same in a decades old ISP and a VC-funded startup.

Merely being profitable today isn’t a good predictor for stable prices and function over the long run for VC-funded services. I’m not planning to do major surgery to my setup every few years as yet another service shits the bed. The workstation/server where my self-hosted services run has last been reinstalled in 2014. Most of my config-as-code was written in 2019. I support a few families with this and I aim at maximum stability with minimal maintenance. So I use open source whenever I can and I often pay for development. I only integrated Tailscale in my setup because the clients are open source and because there’s an open source server option.

I’m not saying to people - don’t use Tailscale. In fact I often recommend it to new self-hosters. But I do that because there’s a way out. So here I’m reminding people who care about a way out to check if this feature is escapable. :D

• you can create memorable links instead of memorizing port numbers: jellyfin.foo-bar.ts.net

BTW, I’m doing something similar with standard DNS records that point to an internal Tailscale IP. I can go to https://immich.mydomain.com/ which only works if Tailscale is active. Let’s Encrypt works too. Obviously the setup isn’t automatic but it’s automateable for more adept self-hosters.

Can you share what components are you using for SSO, UI, etc.?

Ownership, size and profit growth strategy. My ISP is a massively profitable poorly regulated oligopoly. The deal there is clear - they’re already charging as much as the market can afford. They aren’t providing a free service today that they’ll have to monetize down the line to compensate for the time operating on VC funding. Tailscale, awesome as it is today, is in my view guaranteed to enshittify over time as they start getting pressed to grow profit. That’s not too much of a problem for me since the clients I use are open source and there’s an alternative open source server. If I used features unavailable in Headscale or were in over my head and unable to self-host Headscale, I might be in a bad time some time down the line.

You’re talking about Headscale right?

While this is great, especially for smaller self-hosters, as a setup gets more and more dependent on Tailscale, one should think about self-hosting Headscale, and therefore not being over-reliant on services not offered by it. I’m in that boat and I haven’t done the Headscale migration yet.

A Perplexity knockoff?

This is the kind of thing I’ve been waiting for. Looks brand new. Is there anything more mature or this is the first of its kind?

I use their WiFi access points. They’re great. That’s about it.

I just got burned by accidental latest tag on a pg container for Nextcloud. They moved some paths internally and it could no longer find the db.

The cells achieve an energy density of up to 175 Wh/kg

This is kinda shocking density for a first gen mass-priduced batteries.