Only when you have to write out to swap. In the case of something like ZFS, it stores data in RAM, looks for it there, then looks on the disk. So freeing up the RAM is effectively instantaneous; you just mark the space as free, then the other process writes into it.

It’s part of ZFS 2.3.0, so it just depends what version TrueNAS is shipping with.

ZFS seems pretty RAM hungry

This is a common misunderstanding.

Short version; ZFS isn’t RAM hungry, it’s RAM aware. If your system has unused memory lying around, ZFS will use it to improve read performance. But it will give up that memory the moment anything else needs it.

and I don’t believe you can add new drives to an existing volume.

No longer true

Xpipe is fantastic. I have to manage a LOT of SSH connections between work and homelab (well over 200 now) and Xpipe has been a god send.

This is the only correct answer.

At work we use Meshcentral. It requires you to host your own server, but it’s very powerful, and very reliable. We’re managing something like 400 remote systems with it currently. We also use Netbird as a secondary access layer (I prefer it to Tailscale for the simplicity of setting up ACLs, and the really easy deployment).

For most home server usage though, I wouldn’t bother with Meshcentral. It’s a lot of overhead if you’re only managing a couple of systems. If you really need remote desktop (why do your servers even have desktops?) use RustDesk instead.

RustDesk really is fantastic. No shade to any of the other solutions suggested in this thread, but 99% of the time when someone needs remote desktop access, RustDesk is exactly what they need.

My bad. Thanks for the correction.

He’s She’s talking specifically about the idea of embedding AI agents in operating systems, and allowing them to interact with the OS on the user’s behalf.

So if you think about something like Signal, the point is that as it leaves your device the message is encrypted, and only gets decrypted when it arrives on the device of the intended recipient. This should shut down most “Man in the middle” type of attacks. It’s like writing your letters in code so that if the FBI opens them, they can’t read any of it.

But when you add an AI agent in the OS, that’s like dictating your letter to an FBI agent, and then encrypting it. Kind of makes the encryption part pointless.

Even if you use AI tools for drafting, you’ll want to know enough CSS and HTML to be able to parse the code and make adjustments where needed. Being completely ignorant of how to read the output from AI coding tools is never a good idea.

On their ARM platform you get something like 8 cores and 24GB of RAM. Honestly, that’ll run a hell of a lot more than an RSS server.

I have one that’s running three different minecraft servers simultaneously.

You can also set these limits in your compose file, if you use compose (which you should).

Most of those containers are probably grabbing more memory than they actually need. Consider applying some resource constraints to some of them.

Dozzle is an excellent addition to your docker setup, giving you live performance graphs for all your containers. It can help a lot with fine tuning your setup.

Do they have to edit your domain into a connections file or some such

Pretty much. I don’t remember the exact details, but that’s close enough for the cat.

No, as in the person installing the app to use the service has to edit a config file.

Yes, I have no issue editing config files. I’m self-hosting, that’s the point. All the technical load should be on me. But my completely non-technical friends should not have to edit config files to be able to access my self-hosted services. Everything, for them, should be as simple as possible.

It can be self hosted, but to connect the clients to self-hosted servers you have to edit config files, so it’s a very user hostile solution.

Oh, perfect. Thank you.

Honestly, I think this is just one where you try it for yourself. The compose file is about 4 lines long, I had the whole thing up and running in about 30 seconds (OK, 45; I forgot a port was already in use and had to redeploy).

So far my one big complaint would be that the self-hosted version replicates the entire website, including all of the “Why choose Bento PDF” and “Try now” and so on. It’d be nice to just have the tools right there when I load it up. Other than that, well, it looks cool, I’ll know more once I actually try out the available options.

So, they had multiplayer, and it worked very well, but then they went through a whole bunch of major reworks to underlying systems that broke multiplayer, and they basically went “Yeah, that’s gonna stay broken for a while until we get all this shit done, please be patient.”

Project Zomboid is a blast, especially when you really dig into the options for changing game rules. You can basically craft your own custom zombie apocalypse. You can decide how the virus works, whether zombies are slow or fast, whether they have good eyesight, good hearing, how strong they are, where they spawn. You can change loot rarities, how long it’s been since the outbreak started, when the power gets shut off, etc, etc.