I’d go for Alpine Linux in such case.

You may want to check out numen, targeted towards power-users.

Yep, people are enthusiastic about self hosting and like talking about what they host :)

When I am sending? Well, once things are set up properly I’m pretty confident that things arrive (though nobody can ever be 100% sure of course). I also tend to mail to the same recipient domains a lot, like for work and hobby projects, so once those are tested you get pretty confident.

Unnoticed downtime is usually quickly noticed, I depend on my server for a lot of things. Senders are often resilient enough to keep things in their queue and try a few times. There’s also a fallback MX registry at my (3rd party) DNS host which will queue stuff in case the primary MX goes down.

Interesting hardware list, that indeed is a bit more complicated (and probably more expensive) than most are running.

Nice, RSS is great indeed. I use it extensively as well, but I didn’t even realize it was a thing people ran as a service on a server. I hadn’t heard of FreshRSS etc. I personally just run newsboat from my desktop/laptop, even my phone if need be.

I’ve been self-hosting e-mail for over 15 years and hope to continue doing so. Although it’s being made increasingly difficult by big tech players. I wrote about it here: https://proycon.anaproy.nl/posts/rant-against-centralising-e-mail/

To answer my own question:

• E-mail (postfix, dovecot, rspamd, clamav)
• Web (nginx), various small websites including my homepage
• Fediverse Microblogging (Mastodon)
• Matrix Chat (synapse)
• XMPP Chat (prosody)
• Music streaming (mpd, snapcast)
• Home automation (home assistant and my own lighthome stuff, mqtt)
• IRC bouncer (znc)

And the basics of course:

• SSH (openssh)
• NFS

All running on an Ubuntu Linux server, but everything is containerised into mostly Alpine Linux podman (rootless) containers (and a few lxc containers which I’m phasing out).

Nice, you must be into deep learning with such a setup, any particular reason the deep deep learning models and GPU run in your server rather than in a powerful desktop system? Maybe you’re actively offering AI services to the outside world?