Two pitfalls I had that you can avoid:

• look at efficiency. It’s not always neglible, was like 40% of my energy usage because I oversized the UPS. The efficiency is calculated from top power the UPS can supply. 96% efficient 3kW UPS eats 4% of 3kW, 120 watts, even if the load you connected is much smaller than 3kW
• look at noise level. Mine was loud almost like a rack server, because of all the fans.

I replaced that noisy, power hungry beast with a small quiet 900W APC and I couldn’t be happier

No no, that’s how i’m working around the problem now, but i’m sure sni sniffing will sooner or later make my domain well known

I second the complaint about subpaths. I have all my services on a single domain, except for HA. It’s for security by obscurity, when you issue a certificate for a subdomain you start getting malicious traffic probing for vulnerabilities almost immediately. I don’t have this problems for services with non-obvious subpaths.

I can’t understand the stubbornness of developers to accept patches for fixing this problem.

Kodi is good for many streaming services too, just not Netflix. It has been good with HBO Max.

You did not say what kind of streaming services.

For anything self-hosted or torrents/debrid, just get a Raspberry Pi with LibreELEC.

If you use Netflix and the likes, you will likely want something officially supported. My partner likes Netflix for some reason and after years of using the unofficial addon by CastaginaIT, I gave up and got her a Firestick this winter (having set up a separate VLAN for it and ripped out the microphone, of course).

The unofficial Kodi addon is an amazing piece of reverse engineering work, but it’s not really great that you have to log in using your computer every month or two, and occasionally download a 2GB binary, before you can watch a movie on Netflix half-asleep.

I actually wrote to the author, not being from the US too, and they said they are working on a version where you can specify your country/location. Nothing about adding your news sources, though

No real reason. It’s stable so I keep using it. Look at the official list: https://jitsi.github.io/handbook/docs/community/community-instances/

Yeah, Windows and hosting are a tough match. But you have native Photoshop and Autocad.

I ran it via Docker, it was easy enough, consumed close to no resources, upgrading was tricky a few times. Then I got rid of that server and did not continue hosting Jitsi, started freeloading on Freifunk Munchen public instance.

There was some kernel command line argument, related to cgroups I think, that fixes the problem. But yeah, far from frictionless

The installer is much easier, and the github keys import is a nice touch. Last two servers I set up were Debian, because fuck forcing LXD as snap, and the installer was inferior to that of Ububtu took me about 3x the time to do the same setup.

I’ve got the cron job and used Postgres in my previous instance. Limiting fpm concurrency is good insight, Ihavee not done that.

But do you know what works without all that and loads instantly? Photoprism.

Lucky bastard!

It was around 10k-15k photos that the server stopped working well. I’m surprising syncing on Android worked well for you with 8k photos though, there are so many bug reports on Github with cases like mine where it just stops working with 2k photos. Not slow, but failing to sync at all.

Some people have luck like this and guess they’re the ones advocating. I first had a hand-installed instance with all the recommended optimizations like Redis, and then I started fresh with Docker + Redis. In both cases, after 10-15k files it was extremely slow. Both on server hardware with plenty of RAM.

Exactly, same here. Nextcloud is the only container for which I disabled auto upgrades, because I don’t always have 8 hours to deal with the aftermath.

How many photos do you store there that you recommend it? Like 500? With any serious number of photos Nextcloud starts loading for minutes and the android app autoupload goes tits up.

That bug is open for years and if with so many users affected and a commercial entity behind this project nobody has fixed it yet, I doubt it’s fixable without a very major rewrite.

Maybe OP is not doing this just as a hobby and has actual serious workloads?

I’ve had this kind of problem with Vultr, i was very pissed off when I found it but their support raised my limits when I explained what I’m migrating. I also had the comfort of being able to migrate in stages, 10 machines in the first month, then the rest. Maybe this appropach would work for you.

Credit card companies and Paypal are a big problem to hosting companies. They will happily apply chargeback after you provided a month of service to your client, because it took them a month to detect the transaction was fraudulent. How is it the hosting company’s problem?