moving from Linode to Contabo because my shitty personal projects do not need expensive uptime guarantees.

Why would you offer DNS in the year of our Lord 2025 without any kind of API?

I mean, I can fall back to my registrar's services, I just don't understand why.

Why would you take the trouble of configuring it using an API if you only have to do it once every few years or so?

Heh, uptime, everyone guarantees 99% right? Still means it may be down for three days a year.. I do not think that makes it more expensive.

EDIT: HOLY FUCK THOSE FUCKERS ARE CHEAPIII. DAAAAAAAMN, thinking about migrating on instant almost. DAMN.

@retoor I think Linode offers five 9s, Contabo doesn't have an SLA for private users but they''re in the order of two 9s based on past performance.

I need an API for ACME challenges because I use a wildcard TLS cert for simplicity.

@lorentz they received their first complaint for me already. They don't take my Revolut VISA credit card and told them that I was very happy, but now sad. Asked them to make me happy again. It's in their hands now.

@retoor strange, they took my Wise Visa card, so clearly they don't have a problem with Visa or fintech pseudo-banks.

@lorentz everything about you is Wise. This is the first time my card got refused so i think it's a hickup, but still I've shared my feelings towards them. I was a centimeter away to include a picture of myself sad looking in the camera.

@lorentz they send me the advise to use a nazi browser called FireFox and that helped indeed. I'm poorer now. For so far, everything works great.

@retoor :D that is fucking hilarious, I like them more for targeting firefox by default.

@lorentz I just moved a 3.3Gb app to a different data center with NO DOWN TIME without current users using the app noticing it. HOW DOES SHE DO THAT? WRauwr

@retoor snek?

@lorentz yeah, snek is used a lot. It has 3.3Gb of data. After the move: all bots kept working like nothing happened (the ones made by me and @BordedDev). Enterprise shit!

Can your rocket chat do that? I don't freaking think so.

But thanks for telling me about the hosting service. WIth my new storage, i can focus on the drive functionality for snek.

With your snek account you can login under ssh soon and store data.

@lorentz those servers are not hetero, very heavy oversold is my estimation. That's how they are so cheap. I know how those businesses work, I've worked for such. Same for Strato for example (but they're ten times worse). Removing server again. Couldn't have a stable persistent ssh connection from their server to my molodetz server / performance hick-ups.

Sad, too good to be true :(

Also see their disease of an web-interface from the 90's after you login :P Together with their inability to accept a credit card during chrome, i can't take this hosting service serious at all :P

Your rant is valid is f! :) Good host for not many uptime I guess :P

@retoor well, cheap hosting is cheap. But you wouldn't design a networked app these days that self-destructs if a single TCP socket momentarily disconnects, so this is only really a problem with the SSH protocol specifically. Actually, this might be a good opportunity to find a robust alternative to SSH that actually offers the resiliency expected from an application layer protocol to the user and not whatever the transport layer feels like doing at the moment.

@lorentz completely. False. I do it this way for years without issues and ssh can take a hit actually. Ssh is the protocol capable for this.

@retoor SSH sessions end when the TCP socket breaks, do they not?

Tying sessions to TCP sockets is not taking a hit, that's breaking on literally the first hit. It's like resigning from your job because construction work prevents you from taking your usual route, and then re-applying if you find a new route. The only applications that do this are ancient programs from a different internet like SSH and shitty indie games.

And the only SSH module where this behaviour is actually justified is tunneling since this way the socket breakage gets transparently propagated to sockets within the tunnel. SFTP should be a sessionless request/response protocol that allows the underlying socket to timeout to save resources and transparently reconnects for each operation if needed, and shell access should ping by default and auto-reconnect with exponential backoff.

GNU Screen solves a problem that only exists because SSH is an underdesigned protocol that conflates transport layer connections with user sessions. I need to repeat that no modern program does this on purpose, not even developer-oriented ones, not even security critical ones.

@lorentz tell me an real world alterntative. Servers get synced with SSH rsync. It's professional stuff.

Doubt last statement. Also by using -f -N you can just run ssh in background and logout without using screen.