I run a 2 node k3s cluster. There are a few small advantages over docker swarm, built-in network policies to lock down my VPN/Torrent pod being the main one.

Other than that writing kubernetes yaml files is a lot more verbose than docker-compose. Helm does make it bearable, though.

Due to real-life my migration to the cluster is real slow, but the goal is to move all my services over.

It’s not “better” than compose but I like it and it’s nice to have worked with it.

You’re asolutely right, IP addresses are kind of a grey area since the are needed for lot of troubleshooting and debugging.
Nevertheless, you can always strive to reduce the stored data.
For your application, you wouldn’t even need to store the historic IP adresses, just a rough geo-location and maybe a mobile/landline/whatever-flag and comparing the current login attempt to that. Even saves you some performance by not repeating the geo-lookups everytime.
Implement your failed-login counter separately by account and source IP and you’ve got decent security without linking an account to an IP.

Because collecting data that is not strictly necessary is almost always a bad move. IP addresses might be relatively harmless, but might link you to other activities.

You personally might be okay with reddit knowing your IP addresses, but some people might get into trouble.

Take the insane anti-abortion laws in some US states. If an IP address from those states accesses pro-choice subreddits, that might be enough for law enforcement to start harassing someone.