AtomHeartFather

I’m just an old guy with a lot of opinions. I am a sysadmin by trade. I like Linux, cool gadgets, Sci-Fi, DC comics, bass guitar, prog rock/metal, and annoying my kids with dumb dad jokes.

  • 4 Posts
  • 17 Comments
Joined 1 year ago
cake
Cake day: June 10th, 2023

help-circle


  • I do appreciate your feedback, but I think at a minimum that anyone trying to run a Lemmy instance in Docker should know how to install docker and docker compose and how to run basic commands like docker compose up -d. There are many tutorials out there for doing just that and I’m not trying to reinvent the wheel. Once you have gotten that part done my document kicks in and picks up where the official documentation is currently lacking (in my opinion).

    I do explain a lot, but I did my best to explain it in terms that most anyone could understand.

    I will take your feedback to heart and maybe try to write a step by step tutorial for people who are completely new to Docker as well.


  • Thanks for your comment, but I don’t see much value in pulling a new copy of the docker-compose.yml from the Lemmy GitHub. The only things I would be updating when Lemmy updates is the tag/version. If they added new environment variables some time in the future I could certainly take a look at their updated compose file to see the changes but I wouldn’t want to pull it down and replace my custom compose.

    I specifically don’t care for their (Lemmy devs) choices for logging, docker networking, and the built in nginx, so removing and simplifying all that was my main goal. Everyone has their own way of doing things, and this is mine.

    I will probably take a look at your Traefik configs and add them as a separate document for those that don’t want to use NPM. My goal is to add a subsection for most of the current revproxy choices.


  • I do not recommend using Ansible. It adds additional requirements and complexities that are unnecessary. Ansible is a great tool for managing multiple servers and software installs, in my opinion it is not the right tool to install Lemmy on a single instance. My install instructions require only that you have docker and docker compose installed.

    That said, you could easily replace the docker-compose.yml that Ansible set up for you with the one I am providing. Just don’t run Ansible against your server again or it will wipe out your changes.





  • Community discovery that spans all federated instances should be one of the top things that development should be working on. And it should be integrated into Lemmy, not as a separate website people have to go to and search.

    Peoples are lazy. They don’t want to have to go to some separate website and then search for something. And lets not even get started on the difficulties of adding a remote community if your instance doesn’t know it exists, its wonky at best.

    If a user cant type “Stephen King community” in the search bar on their instance and then get results, they are either going to assume it doesn’t exist and give up OR they are going to be hitting that “Create Community” button.


  • Of the themed instances that exist now, I’d be willing to bet that in addition to their local communities they host that they also subscribe to other communities that aren’t strictly related to whatever theme they are going with.

    For example, I’m sure the Star Trek instance also subscribes to the lemmy@lemmy.ml community so the admin can stay abreast of Lemmy news. And probably also follows other technology related communities as well.

    I think most people would just want to gravitate to whatever they want to be identified with. There’s nothing stopping you from joining a music themed instance and then adding some non-music subscriptions to your list. It doesn’t force those subs on anyone else on the instance.

    And if you don’t want to be identified with any specific topic or community, you can always join a general Lemmy instance like Beehaw or Lemmy.world and subscribe to whatever you like piecemeal.



  • Honestly, I hope not.

    For example, if all the “programming” communities ended up on a single instance, that is still a single point of failure. I think it would be better if they were spread out a bit. That way if the programming themed instance went down unexpectedly it wouldn’t take ALL the programming communities out with it, only the ones it hosts.

    There’s nothing stopping anyone from creating a programming themed instance and then subscribing to various programming communities on other instances and then creating their own local communities to fill in the gaps. And ideally, I think that’s what should happen.




  • Hot take. I think the instances that are trying to be Reddit are the ones that give their users carte blanche to create new communities without any thought of looking to see if the same community exists elsewhere. I’d prefer that community creation be limited to the admins of each instance, that way they could - hopefully - at least do a cursory search to see if the community exists already and then just add it to THEIR instances subscriptions. There’s a reason why every community shouldn’t be on a single instance. It’s a single point of failure.