Anyone with experience using NixOS to create a Lemmy instance, please share any caveats and troubleshooting tips you have.

  • neoney@lemmy.neoney.dev
    link
    fedilink
    English
    arrow-up
    4
    ·
    edit-2
    1 year ago

    Oh hello, my instance is on NixOS!
    There’s a NixOS module for lemmy, but it’s kind of broken right now and in the middle of a PR. If you want, you can replace the current nix package with the one from the fork, and successfully deploy it, however, if I were you, I’d just wait a week or so.

    The configuration right now requires a bunch of hacks, and even doing some non-reproducible stuff.

    • ericjmorey@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 year ago

      That PR link returns a 404.

      This is my way of learning Nix/NixOS so I’d rather give it a go before the PR is merged.

          • neoney@lemmy.neoney.dev
            link
            fedilink
            English
            arrow-up
            2
            ·
            edit-2
            1 year ago

            Okay, I’ve actually figured simplifying it out, it’s not that bad. Let me share my config:

            First, since the PR isn’t in nixos-unstable yet, I’m adding the fork as a flake input

            inputs = {
              nixpkgs-lemmy.url = "github:CobaltCause/nixpkgs/lemmy-module-improvements";
            };
            

            then, in my system configuration, I add this:

            # Not sure if this is required, maybe caddy auto-allows it
            networking.firewall.interfaces.eth0.allowedTCPPorts = [443 80];
            
            # Override the lemmy module with the one from the PR
            disabledModules = ["services/web-apps/lemmy.nix"];
            imports = [
              "${inputs.nixpkgs-lemmy}/nixos/modules/services/web-apps/lemmy.nix"
            ];
            
            services.lemmy = {
              database.createLocally = true;
              database.uri = "postgres:///lemmy?host=/run/postgresql&user=lemmy";
              enable = true;
              settings = {
                hostname = "<YOUR_HOSTNAME>";
              };
              caddy.enable = true;
            };
            

            and, that’s it!
            However, I’m not sure if it will cleanly deploy, as you might get an error while starting.
            If so, please check postgresql logs sudo journalctl -fu postgresql. The error will most likely be something like this:
            […] [10289] ERROR: permission denied: "RI_ConstraintTrigger_a_16639" is a system trigger […]

            If that happens, you need to manually run the migration until the fix is merged into Lemmy. Here’s how I did it:

            1. sudo su - postgres
            2. psql -v ON_ERROR_STOP=1 lemmy postgres
            3. (in psql) SET ROLE lemmy;
            4. Paste the SQL code from here: https://github.com/LemmyNet/lemmy/issues/2784#issuecomment-1578337686

            After that’s done, you can exit the postgres CLI by typing \q, exit the postgres user and just simply sudo systemctl restart lemmy which should start properly now, and be accessible to the outside network.
            Open it and it will give you the initial setup screen. Good luck!

            • neoney@lemmy.neoney.dev
              link
              fedilink
              English
              arrow-up
              2
              ·
              1 year ago

              Note that the PR got merged into nixpkgs now, but hasn’t made it’s way forward (see https://nixpk.gs/pr-tracker.html?pr=236295), so that’s why you need to do the override for now Also, a fix for the issue with the migrations (if still occuring) has been merged yesterday, but it’s not in any release yet, and especially not in nixpkgs. You could most likely get around it by using overrideAttrs to change the source to make it build from a newer commit.

              • ericjmorey@lemmy.worldOP
                link
                fedilink
                English
                arrow-up
                1
                ·
                1 year ago

                Is this sort of lag the nature of package managers that will inherently be part of NixOS and nixpk?

                • neoney@lemmy.neoney.dev
                  link
                  fedilink
                  English
                  arrow-up
                  2
                  ·
                  1 year ago

                  Nixpkgs is by nature fully reproducible and declarative - that means you can’t write a nix package to just build from master branch. You also have to specify the sha256 hash to absolutely make sure the source is the same for everyone.
                  With flakes, it’s a bit easier, because if the package you’re trying to build has a flake, you can just update it in your own lockfile and it will be rebuilt from the latest version.
                  If using something from nixpkgs, you have to wait (or PR yourself) for someone to PR the update, or you can also use overrideAttrs to do a “private” fix.
                  For example, I’m using the wlsunset program, but wanted to build from a different branch, so I did this:

                        (wlsunset.overrideAttrs (old: {
                          src = fetchFromSourcehut {
                            owner = "~kennylevinsen";
                            repo = old.pname;
                            rev = "81cfb0b4f8e44db9e5ecb36222a24d53a953e6aa";
                            sha256 = "sha256-Lxuhsk4/5EHuKPkBbaRtCCQ/LFvIxyc+VQYEoaVT484=";
                          };
                        }))
                  

                  This just changes the source attribute to another commit and another sha hash, and it works fine.

                • neoney@lemmy.neoney.dev
                  link
                  fedilink
                  English
                  arrow-up
                  1
                  ·
                  1 year ago

                  While we were talking I updated lemmy-server using overrideAttrs for my own use. It’s honestly not that hard. You change the commit, specify some random incorrect hashes, build it, nix screams at you that the hash is incorrect (and generously provides the correct one), put the correct hash, and build again. Done.

  • yiliu@informis.land
    link
    fedilink
    English
    arrow-up
    2
    ·
    1 year ago

    FWIW, I’m running NixOS but gave up on running the Lemmy module. I gave up when I realized that Lemmy seems to need superuser access to the Postgresql server, to install plugins or whatever.

    So instead, I used Arion to make a docker-compose image, running in podman. Works great so far.

  • kylian0087@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    arrow-down
    1
    ·
    1 year ago

    I have not read the full docs on setting up lemmy yet. But i don’t think it matters what distro you use as long as the distro can run docker. So setup docker and from their it is the same for any distro.

      • kylian0087@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        1
        ·
        1 year ago

        That is cool. I did not know this. Still i think docker would be the most universal and simple way.

        • Something Burger 🍔@beehaw.org
          link
          fedilink
          English
          arrow-up
          3
          ·
          1 year ago

          If someone is using NixOS, they would prefer native support. It’s easier to set up than Docker and fully integrated with the rest of their system’s configuration. Though NixOS also supports declaring OCI containers (with Docker or Podman) if needed.

          • psyix@infosec.pub
            link
            fedilink
            English
            arrow-up
            1
            ·
            1 year ago

            do you have working config or flake available? im on latest unstable channel and it builds and installs fine but when trying to login or setup admin account i receive an error

            syntaxerror: unexpected end of json input

            which when traced in browser console it seems to do with captcha endpoint downgrade through reverse proxy. i use traefik so will have to investigate further