so it’s an electron web app you could instead of in-browser web ui?

looks very nice, thanks. would appreciate better documenting of SMTP options (login & password) and support for Authelia.

thanks, I guess I missed it. gonna update ASAP just in case, even though I’m the only user of my instance.

unfortunately there’s no images for 0.18.2-anything yet :(

you’re looking for zfs set sync=disabled which would bring performance back to acceptable levels. the alternative is using enterprise SSDs.

I’m still not sure why would anyone expect privacy on a distributed public forum and how would this even work.

3 days

very nice, updated with zero issues.

nothing to suggest but i’m absolutely fascinated by your post, op. please post results.

if you pass all outbound connections through vpn you’ll be fine.

i’m hosting lemmy on a vds using an S3 storage hosted at home (over wireguard). :)

i’ve decided against hosting lemmy at home because there’s no way to setup a proxy for outbound connections (or i couldn’t find it) so any federation request your lemmy instance makes reveals your home IP to owners of that particular instance.

re: container logs issue, the proper fix is setting this once in daemon.json (which is what you did) and fixing this issue node-wide instead of adding custom logging into compose files, unless the situation requires it (we had an application which had to use gelf for, oh god) which isn’t the case here. i just have an ansible role to deploy docker which configures logging, among other things.

agree that multiple networks may have a use but definitely not in this particular case.

I always see docker compose files like this as a starting point

same here, unfortunately not everyone has necessary expertise to do that, default setup linked on the project page shouldn’t be this half-baked.

also, unless they fixed it already, 3 days ago compose file and nginx.conf had mismatching ports so it would just never work ootb.

every time i see this compose file my head hurts

• so much debugging enabled
• two networks for whatever reason
• bespoke logging setup instead of just using system default
• host mounted directories instead of docker volumes
• postgresql exposed for whatever reason

it’s a development setup. devs really need to put a production enduser version of this out there.

replace the whole build: yaml block with image: dessalines/lemmy:0.17.1

also bump both versions from 0.17.1 to 0.17.4

100% this, it’s kind of weird that it’s not.

unless you depend on a third party, this is a nontrivial problem.

recent events with beehaw have shown that being on a large instance does not guarantee federation.

i’m all for roaming profiles even though i think the best option is hosting a personal or a friends-only instance - unless you’re a colossal asshole nobody will defederate you, you’re are not depending on anyone but yourself for your profile and subscriptions, etc etc. this obviously isn’t for everyone but the barrier of entry is sufficiently low for people interested in tech and fediverse to do just that.

email has started as a fully distributed system but - for reasons too many to count here - ended up centralized over several huge providers, openly dictating rules to everyone else. i’d rather fediverse not followed this road.

you’re right, i completely forgot that it has this strange local DB. that’s disappointing.

personally i would just run it outside of k8s then because it’s not stateless and pvc is a PITA.

i suggest pointing pictrs to (an external) S3 backend instead of using a pvc in your README example. minio is much easier to deploy than ceph.

i’d rather have a mesh of small-ish instances instead of several huge ones, that aside the more people the better.

as far as i remember, processes run inside container as root (0:0) end up under your own UID.

everything else gets mapped to those weird UIDs.