this post was submitted on 15 Jun 2023
12 points (100.0% liked)

unRAID

1119 readers
1 users here now

A community for unRAID users to discuss their projects.

founded 1 year ago
MODERATORS
[email protected]
12
Anybody self-hosting Lemmy? (self.unraid)
submitted 1 year ago by blurredweasel to c/[email protected]
14 comments fedilink hide all child comments
 

I am not sure how to self-host docker compose environments. Maybe somebody can help out.

I'm a software dev and not afraid of some sysadmin work.

Options I see:

  1. I see a docker-compose community app that lets you start setting things up, but haven't investigated very much.
  2. Spin up a linux vm, pretend its not unraid and follow instructions. This is heavier-weight so I want to investigate the docker approach at least some.
  3. Individual dockers, sorta extracting each item in the docker-compose file and running them one-by-one.

I already have nginx proxy manager setup with other services, so that should be fine to expose things to the wider internet when I get that far.

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 8 points 1 year ago (2 children)

We are :P

Tiff wrote a post here on self hosting it: https://reddthat.com/post/19103

Also nearly everyone is using docker for Lemmy and as long as unraid lets you run docker commands you should be fine.

Docker compose is (now) part of docker. It's a way to describe all the containers you want, and then docker goes and does it. You should 100% learn docker compose, and Lemmy already has a compose file in their repo. So you should be able to copy paste and say docker compose up -d

[–] [email protected] 1 points 1 year ago (2 children)

Hows the CPU+memory+network load running an instance your size?

[–] [email protected] 4 points 1 year ago

I write about that here: https://reddthat.com/comment/66766 and allude to our problems in the future funding post here: https://reddthat.com/post/25633

But it's only a 2 core, 4gb ram, and I think a 100mbit line, though I have yet to test the network throughput to its maximum.

We average about checks 15-25% CPU usage at the moment over the past 24 hours.

We are using Cloudflare Infront of our server to help with spikes, but honestly it doesn't do much except for serving the images.
This is due to the fact that lemmyui interacts via websockets and all federation are pretty much POSTs.

Lemmy itself is very CPU and memory efficient. The biggest problems all other instances are having are database related and the inefficient nature that some of the queries are.

In the long run, I would like to get a dedicated colo instance for the database server, which would give us enough growing room for up to 10k+ users even with the current database inefficiencies, but until we get about $200-250 per month in recurring monthly donations there will no way we can get to that.