Hello all! Yesterday I started hosting forgejo, and in order to clone repos outside my home network through ssh://, I seem to need to open a port for it in my router. Is that safe to do? I can’t use a vpn because I am sharing this with a friend. Here’s a sample docker compose file:
version: "3"
networks:
forgejo:
external: false
services:
server:
image: codeberg.org/forgejo/forgejo:7
container_name: forgejo
environment:
- USER_UID=1000
- USER_GID=1000
- FORGEJO__database__DB_TYPE=postgres
- FORGEJO__database__HOST=db:5432
- FORGEJO__database__NAME=forgejo
- FORGEJO__database__USER=forgejo
- FORGEJO__database__PASSWD=forgejo
restart: always
networks:
- forgejo
volumes:
- ./forgejo:/data
- /etc/timezone:/etc/timezone:ro
- /etc/localtime:/etc/localtime:ro
ports:
- "3000:3000"
- "222:22" # <- port 222 is the one I'd open, in this case
depends_on:
- db
db:
image: postgres:14
restart: always
environment:
- POSTGRES_USER=forgejo
- POSTGRES_PASSWORD=forgejo
- POSTGRES_DB=forgejo
networks:
- forgejo
volumes:
- ./postgres:/var/lib/postgresql/data
And to clone I’d do
git clone ssh://git@<my router ip>:<the port I opened, in this case 222>/path/to/repo
Is that safe?
EDIT: Thank you for your answers. I have come to the conclusion that, regardless of whether it is safe, it doesn’t make sense to increase the attack surface when I can just use https and tokens, so that’s what I am going to do.
Opening ports on your router is never safe ! There’re alot of bots trying to bruteforce opening ports on the web (specially ssh port 22)
With SSH I would disable the password authentication a only used key based authentication. Also disable root access. (Don’t know how it works with forgero though)
I would recommend something like wireguard, you still need to open a port on your router, but as long as they don’t have your private key, they can’t bruteforce it. (You can even share the wireguard tunnel with your friend :))
Also use a reverse proxy with your docker containers.
There are a lot of things you could do to secure everything, but If you relatively new to selfhosting, there’s a steep learning curve and a lot of time needed to properly secure everthing up. You could be safe by doing nothing for a few months but as soon as someone got into your system, you’re fucked !
But don’t discourage yourself, selfhosting is fun !
This is both true and highly misleading. Paranoia isn’t a replacement for good security.
The same is true of ssh when using keys to authenticate.
You’re right, but only if you are an experienced IT guy in enteprise environnement. Most users (myself included) on Lemmy do not have the necessary skills/hardware to properly configure and protect their networking system, thats way I consider something like wireguard way more secure than opening an SSH port.
Sure SSH key based configuration is also doing a great job but there is way more error prone configuration with an SSH connection than a wireguard tunnel.
But it doesn’t help to just tell newbs that “THAT’S INSECURE” without providing context. It 1) reinforces the idea that security “is a thing” rather than “something you do” and 2) doesn’t give them any further reference for learning.
It’s why some people in this community think that putting a nginx proxy in front of their webapp somehow increases their security posture. Because you don’t have “direct access” to the webapp. It’s ridiculous.
In this case it’s handled by forgejo.
The reason why I am asking this question is because I think that the ssh port I am opening only has access to my repos (which means that even if I somehow get hacked the damage is minimal) and it doesn’t accept any keys aside from mine and my friend’s, which we set up through the web interface :).
I have wireguard setup and I’d also thought about sharing a tunnel with my friend, but it seemed much more hasslesome than simply opening the port, not to mention the fact that if anyone wanted to join too I’d have to do that again.
It is exactly because I am afraid of getting fucked that I am asking this and being careful. For now, my idea is to only open the port when someone is about to use it, since I am not absolutely sure that it won’t somehow accept a request from a person with less than noble intentions. (either that, either simply use tokens)
Reverse proxying was also my intention at first, but I just couldn’t get it to work with cloudflare for some reason!
Thanks for the insight!