For me it’s https://nginxproxymanager.com/ it’s just so easy to setup and use. One docker command and you’re up and running with a nice webinterface to manage access to your docker instances with ssl. I heard good things about Traefik too but I have no personal experience with that one. NPM does everything I need and if it ain’t broken… :)
I used Traefik on my Docker stack and it’s pretty neat, though it took some time for me to get my head around how to configure it correctly.
Yeah seems like I was lucky to find what I needed on the first try. A colleague of mine was using Traefik but switched to NPM because it’s so easy to use.
I second that. Amazing easy to use, configure, supports (LetsEncrypt) certificates via DNS-01 challenge and integrates with ease with most DNS providers.
Paired with authentication providers (keycloak, authelia, authentik), the “advanced” textbox lets you do forward proxying really easy, or customize your “basic proxy”.
I’m not sure how many of these features are present in Traefik, it would be really nice if any of you know if any of these are easily supported in it:
- Forward proxying
- Custom rewrites (nginx
internal;
rewrites) - Unattended DNS-01 support with ACME (LetsEncrypt)
I second NPM. As you mentioned it’s been very easy to use, but I also haven’t been trying to do anything complicated.
I’ve never used load balancing so perhaps Caddy or Traefik is easier to use than NPM in that regard, but I wouldn’t know.
Yes NPM is for basic reverse proxying, so one URL to one server. If you wanted to scale and load balance across multiple servers you’d need regular nginx with a text config file since you literally can’t configure a second or third server.
And I’d still find that easier than Traefik, but maybe that’s just because I’ve been using Apache2 and nginx for like a decade at this point so it’s what I know.
I’ve been using NPM for years… but since 2.10.3 broke SSL certificates and there’s been literally no interest from JC21 to fix the problem (there’s a PR ready to go) i’ve been forced to look elsewhere and have settled on caddy for now…
To be fair, the pull request was last week. It’s inconvenient but life/work balance.
Agreed but it’s more the worry that it’s been broken for over 3 weeks and the dev(s) seems to have no interest in resolving it… to me that is a bad sign of things to come and projects being abandoned.
I’m not challenging you, so please don’t take offence here but is the issue sincerely a ‘lack of interest’ or is it just that NPM is FOSS and the maintainer is bogged down with life? You could fork it and fix it.
It’s a very good question and of course… i could fork it and fix it using the PR… but then that would be it… I’m not experienced enough to even achieve that to be honest…
My issue I guess is not so much with the fact that there is a problem… it’s with the fact that i can’t afford for my homelab to be down because it’s never fixed or takes time to fix… i appreciate all of this is free… i think i may of even donated at some point because i was so thankful it existed… but now it’s such an integral part of my and my families life that i cannot have something in my stack that isn’t going to be fixed rapidly.
JC21 created an amazing product and if it’s fixed or V3 ever appears i’ll 100% check it out… but for now whilst it’s not as pretty… i have to fall back to caddy.
I used NPM for a very long time, but after I switched to podman, DNS name resolution for containers stopped working in NPM, they work fine in every other container. Switched to caddy and it’s okay, it only supports HTTP transports so I can’t use it as a gateway for my DoH/DoT server, but that’s not a huge deal. Once NPM works properly on podman I may switch back
Nginx, because it works well and most open-source projects provide good examples for it when setting up things.
Same for me. You need the read into the documentation a bit, but once you understand how it works its fairly easy.
I use NGiNX and have ever since I started. It just works and is easy to configure.
Same. I know it’s more work than caddy etc, but I’ve been doing it for eons now so it’s muscle memory at this point.
For me it’s traefik. It’s took me a while to get it working, but it’s actually really easy now. Setting up container access with labels is very convenient!
nginx. Traefik is near unusable if you ever need something that isn’t dockerized. Caddy seems neat, but I miss some options you get with nginx.
nginx is just… good in all aspects.
deleted by creator
Caddy for general reverse proxy stuff, works like magic and makes certs, routing, etc just work.
I also have a lot of my stuff subsequently reverse proxied behind Authentik for anything that shouldn’t be exposed to the public internet
I love that about Caddy as well, it just works!
Do you know of any tool that can help me look at overall traffic that goes through it?
Right now I am using Mullvad through gluetun to essentially route traffic to my services without opening ports on my router and I am just curious what sort of traffic is hitting my server seeing how (I hope) isolated my address seems to be (servicename.mydomain.tld:<random port recieved from mullvad port forwarding>)
I will soon migrate this reverse proxy setup to a VPS since Mullvad will be sunsetting their port forwarding feature soon but I am still in need of a tool that can show me what sort of traffic goes through Caddy. Something like countries, IPs and services that they are trying to access as well as the request types.
Do you know of any tool that can help me look at overall traffic that goes through it?
I haven’t looked in detail at the Monitoring Caddy documentation page and haven’t used this myself, but apparently it can be configured to emit a bunch of metrics in Prometheus format.
Something like countries, IPs and services that they are trying to access as well as the request types.
Oh, for that kind of thing you’d need to parse the log files instead. GoAccess maybe?
Apache.
I started my self hosted journey over a decade ago and from what I remember most of the guides were for apache so that’s what I learned. Over the years Ive added so much that to re-do everything would take down my stuff while I figure it out and I just haven’t found it worth it.
Although it’s harder to keep it up these days, even setting up my Lemmy instance was a pain because nobody has apache guides anymore so you have to figure it out yourself
If you published an Apache guide I’m sure it would be popular.
Fellow Apachian! It’s how I learned how to make a reverse proxy initially and just never saw the point in learning something else (though to be fair haven’t had to make a reverse proxy recently).
I’m so used to Nginx I have trouble caring enough to learn anything else. If I were to eventually learn another, Caddy looks like the most attractive option, but I’m super open to hearing from people who have used both Traefik and Caddy.
Traefik, because I can configure it with labels on my containers and don’t have to deal with the proxy config every time I add a new service.
Used nginx for years but it’s starting to show the signs of its age, same as Apache did a few years before that.
I use Caddy, but recently realising it’s not good enough. Dealing with any traffic that’s not HTTP/s puts you in a pickle.
One more vote for Caddy, everything just works, simple things are simple but you have a lot of flexibility for more complex situations.
deleted by creator
I made the switch from NGINX to Caddy. For me, configuring Caddy is much more simple than configuring NGINX. Also Caddy automatically obtains and renews SSL certificates.
So, Caddy’s simplicity is what won me over. I don’t care about speed since I’m the only user of my self-hosted services.
When Nginx (and I pushed it hard on others to use) to NPM to Traefik to Caddy. Caddy is just simple, easy setup for Certs, and integrates easily with Authelia.