Home

Hi,

I an currently trying to add remote access to 2 of my servers but didn’t manage to get a working setup as is.

Right now I want to access 2 servers: * one is for media stuff (navidrome, jellyfin, managing the arr stack) * one is for my data syncing with rsync and after set a backup from borg to another server not on my domain

I was trying at some point to add stuff such as tailscale, but somehow I always had issues with having both servers reachable within the IP range I use on my local network, so everything would work as is with the current config at home being away. I have also heard of cloudflare tunnels as well, but that I didn’t try yet. At some point I tried to do just a regular wireguard from my opnsense, but I would prefer not to have open ports to worry about (and also had issues with internal IP not being assigned from wireguard as well).

Does anyone here has experience with this? If so, what was your solution and/or caviats to it?

I was going to say Tailscale. You will need to give more details about what didn’t work with Tailscale, but it has been pretty seamless for me.

Basically when I connect to tailscale I just can’t get it to give an internal IP so I can access everything with my configs. Unless I am missing something obvious, I don’t understand what is going on here.

Sure, you’re not behind CGNAT?

When you connect to Tailscale, via cli it should give you a link to use to connect it to your account. You can get the ip address in various ways. One of which is just looking on your Tailscale dashboard. You need to have Tailscale installed on every machine you want to talk with.

I recently switched from tailscale to NetBird. Similar solution but FOSS and self-hostable.

Have you exposed the subnet the services are on, onto the Tailscale network?

I do not have Nat of any kind that I am aware

Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:

[Thread #127 for this comm, first seen 2nd Mar 2026, 10:00] [FAQ] [Full list] [Contact] [Source code]

Do you want to expose port ⁸⁰⁄₄₄₃ and set up a reverse proxy or do you want to use a VPN tunnel? You could just use SSH to port 80 and 443 like so: ssh -L 80:<local-server-ip>:80 -L 443:<local-server-ip>:443 <username>@<domain>

I expose port ⁸⁰⁄₄₄₃ and use Caddy as a reverse proxy together with Authelia to protect anything that I deem needs an extra layer of security. I followed this guide: https://caddy.community/t/securing-web-apps-with-caddy-and-authelia-in-docker-compose-an-opinionated-practical-and-minimal-production-ready-login-portal-guide/20465

Once setup, it is easy to remove or add a backend to Caddy and Authelia. This way does mean that you sometimes need to log in twice, but that is a small price to pay if your backend app does not support SSO (like n8n community edition).

I think I did set this option, but still no internal IP. Can try again later to be sure

I just enable SSH, configure it to run on some non-standard port and enable Fail2ban… Make sure if use a certificate or secure password and also check if fail2ban is actually doing its job. Never had any issues with that setup.

Have you tried adding Tor hidden services? It was the easiest solution for me to expose ports from behind the provider’s NAT to my phone when not at home.

If you have IPv4 addresses, I guarantee you’re behind at least one NAT gateway. What you need is a Tailscale subnet router, or something equivalent from another service.

In the most basic configuration, the Tailscale client facilitates communication between one host it is running on and another host it is running on that are both connected to the same tailnet. For this purpose, it uses addresses from the 100.64.0.0/10 “shared address space” subnet. These addresses will only be reachable from within your tailnet.

If you want an entire subnet (e.g. your LAN) to be accessible within your tailnet, you need to set up a subnet router. This involves configuring the Tailscale client on a device within the target subnet to advertise routes (tailscale set --advertise-routes=192.168.1.0/24), allowing the host to advertise routes in the admin page (Machines -> … -> Edit routes), and configuring the Tailscale client on external hosts to accept advertised routes (tailscale set --accept-routes).

If you want your servers to be accessible from anywhere on the internet, you’ll need Tailscale Funnel. I don’t use it personally, but it seems to work.

I’m behind CGNAT. My OpenWrt router is a Netbird server that can be connected externally. Having the Netbird server in the router allows me to ssh devices or use services as if I was connected via WiFi.

There’s documentation for Opnsense as well -(https://docs.opnsense.org/manual/how-tos/netbird.html)

I switched from tail scale to pangolin for reverse proxy. Does everything. Auth, VPN, hidden services, public services. Fantastic piece of software

I will check if this can work for me, but sounds like it is the kind of solution I am looking for

Ah great, this sounds like what I was missing with tailscale. With try once I can, thanks!

Never tried hidden services from tor. Can check how that works but not sure if it is the solution I am looking for. Thanks for the info anyways!

“how do I add remote access to my servers?”

don’t.

create a new server that’s accessible via VPN and then access your servers from there. then actively log all connections from that device and alert anytime someone or something connects to it.

what is more secure? a house with twenty front doors or a house with one front door and an alarm on it.

If you check my edit that is kind of what I was hoping to do from the start: have a hop server (or stepping stone, both terms apply), and from there I do what I need to do

The way Tailscale works, you don’t need to worry to much about your local IP address. You can just use the Tailscale IP address and it will connect as if you were local using the fastest route. That’s the beauty of a mesh VPN. Each device knows the fastest route to each other.

Without more information I can’t really tell what issue you are actually having, but if your system has internet, you have a local IP and if the system is showing as up on your tailscale dashboard than it will have a tailscale IP. Not being able to connect using one or the other would be a configuration issue. Whatever service you are having trouble with is probably only listening to one of the interfaces but not the other.

I’m assuming you are running a linux or unix box, but try running the command ip addr. Assuming you have the package installed, it will tell you all of your IP addresses for the system you run the command on. The list may be quite long if you have a lot of docker containers running. The command tailscale ip will do the same but limited to your tailscale IP addresses.

After everything is setup, create a network route to distribute an ip to machines connected to you lan. I can’t recall exactly but setting up Netbird was pretty straightforward when following the documentation. They also have their own for Opnsense - (https://docs.netbird.io/get-started/install/opnsense)

Managed Networks documentation

After everything is setup, create a network route to distribute an ip to machines connected to you lan. I can’t recall exactly but setting up Netbird was pretty straightforward when following the documentation. They also have their own for Opnsense - (https://docs.netbird.io/get-started/install/opnsense)

Managed Networks documentation

Just expose it on single-stack IPv6. Nobody ever knocks. The address space is not scannable.

How did you config tail scale though? Are you using some Apple or MS author account? I want to stay away with using one of their services to “authorize” connecting to my own server

How did you config tail scale though? Are you using some Apple or MS author account? I want to stay away with using one of their services to “authorize” connecting to my own server

For remote management, I just enable SSH, configure it to run on some non-standard port and enable Fail2ban… Make sure I use certificates or secure passwords and also check if fail2ban is actually doing its job. Never had any issues with that setup.

This is what I’ve done for years, but I sometimes feel like it’s not a great solution from a security standpoint.
Though I have switched from fail2ban to Crowdsec, which did end up banning my own connection attempts when I forgot to whitelist myself, so that seems secure enough.

Are you using some Apple or MS author account?

Google and Github SSO were the only options when I originally setup tailscale. There are a few more options now including what looks like every self-hosted OIDC provider I’ve ever heard of, and a few I hadn’t.

How did you config tail scale though?

There are a couple options depending on how you are using it. Most of the time I just use the tailscale command to configure each node.

Most systems were just sudo tailscale up --ssh to get it up and running, although I have one system setup as a subnet router to give me outside-the-house access to systems that I can’t put tailscale on. That was a little more involved but it was still pretty straightforward and well documented. Their documentation is actually very well written and is worth the read.

Give it a robot that can read your handwriting, & write snail-mail lettres to it?

d :

_ /\ _

Hmmh. I’m not entirely satisfied with any of them. Crowdsec is a bit too complex and involved for my taste. And oftentimes there’s no good application config floating around on the internet. Whereas fail2ban is old and eats up way too much resources for what it’s doing. And all of it is a bit too error-prone(?) As far as I remember I had several instances when I thought I had set it up correctly, but it didn’t match anything. Or it was looking for some logfile per default but my program wrote to the SystemD journal. So nowadays, I’ll double-check everything.

I think you’re overthinking it. Wireguard is considered the “gold standard” and an excellent solution for what you’re trying to do. Open ports can be a concern, but an open Wireguard port is completely silent when not in use and does not respond unless it receives the correct access keys. That makes it invisible to port scanners.

Wireguard on my OpenWRT router works flawlessly. If the router is working the WG endpoint is too, and there are no 3rd parties involved. Tailscale provides much the same thing, but as I understand it requires the involvement of multiple 3rd party services. I’ve been burned too many times by terms of service changes and security breaches so I wanted to avoid relying on any corporate entities wherever possible.

Tasker brings up the tunnel on my phone automatically whenever I’m not connected to my home wifi and drops it when I get back home, so my home servers are always available. My biggest problem when not at home is Verizon’s crappy mobile network.

IMO it’s worth the effort to properly configure Wireguard and get your servers. Once you get it set up you probably won’t have to touch it for years.

Did you consider Headscale? What made you chose NetBird?

Nope.

NetBird is European. The stack itself is FOSS and self-hostable instead of relying on third party projects, like Headscale. It has a reverse-proxy feature in beta that was also appealing.

NetBird also utilises Coturn for STUN and TURN, and I’ve other software that depends on Coturn, so that kind of went hand-in-hand.

I get what you’re saying, but how exactly the whole IP rotation is done in your case? How did you manage to have it accessible at all times even when your home IP changes? In my I actually have ipv6 which I am not sure if it does not make things more difficult

If the servers have public IPs and you want the minimum possible ports open, just SSH? With passwords disabled and large keys, it’s quite secure.

If that’s still not enough for you or you need a private gateway, then Wireguard. I can strongly recommend Tailscale - It’s really an orchestration layer on top of Wireguard. You can setup your own Derp relays and head scale if you are truly paranoid. But 99.9% you don’t need all that and Tailscale out of the box will work well.

Also Tailscale isn’t a single point of failure the way you’re imagining. It’s certainly possible for Tailscale’s servers to go down, but that won’t drop existing connections.

DDNS (Dynamic DNS), one 3rd party service I do use.

My network is reached by URL, not IP (although IP still works). When my IP changes the router updates the DDNS service in minutes. Lots of providers out there and it’s easy to switch if needed. I like DuckDNS. It’s free or you can choose to donate a bit to cover their expenses.

Can also check one more time wireguard directly. Thanks!

Noted, moved to the top of the which VPN-ish tool to use soon stack.

It’s nice, the quickstart script is super easy to use and gets you started… well quickly. I’m still figuring the reverse proxy bit out, but it fully replaced tailscale for me in about ~10 minutes?

As far as I know you need to authenticate through them.

Alternatively, you could setup your own vpn to do all this, but it is much more work and will likely have some cost.

This is essentially what I’m doing. I have a firewall DMZ that drops everything that goes to it by default, then from there it port forwards my publicly available services and then anything that isn’t public is through a wire guard network

For the reverse proxy you may find Traefik marvelous, in case if Nginx/HAproxy is too featureful.

All you need is Wireguard with IP forwarding allowed on the host, maybe some firewall rules if you have one. You configure your wire guard client to only route traffic for your network IPs. I leave my wire guard client connected 100% of the time.

The moment you get a TLS cert, it’ll show up in Certificate Transparency logs and apparently the attack bots scan that for targets.

Make it a subdomain on a wildcard cert if you’re concerned about that.

It uses Traefik by default, actually. I’m struggling to get the reverse proxy function to cooperate with me still hosting other things on the VPS. I use it not just as my Netbird coordinator, but also to host my Forge and site.

Oh! I see! Just in case, Traefik has great options for debugging, and logging (e.g. TRACE log level)! Not to mention custom plugins! There’s a great Discourse based forum, and there’s both IRC and Discord Communities, too!

This is the way. Quite secure and private. It is not complicated to set up, just have to get the keys and copy them in the right places (and protect the private keys) and do the forwarding to a VPN endpoint on your network.

Yup. It gets more involved once you start adding DNS and SSL. But if you’re ok typing IPs and you’re not opening your firewall to the public, it’s all you really need.

You should keep trying with tailscale, did you read the docs? (tailscale provides amazing documentation), you dont need to install the client on every device, for that use subnet routers, all is in the docs. Give it another try and post back what issues you are having.