Struggling to connect to your ngrok tunnel from your phone or another device? The problem is almost always your local server binding to localhost (127.0.0.1), and here are the real-world ways to fix it.

Why Your ngrok Tunnel Won’t Connect on Mobile (And How to Actually Fix It)

I’ve been there. It’s 1 AM, you’re trying to show a product manager a hotfix for the mobile web view on their actual iPhone before a major release. You spin up your local dev server, run ngrok, send them the shiny `.ngrok.io` URL… and they get a “This site can’t be reached” error. You refresh it on your own machine and it works perfectly. It’s one of those infuriatingly simple problems that can cost you an hour of frantic searching while someone important is waiting. It’s not you, and it’s not ngrok—it’s a fundamental networking detail that bites every developer at least once.

The “Why”: Understanding The Localhost Trap

Let’s get this straight. When you run a typical command like ngrok http 3000, ngrok is doing its job perfectly. It’s exposing port 3000 on your machine to the public internet. The problem isn’t ngrok; it’s your local development server.

Most web frameworks, by default and for security reasons, start their server listening on localhost, which is an alias for the IP address 127.0.0.1. This special address is a “loopback” interface. It means “this machine, and only this machine.”

So, here’s the flow of failure:

1. Your phone sends a request to your-tunnel.ngrok.io.
2. Ngrok’s cloud service receives it and forwards it through the tunnel to the ngrok agent running on your computer.
3. The ngrok agent then tries to pass that request to your local server at 127.0.0.1:3000.
4. Your server, which is only listening for connections originating from itself, sees a connection coming from the ngrok agent (a separate process) and often rejects it. This results in a “Failed to complete tunnel connection” or a browser timeout.

The solution is to make your local server listen for connections from more than just itself.

Solution 1: The Quick Fix (The `–host-header` Flag)

This is the solution you’ll find most often, and it works in a pinch. You can tell the ngrok client to rewrite the request’s host header before forwarding it to your local server. This essentially tricks your server into thinking the request came from localhost.

Instead of your usual command, you run this:

ngrok http --host-header=rewrite 3000

This is a great “get me out of this jam right now” fix. However, I don’t love it because it masks the underlying issue. It’s a patch, not a proper solution.

Solution 2: The Permanent Fix (Binding to 0.0.0.0)

This is the right way to solve this problem. You configure your local development server to listen on 0.0.0.0 instead of 127.0.0.1. The IP address 0.0.0.0 is a special address that means “listen on all available network interfaces.” This allows your server to accept connections from your machine itself, from other devices on your local network, and crucially, from the ngrok agent.

Here’s how you typically do it for common frameworks:

app.listen(3000, '0.0.0.0');

app.run(host='0.0.0.0', port=5000)

rails server -b 0.0.0.0

HOST=0.0.0.0

Once you’ve made this change, you can go back to running ngrok normally, and it will just work:

ngrok http 3000

Heads Up: A Quick Word on Security
Binding to 0.0.0.0 makes your dev server accessible to any device on the same network (like your office Wi-Fi). This is generally fine for development, but it’s something to be aware of. Never run a production server bound to 0.0.0.0 unless it’s properly firewalled.

Solution 3: The ‘Nuclear’ Option (Specify The Network IP)

Let’s say you’re working with some compiled binary or a weird proprietary server and you can’t change the host it binds to. Or maybe you’re just not sure how. There’s a brute-force method that almost always works.

First, find your computer’s local network IP address. On macOS/Linux, run ifconfig | grep "inet ". On Windows, run ipconfig. You’re looking for something like 192.168.1.15 or 10.0.0.42.

Once you have that IP, you can tell ngrok to forward traffic directly to that IP address and port, completely bypassing the localhost name resolution issue.

Let’s say your IP is 192.168.1.15 and your server is on port 8080:

ngrok http 192.168.1.15:8080

This explicitly tells ngrok, “When you get a request, forward it to the server running at this specific network address.” This is less elegant because your IP can change, but it’s a powerful diagnostic tool and a surefire way to connect when nothing else seems to work.

So next time you’re stuck, don’t just add the --host-header flag and move on. Take a minute to understand the “why” and fix it properly by binding your server correctly. Your future self will thank you.

Darian Vance

Lead Cloud Architect & DevOps Strategist

With over 12 years in system architecture and automation, Darian specializes in simplifying complex cloud infrastructures. An advocate for open-source solutions, he founded TechResolve to provide engineers with actionable, battle-tested troubleshooting guides and robust software alternatives.

🤖 Frequently Asked Questions