Running ngrok tunnels?

ngrok tunnel dropping or edges misrouting? We bring it back.

ngrok started life as a developer tool for exposing a local port and has grown into a full ingress platform with edges, reserved domains, OAuth, mTLS and traffic policy. That shift means a lot of teams are using it for production webhooks, demos, and B2B integrations without really planning for what happens when a tunnel drops or a policy change silently breaks auth.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

ngrok.log

still flaky

What it is

ngrok at a glance

ngrok is an ingress-as-a-service platform. A small agent process establishes an outbound tunnel to ngrok's edge, and inbound HTTPS, TCP or TLS traffic on a public hostname is routed back down that tunnel to a local service. You get a public URL without touching firewalls, DNS or load balancers.

The platform has grown well beyond the free single-tunnel sweet spot. It now covers reserved domains, traffic policy rules, OAuth and SAML enforcement, IP restrictions, mTLS, and stable regional edges. That makes it useful not just for webhook development but also for production ingress to on-prem or edge devices that cannot easily expose a public endpoint on their own.

Type
Ingress platform and secure tunnelling service
First released
2013 open-source agent
License
Proprietary cloud service, open-source agent
Maintained by
ngrok Inc.
Typically used for
Webhook development, demos, IoT ingress, B2B integrations, remote access to on-prem services

The honest version

ngrok pros and cons

No technology is all upside. This is where ngrok earns its place and where it tends to cause trouble.

Where ngrok is strong

  • Webhook development stops being painfulPointing Stripe, Shopify or Slack at a local service during development is a one-liner rather than a half day of ngrok-adjacent plumbing.
  • Stable hostnames for stable integrationsReserved domains mean a webhook consumer does not have to be updated every time the dev agent restarts, so production integrations survive a laptop reboot.
  • Policy lives at the edgeOAuth, SAML, IP allow lists and rate limits are enforced in the ngrok edge rather than inside your app, which keeps those checks out of application code that often has no business doing them.
  • Agent is small and portableThe ngrok binary runs on Linux, macOS, Windows, Raspberry Pi and container bases, which makes it practical for IoT, demos and short-lived environments.
  • Observability on inbound trafficRequest inspection, replay, and traffic policy dry-runs let you see exactly what a webhook sender is sending, which saves a lot of guesswork.

Where ngrok bites

  • You are routing production traffic through a third partyEvery request to your service lands at ngrok's edge first. For many teams that is fine, for others it is a compliance conversation that has to happen before the tool is adopted.
  • Quotas bite at surprising momentsConnection-per-minute and bandwidth limits on lower tiers produce errors like ERR_NGROK_3004 under modest real-world traffic, and the fix is usually a plan upgrade rather than a config change.
  • Error codes are opaque at firstERR_NGROK_108 and ERR_NGROK_6024 look similar but mean completely different things. New teams spend a lot of time cross-referencing codes against the docs.
  • Traffic policy is powerful and unforgivingRules evaluate in order, and a slightly wrong action can silently block every request without an obvious failure in the dashboard.
  • Cost model is per edge and per featureReserved domains, static IPs, edges and seats are all separate line items, and the bill for a small team with several production tunnels is no longer the hobby-tier number.
  • Latency floor is set by the nearest edgeEven a direct connection is going through a POP, which adds milliseconds that matter for latency-sensitive services and are not under your control.

Sound familiar?

Common ngrok problems we fix

The ngrok problems that bring people to us, and what you have once each one is fixed.

ERR_NGROK_108

Only one tunnel runs at a time

Multiple agents trying to connect on an account limited to a single concurrent session, usually because a stale ngrok process is still running on a forgotten machine.

Session cleaned up, tunnels stable

ERR_NGROK_3200 tunnel not found

Reserved domain points nowhere

A reserved hostname that is not currently claimed by a running agent, either because the agent is down or because the hostname in the tunnel config was mis-typed.

Reserved domain bound correctly

ERR_NGROK_8012 connections refused by local server

The tunnel is up but the app is not

ngrok connects fine, but the upstream port is not listening, often because the dev server crashed or the agent is pointed at the wrong port after a config change.

Agent and app aligned

traffic policy blocked request

A policy rule locks everyone out

A newly added require-oauth or ip-restriction rule misordered against a bypass rule, so legitimate traffic is denied before the allow rule evaluates.

Policy order corrected

ERR_NGROK_3004 40 connections per minute

Webhook sender hits the rate limit

A chatty webhook provider bursting more connections than the current plan allows, surfacing as intermittent 429-shaped failures that are actually ngrok tier limits.

Plan or policy sized to real load

ERR_NGROK_15002 TLS handshake failure

mTLS edge rejects the client

A client certificate that was renewed on one side and not published on the other, or an edge expecting a CA that was rotated without updating the trust bundle.

mTLS trust aligned

monthly bandwidth exceeded plan limit

Bandwidth limit overshoots the plan

A video or file transfer through the tunnel burning quota far faster than webhook traffic, pushing the account over its monthly cap mid-cycle.

Traffic classified and plan sized

What we do

ngrok help, from first tunnel to production ingress

Call us for one dropping tunnel, or hand us the whole ingress setup.

Rescue
A production tunnel flapping, a reserved domain pointing at nothing, or a traffic policy locking everyone out. We stabilise the edge and then explain what tripped it.
Production tunnels
Reserved domains, static IPs, redundant agents and health checks, so an ngrok-backed integration does not go down when a single process restarts.
Traffic policy
OAuth, SAML, IP allow lists and rate limits implemented at the edge with a test plan, not inferred from documentation in a hurry.
Webhook development workflow
Stripe, Shopify, GitHub and Slack webhooks wired to local services with reliable hostnames, replay and signature verification.
Cost and plan review
Bandwidth, edge count and feature usage reviewed against the tier you are paying for, with a plain-English recommendation.
Managed Tech
Agent updates, policy reviews, quota alerting and an engineer available when a tunnel misbehaves outside office hours.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running ngrok alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

ngrok questions

Something else? Call+1 737 727 7327

Is ngrok safe to use in production?

For the right workloads, yes. B2B webhooks, admin tools, IoT ingress and remote support flows are common production uses. For public consumer traffic at scale it is less typical. The real questions are compliance posture, acceptable extra latency, and whether a tunnel-based ingress is a dependency your business is comfortable having.

Why does our tunnel keep dropping?

The usual suspects are an aggressive local firewall killing the long-lived outbound connection, a plan that limits concurrent sessions, or a flaky residential network. Running the agent on a small stable host with a reserved domain and clear logs usually turns a flaky integration into a boring one.

Can we use ngrok with a custom domain?

Yes. Paid tiers let you point your own domain at an ngrok edge with a CNAME, and reserved domains hold the mapping even when the agent restarts. Setting that up properly, including the TLS certificate side, is one of the most common bits of help we are asked for.

What is traffic policy and do we need it?

Traffic policy is ngrok's rule engine for inbound and outbound requests at the edge: auth enforcement, IP restrictions, rate limits, transformations. Not every account needs it, but if your integration demands OAuth, SAML or hard rate limits before traffic touches your app, it is cleaner than writing it yourself.

Should we use ngrok or Cloudflare Tunnel?

Both are serious options. ngrok has richer traffic policy and request inspection. Cloudflare Tunnel integrates tightly with the rest of Cloudflare's security and performance products, and the bandwidth story is often better for high-volume traffic. The choice usually comes down to the rest of your stack rather than the tunnel features themselves.

How much does ngrok help cost?

First estimates are complimentary. Describe what you are tunnelling, how much traffic it sees and what is going wrong, and you get a quote before any agent configuration is touched.

Free estimate

ngrok giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.