The app crashes on an unhandled rejection
An async route throws, nothing catches it, and the whole Node process exits. In older Node versions it only warned; now it kills the server.
Every async route caught
Working with Express?
Most Express codebases that reach us were started quickly, grown by several hands, and then left alone for a few Node releases. We read the middleware chain carefully, write down what it actually does, and move it forward without rewriting what did not need to change.
What it is
Express is the minimalist Node.js web framework created by TJ Holowaychuk in 2010. For years it was the default way to build a Node API: routing, middleware chain, template rendering helpers, almost no magic.
Now under OpenJS Foundation stewardship. Express 4.x was the status quo from 2014 until Express 5.0 released in October 2024 with async-friendly defaults and modernised internals. Still the most-downloaded web framework on npm.
The honest version
No technology is all upside. This is where Express earns its place and where it tends to cause trouble.
Sound familiar?
The Express problems that bring people to us, and what you have once each one is fixed.
An async route throws, nothing catches it, and the whole Node process exits. In older Node versions it only warned; now it kills the server.
Every async route caught
A long-lived service on a Node version the registry no longer supports, with middleware that has not been updated in years and a package-lock full of advisories.
Upgraded to Express 5 on current Node
An Express codebase handed over with no README, no tests, and routes defined in half a dozen different styles. Nobody is sure which endpoints are still used.
Documented and back under control
A middleware that calls res.send and forgets to return, so the handler runs afterwards and tries to send again. Looks fine in testing, throws under real traffic.
One response per request, always
The default session store keeps data in process memory, so a restart logs every user out and a second instance never sees the first one's sessions.
Sessions in Redis or Postgres
CORS configured for one origin and then copied between projects, so the new front-end domain is quietly rejected by the preflight.
CORS set per environment
A process that never shuts down cleanly, so the next deploy finds port 3000 still held by a zombie and refuses to start.
Graceful shutdown wired in
Dozens of transitive packages flagged, some abandoned upstream. Upgrading one breaks another, so nobody has done any of them.
Dependencies current and audited
What we do
Call us for one broken route, or hand us the whole service.
How it works
Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.
We fix the cause, not the symptom. If something is down, getting it back up comes first.
Backups, monitoring, updates and access, so the same problem can't quietly come back.
You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.
Often found together
We work on the whole stack, so the problem can be wherever it actually is.
Yes. We run the upgrade on a branch, sweep the middleware list for anything that no longer works with Express 5, migrate async error handling, update the Node version alongside it, and only merge when the test suite and a staging run look clean.
Routinely. We read what is there, write down what each route actually does, get the project building on current Node, add the tests that were missing, and hand back a codebase that another developer can pick up without a briefing.
An async route handler threw and nothing caught it. Node now exits the process on unhandled rejections instead of just warning. The fix is either wrapping every handler, adopting express-async-errors, or moving to Express 5 where async errors flow into the error middleware on their own.
Often not. A well-kept Express app is a fine thing to run. If the service has grown beyond what the framework helps with, we move it piece by piece rather than in one go, so you keep shipping while the rewrite happens.
It depends on who the users are. For first-party users a session cookie backed by a real store is usually right. For third-party clients or mobile apps, signed tokens through a vetted library. We pick one, put it behind a small module, and keep it out of every route handler.
Estimates are free. Describe what is wrong or what you want upgraded and you get a price before any work begins.
Free estimate
Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.
Your message is with a real person, and we'll get back to you soon.
Rather talk now? Call us: