The Most LOVED Server On The Planet

​Caddy is sitting at nearly 70k GitHub stars for a reason: after 15 years of being told Nginx is the only "professional" choice, I spent 30 days trying to break Caddy just to see if the hype was real.

Many of us developers treat web servers like a pet: we feed it certificates, groom its config files, and stay up late when it gets sick.
I lived that life.
Back in August 2016, I woke up at 5 AM to my pager app screaming.
Production was down, and my phone was melting with messages.
The reason?
A wildcard SSL certificate had expired.
One. One manual, avoidable task paralyzed our entire infrastructure.
If you've noticed why all Fox News articles comments not only were missing but slowed down their system as well that month... well, it was kinda my bad 🙀
​

Caddy, isn't a pet, it’s a service dog.
Actually, a robot service dog (don't think about AI.don't think about AI)...
It stays out of the way and handles the hard stuff like SSL without you ever asking.
​

The King has complicated rules

For a decade, the standard solution has been Nginx or if you're more "modern" -> Traefik.
The latter, up until last month my go-to for anything, from side projects to K8s in production, is not to be disrespected, but it can take a page from this book...
We’ve been taught that if you want "production grade," you need complex configurations and external tools like Certbot to hack together HTTPS.
Most people solve their server needs by manually managing certificate renewals or paying for "enterprise" features that should be standard.

If you want to get Caddy running the right way, use Mise like I do, or your favorite package manager (their list is quite extensive even in my standards).

Once it's in, you can see the sheer breadth of what it handles: from reverse proxying to authority management, just by checking the help menu:

To start a server:

If you have a Caddyfile and want to see how the server translates it into that complex JSON while adapting it in realtime:

And to push changes without killing the process:

​
Why your infra is a high maintenance

I ran stress tests comparing Caddy to Traefik; over 30 different permutations.

Even with 4 gigs of RAM, Traefik kept eating memory like a cookie monster and spiking under load, while Caddy stayed relatively lean and stable.
This was mostly noticeable on very little resources (which suggests Caddy is the better option when you're constrained), but also, when the system was extremely stressed.

The zero config server

It turns that you don't need horrendous config structures to run a multi-node, multi-endpoint reverse proxy.

A Caddyfile can be as easy as

Want a file server with a UI to browse your assets?

But here is the trick that blew my mind: serving localhost on HTTPS (yes forreal).
Usually, that's a manual certificate nightmare.
With Caddy, you just put this in your Caddyfile:

Caddy will literally ask for your password, install a root cert, and boom 💥 localhost on HTTPS.
No more insecure flags needed in dev.

If you’re using it as a proxy (which most of us are), it’s just one command:

In a Caddyfile, it’s even cleaner:

You can even run multiple sites in one block without the Nginx-style clutter:

For the Kubernetes crowd, Caddy is already ahead. It doesn't have a flashy UI, but it ships with a metrics endpoint for Prometheus:

And if you need to extend it, like using my favorite DNS provider, DNSimple, to handle DNS challenges: you use xcaddy to build a custom binary with the modules baked in:

Is it the undisputed heavyweight king?
Maybe not for a massive, thousand node Kubernetes cluster ingress just yet.
But for everything else? local dev, side projects, and high-performance gateways --> it has officially dethroned Traefik as far as I'm concerned.
It’s zero config, it’s faster, and it actually lets me sleep through the night.

I hope this was valuable! Thank you for reading.

Feel free to reply directly with any question or feedback.

Have a great weekend!