Serving a website on a Raspberry Pi Zero running in RAM | Hacker News Reader

Serving a website on a Raspberry Pi Zero running in RAM

btxx.org

253 pointsxngbuilds99 comments

This feels a little weird because while they are running the website itself (HTTP) off the Pi, they are handing off all TLS to a cloud provider.

So while the content is in RAM on the Pi, a lot of the heavier lifting (TLS termination) is done elsewhere, which saves a ton of CPU load on the Pi.

Sometimes these demos enable caching on the reverse proxy. So then for these tiny demo html pages you request, you may not even reach the fun tiny computer it is supposed to demonstrate.

Yeah, I've seen this in more than a few places. There was a blog "running on a Wii" that, IIRC, was doing the same thing.

On the one hand I get it, TLS is pretty heavy, and it makes sense to take advantage of a VPS or Cloudflare or however you want to do it.

But once you are spinning up a VPS, the question is ... why the Pi? The VPS in the article has less RAM, but more storage. If you're already doing TLS termination on the VPS (the most RAM intensive part), you might as well just do the whole shebang there.

I know this is all for fun, I'm just wondering -- is the Pi Zero really too slow to handle TLS, especially with an optimized TLS library? In this setup, the Pi is already being directly exposed to the Internet anyway, there's no VPN being used. That ARM11 isn't "fast", but surely a 1 GHz ARM11 can handle an optimized TLS library serving some subset of TLS1.2.

The TLS termination isn't actually on the VPS. The article details that Tierhive has an haproxy edge service (handling the TLS), that then has the vps as the backend, but that vps is just doing tcp proxying with socat to the ddns exposed home server fqdn. Feels like a lot of unnecessary loops. Kinda fun I guess but, just, why

Yes it is, "we plan to use our external VPS for handling the TLS termination". Edit: Ah I see you are just pointing out termination is on haproxy service not VPS. Thought you were implying it was terminating on pi, my apologies.

The VPS is running socat only and just doing tcp forwarding. There is a shared haproxy also run by their same host, sitting in front of the VPS and is handling the TLS. I encourage you to read the article fully. They probably should have said "VPS provider" instead of VPS for the TLS bit.

But it's plain text like you said in another comment after the haproxy, so two more plain text paths (with at least one going through the internet (vps->pi), not sure if haproxy->VPS is internal to the provider network (maybe)), so not ideal in my book

Not disagreeing with you, but that makes it even worse.

Agreed

This reminds me of the recent "running Doom on DNS" post which in actuality was "running Doom from DNS [as a storage device] on my PC" which is multitudes less impressive.

It is more than a little weird. A pi zero is more than capable of handling HTTP/1.2 and TLS 1.3 for a handful of connections per second. This machine is 10x what we were running web servers on in the '90s.

Also, all web pages are served from RAM. It's automatic that modern OSes will cache this stuff on first access.

>This machine is 10x what we were running web servers on in the '90s.

Kind of irrelevant since operating systems and web pages in the 90's were significantly smaller in footprints, as the web was mostly plain text back then. Windows XP with its GUI would run Max Payne on 128MB of RAM. You could do a lot more back then that You can't do modern stuff like that today with 128MB of RAM.

You can host such sites perfectly well nowadays. I’ve often served hand-written HTML pages of only few lines

LLMs, including open ones, are really good at this it turns out. It stands to reason, there is tons of training material out there no doubt they have consumed and are ready to regurgitate.

Yesterday I one-shotted several interactive pages, that Qwen built out of straight HTML and Javascript. I handed it my API (source code, not even a swagger, via an MCP that Qwen wrote for me), asked for a frontend, and it delivered. One page at a time to keep context down, and mightve gotten lucky on the first draw but after the first one I told it to make the next ones like the first.

Can't say I've had that experience with backend languages & frameworks, incl writing that same API, but perhaps I'm off the beaten path with those, or perhaps there's greater breadth of things to do vs a narrower set of acceptance criteria? IDK.

Here I was sweating that I'd have to research and learn a current-day frontend framework. It felt like a magic wand using consumer-grade AI. HTML and plain old Javascript was plenty.

Tangent but apropos of other contemporary threads on HN, it puts a spin on supply chain threats. There's no NPM or anything, except perhaps whatever mysteries are baked into the model.

The contents of webpages are largely the same.

HTML code, CSS, Javascript, Images.

In this case, they are static elements, which can even be cached locally to share more easily.

If someone wants a massive build system to render a static HTML page, that's on them, and their personal interpretation. Increasingly, and maybe more often than not, there is more than one way to get the same outcome.

The fact that there's hundreds of downloads for a single web page is up to the constructor of that page. Still, these things can be reasonably cached. For example, host it on the Pi, then put a cloudflare in front of it or something.

The Pi Zero might not be for you, or easy to try to undermine. Which criticisms would go away if it was on a regular pi?

Even then... it's usually built before it's deployed n the server.. the server is still delivering text, css, js, images and images have always been pretty large. So your connection is tied up for a little bit longer... and as content was smaller in the 90's, connections themselves are much faster today... in the 90's you were lucky to be hosting on a T1 or faster and clients on modems. Today, you've likely got between 100mb to 2gb uplink on your home connections, let alone business connections that generally start at 1gb. 600x the bandwidth for the server from a T1

It's possible enough to work from the publication point backwards.

A T1 today put behind relevant caching like CloudFlare could still get a lot done.

Today's home internet speeds between 1gig fibre and up can handle a lot too.

Understanding what's needed, how it could be delivered, constraints can all factor into how to approach structuring, developing and publishing any web application.

>Which criticisms would go away if it was on a regular pi?

Maybe you misunderstood. Which criticism did I make of the pi zero? I criticized present day SW.

Yeah, I ran a phpbb forum (alongside my normal static site) on a 486 in 2003 or so. It worked. It was slow, but it worked just fine for my friends and I! I remember it took multiple minutes to generate the SSH server key after the initial install lol

A 486 in 2003? Pentiums were shipping by the mid-90s, did you just have super old hardware lying around?

I retired my 486 in ‘95 or thereabouts…

In the mid-90s, I retired my 486 hardware and brought it over to a local ISP that we were friends with.

It had a second life doing stuff like delivering mail, handling IRC, serving web pages, and whatever else a few of us wanted from it. The performance was fine.

(The Pentium-ish machines stayed on desktop duty where GUIs devoured resources.)

Yeah, I was actually running it on a 386 at first. In those days I didn't have much money and I just dumpster-dived computer hardware that was thrown out behind computer shops. Back before recycling stuff got regulated, PC repair shops threw out totally-working hardware all the time. Every piece of hardware of the web server was free, including the monochrome CRT monitor I had hooked up to it, and the awesome IBM Model M keyboard. The best dumpster find ever was a working Pentium 4 machine, all they took out of it was the HDD! Good times :)

Anyone remember 32 bit/33 MHz PCI slot SSL accelerator cards? As I recall openbsd had kernel driver support for several

Yes. I still have one. I used it in a Soekris 486-class machine running OpenBSD for many, many years. I stopped using the Soekris when they dropped driver support for the accelerator. I think i386 support might be gone now too. Cool little card; I think I might still have it in a drawer somewhere.

I remember the MiniPCI variants, it makes sense that a full-size PCI one would've existed. Post a photo if you find it!

In unrelated news: I think the Magic-1 is still online. ( https://www.magic-1.org be gentle with it)

I was initially thinking does the pi zero even have floating point? But then I remembered that encryption only uses integer operations, not even very large ones.

I wouldn’t consider “the way most people do TLS in 2026” weird. That said this isn’t all that impressive or interesting, a computer… serving a website.

Is sending plaintext traffic over the open Internet "the way most people do TLS in 2026"? Am I missing something from the post?

Many (most?) are hosting web applications and/or content in separate applications and sometimes servers from where TLS (HTTPS) termination happens. HAProxy, Traefik, Caddy and Nginx as reverse proxy and TLS termination servers are pretty common, even more so if you're containerizing your applications themselves. It dramatically simplifies the application stack.

While I may make the argument that most are probably hosting and doing php on the same server, it's not the typical approach for any custom software at this point.

It's vastly different to do TLS termination within your own network and to do it on a rando VPS and then send normal TCP over the internet. It's not an argument of it being on the same server.

The VPS is your security in this case. It's not sending plaintext over the internet, is it?

Edit: No, the article mentions listening on port 80 at home. I thought they'd be SSH tunneling or something. That is unusual, but I guess for a static website it doesn't really matter.

> That is unusual, but I guess for a static website it doesn't really matter.

It sorta does matter. Either the actual raspi does nothing of value or the traffic has value that should be protected.

Sure, I heard the argument that public HTTP traffic does not need encryption but if it is of any value then both parties have a interest in it unmanipulated, uncenscored, validated or all of the before. Even if it is just preventing the ISP injecting dumb ads.

Yeah that's a valid concern. Idk, nothing about this setup makes sense.

Considering that a 'base' raspbian type install can be something like 160MB of RAM used with openssh running and a lot of other launched-from-systemd daemons in the background, that leaves plenty of RAM available for a stock apache2 or nginx setup with TLS. No it won't be able to serve a ton of simultaneous requests, but I'm in agreement with the other comments here that doing purely port 80/http and putting it behind a secondary TLS proxy is not really "serving the website" from the raspberry pi.

Hmm is linux on arm so frugal still?

Don't have a Pi plugged in now to check, but I have a fresh x86_64 linux and it's using like 600 Mb of ram - server install, I only got around to installing and configuring openssh and samba so far.

Oh and it's Devuan, so not even systemd to blame. I think it was close to 1 G with a systemd distro.

It reminds me of the footage of Doom running on a pregnancy test. And then it turned out it was another computer just displaying to the build in AMOLED display.

What was supposed to be a cool achievement is rendered pointless when one of the key elements is offloaded elsewhere.

Why even serve TLS here?

because we collectively seem to have decided that plaintext http is worse than.. the computational cost to the planet.

I have a self hosting Pi Zero W running Gentoo. It started as a joke, but I kept it because it's actually occasionally useful for testing new kernel releases.

I found a fun bug with it a couple years ago: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin...

It is still able to build software faster than it is released. It takes roughly a month to recompile the entire system :D

I self host some CI runners and do kernel work on a Pi writing some software defined radio things.

For the radio stuff I can just take the Pi, frontend, and a battery pack outside to test.

When I finally move to a place with proper fiber internet I'm going to be hosting several side projects on a handful of Pis.

I've been using Raspberry Pi Zeros for cheap little linux appliances since they were released. Boot them up with the latest Alpine Linux and run the whole thing from ram. You can also remove a card from a running machine with no ill affect, and they easily survive power cuts. I've never had a card fail.

A raspberry zero is more powerful than an enterprise server from the 1990s. A minimalist static website is not impressive. You can fit way more in there.

Hey, it loads! Unlike ~10% of pages on first page of HN, hugged to death.

Also I love the dithered B&W images. The entire aesthetic of the site is great.

The website running on the vape was far more interesting than this. I do wonder if anyone has tried to use the microphone in these devices to listen to audio. Backdoored vape

One could record from the microphone and then transmit via RF side channel.

I hosted my personal email domain on a Zero for almost 10 years. It had about the same capability as the very expensive (and large) Win NT4 machine we used for our 80-person organization when I started my career in tech. I eventually replaced the Zero with a Raspberry Pi 4, primarily because the Zero’s IO ports are annoying (eg, USB is not hot-pluggable!) An RPi 4 is extreme overkill for personal email but it still idles under 1W and when it fails I can replace the entire machine for next to nothing.

The point of failure for all of these machines has been the SD card. They seem to last 4 years almost to the day. I suppose if I set up a RAMdisk they might last longer, but honestly, for the price of an SD card it’s not really worth my time.

>The point of failure for all of these machines has been the SD card. They seem to last 4 years almost to the day. I suppose if I set up a RAMdisk they might last longer, but honestly, for the price of an SD card it’s not really worth my time.

There are "Industrial" SD cards which should last considerably longer, you can look up a few people have done their own testing. They can be slower but that shouldn't be a blocker for an email server on a pi.

You can also boot (full size, not-zero) pis off USB disks. Or just netboot.

They make high endurance microSD cards that can handle a lot more writes before failing.

OTOH, I corrupted a card by turning off the Pi in middle of writing.

I remember in the mid-late 90's how poorly the Exchange server ran that there was a nix server for inbound email just to throttle the ingress. When it was upgraded to a 4-socket server, there was concern when the *nix guys just let everything that was being held during the upgrade through, and it just chugged along. But the moment of panic was palpable. The Unix guys really didn't like that business internals and apps were running from Windows services, so thought it would be funny to try to knock over the new mail server.

Today, you can run mailcow/mailu with all the options on a relatively modest vps. I'm on a cable provider that locks down residential customers and charges over 2x as much for business, so it's cheaper to use VPSes.

On RPi, I've mostly opted to use SSD + USB Adapters as they've been significantly more reliable that SD. There's lots of cases that make this configuration a breeze. That said, I've mostly been running Mini PCs since COVID when the RPi got to be more expensive all-in and slower.

Interesting, what tools did you use for email hosting?

I’m scared of self hosting a mailbox.

https://www.xmox.nl/ is pretty good single binary mail solution for personal email hosting, it not too many features of modern webmails are needed.

Self hosting a mailbox is easy - getting email back out is the hard part.

Have been using Mail-in-a-box [1] for about 5 years [1]. I haven't done any maintenance for at least 3 years, besides occasionally clicking restart in the admin webpanel every time it does serious security updates.

I don't send a lot of emails from it, but the ones I do are delivered.

[1] https://mailinabox.email/

It's a relatively steep learning curve if you're not getting paid to do it. However, in my case, I have been running my personal mail server on a trusted VPS host for well over a decade now. After the initial setup, there is really nothing to do except for regular Linux updates/upgrades. I run postfix, dovecot, roundcube, and rspamd. All of the configuration is in Ansible, so if the host goes tango uniform, or if I want to move it elsewhere, downtime should be minimal.

There are a few open-source one-command mail server deployment solutions that do all of the heavy lifting for you. Some of them might even be pretty good. The problem with those is that if you don't understand how your mail server is put together, you're completely stuck if it breaks.

I just use Postfix. I originally ran on NetBSD. I have an alternative approach for spam, so I have no anti-spam tools.

Indeed, you can even permanently run one using a balcony solar panel:

https://solar.lowtechmagazine.com/about/the-solar-website/

My thoughts exactly. People regularly run Pi-Hole on these things, which not only is "serving a website" (the dashboard) but is also being a DNS server.

I am serving a small web interface to control my shutters on an esp32. I even did the experiment to not parse the request and just always respond with the same response, so a webserver for a single page can be trivial (you would have embed images and all other resources into the html then). But of course I am parsing the request, because I need separate routes for the page and for the actions. Since this is on my home lan it doesn't even need ssl. I guess as long as the traffic is low, an esp32 might be able to do ssl. For me that isn't relevant because it isn't on the internet and when I want to connect to it from outside my home lan, I just use wireguard.

maybe i'm being stupid, but what's the complexity behind this? why is this exciting? is it because it's running on memory or because it's running on the pi? i find this trivial unless you're a new engineer.

for the ones that say that the pi can't handle tls that's just stupid. that's trivial as well.

for the ones saying that you need a vps how cloud native are you people? you can just expose a port on your router (if you're brave enough) and have any dynamic dns service point to the correct ip address.

Came here to say the same. The Zero is a very capable machine

Running a mainstream website on a RPi Pico W is more advanced, but still not really challenging as long as the content is static

I guess "Diskless" was defined by the same people that invented the term "Serverless"? That Pi is still using an SD card to boot from.

A Pi with Ethernet can truly boot diskless via TFTP. And later Pi4 and Pi5 can even boot directly from the internet by getting their initial "boot.img" FAT partition via HTTP from anywhere. That would be diskless.

except in RAM based OS the page size and computation resources reduce significantly as OS is using the same memory to run instead of reading from disk and eating up critical memory area for applications, only thing in between a full crash and OS is the swap storage, macOS and Windows would never be even able to achieve this it is possible only due to Linux/BSD and it's level of optimization.

a better way would be to boot via nvme SSD, ethernet boot has a dependency of network, what if you need to debug when network is down or debug the errors/bugs network itself ?

what makes that truly diskless? that's still getting the bytes from a disk somewhere

i think if the disk is not necessary for the whole runtime after boot, that's fair to call diskless

The OP link is not to Pi zero website, here's the actual website that's being hosted on the Raspberry Pi:

https://zero.btxx.org/

PiZero is a pretty solid machine.

I run my micro-homelab on a Pi Zero from 2018. It’s behind Cloudflare tunnels. It runs the apps i need on a DietPi OS within 180MB and it’s uptime is ~8 months.

First time I hear about DietPi OS, does it offer a headless distro?

Yes, you can setup the entire OS through dietpi terminal utilities over ssh.

The pi zero's are great. I have a bunch of them. I used to use them as a tiny server for live webcams streaming to YouTube for customers, but YouTube now have a minimum sub count before you can go live, which sucks. These boards are pretty powerful.

I have never been able to stream video from a raspberry pi zero's official camera. What tools/software were you using?

I'm using an 8MP camera from freenove on a pi zero 2 - it's great.

Oh, it’s been a few years, if I remember right it was an iso, I burnt to the SD and simply changed the details in a config.txt, I would have to dig out my old laptop to get the details.

we're running a complete production grade cloud storage service with Raspberry Pi Zeros at https://getcloud.doshare.me that's how powerful Rpi hardware we've tested it for upto 10k concurrent requests with storage ofcourse, but still too far powerful

What makes it "production grade"? How are you defining that?

RAM? In this economy?!

It’s ddr2 so it is not really affected by the ram crunch

The Pi Zero has 512MB RAM and a one GIGA Hertz CPU. It's a fucking super computer. Maybe not today, but not that long ago and back then people were running much more intensive things on them than hosting a website. It should be perfectly capable of handling TLS. AES might be a bit haeavy without hardware acceleration, but you can also do only ChaCha20 as the single supported server cipher. It would be easy to DDOS, but you should be able to mostly address that with firewall rules rate limiting connection attempts upstream.

I don't mean to shit on this, exploration is nice and putting perfectly fitting hardware to use instead of throwing abundant unnecessary hardware on every simple problem — just to bring it to crawl with loads of shitty bloates software — is good, but it's not particularly impressive.

For optimal moral support, have one of the spare Pis holding a sign, maybe "Pi is our guy"

One of the things that initiated my MacPro5,1->M2Pro migration (~2023Jan) was the simple fact that the latter's SSD was faster than the former's RAM.

And then that Its newer RAM is even several <more> orders of magnitude faster[-than-new.SSD]

----

I now run a fulltime MacPro6,1 – with dozens of gigabytes of RAMdisk – as an integrated swapdisk on my localnetwork.

[RAM_is_cool.bmp]

After seeing what new R-Pi stuff is selling for I went rummaging in the parts drawer and found the following:

- R-Pi Zero W

- Sixfab UPS hat

- Sixfab Cellular IoT App Shield

- R-Pi model 1B

With all this I should be able to make a multiply redundant always-on bastion host. It's awesome that alpine supports the armhf stuff, many OSes have dropped 32bit support entirely.

In the good old days a decade or so ago where the full fat Pi board was always 35 dollars and the zero was just 5, they were so cheap as to be practically disposable. I have an insane number of Pi 3/4 and Zero/ZeroW boards in projects and drawers around the house, but this has massively tapered off as prices have gone up. At one point I had an 8 pi 3 cluster to learn kubernetes/container orchestration techniques on - completely unnecessary, but building the little rack was 85% of the fun. That cluster ran my home stack for years (DNS, home automation, network admin UI etc).

I've since got a lot more interested in the microcontroller community - so many Pi projects should really be microcontroller projects - the esp32 especially scratches the itch for cheap things to hack on, and you can get them for like 6-7 bucks each with wifi.

Yeah I've been using an ESP32-C6 for the latest wifi connected project I'm working on. The RP2040 and RP2350 look interesting too, I have a couple of them but haven't really done much with them.

I assembled a solar server with those parts laying around last year:

- Victron Monocrystalline Panel 90W 12V

- Victron Gel Battery 12V 60Ah

- Victron MPPT Charge Controller 75V 15A

- Raspberry Pi Zero W

- Witty Pi 5

- Sixfab 4G/LTE Base HAT

- Quectel EC25 Mini PCle 4G/LTE Module

Almost 100% uptime except for a few days after a bad winter storm, pretty neat!

Brilliant write up, just 1 correction, the TierHive VPS should only be $1.20 per year not $4 :)

They are powerful little devices. I used a Pi Zero 2 with an ethernet adapter to host an x86 TrackMania² server using BOX64 and it never had a problem. Only swapped it out recently because I needed the Zero 2 for another project.

I have several of these running all sorts of quickie utilities. The key for making things faster (at least for my tasks) was to write everything I need in c#.

For whatever reason, the speed seems far faster than Python for me.

I mean yeah, but the Pi Zero can also do TLS and much more complex tasks so I really don't understand why this is such a big "feat." It's a linux pc just running a webserver. Am I missing something here?

Instead of having an open port in my router and sending data in plain text, I would use an ssh tunnel or a vpn. Or probably put the entire web site on the VPS.

So what benchmarks did you run or what's the advantage? Might as well just run the site on the VPS at this point since you're paying for it?

People run websites on a VPS with 32MB RAM decades ago and it was fine. What is new about this?

I’ve wanted to get into raspberry pi builds for a long time. Off to order one now and try this.

In which country does OP live where their ISP permits them to host a website?

the zero is the coolest PI imo . You can run nearly any network service on it. Great for adguard or pihole. Smokeping. Lots of fun stuff. Supports usb Ethernet too

Nice writeup. darkhttpd + tmpfs is a clean minimal stack — the diskless approach is also a quiet win for SD-card longevity, which usually the first thing to die in long-running Pi setups.

Umm some people run a website on a conmodore 64. That's impressive.

A Raspberry Pi Zero can just run apache.

i'm almost certain you could do tls termination at reasonable scale on the rpi0 no problem

tell OP about tftp

Reply on news.ycombinator.com