> ## Content Index
> Fetch the complete content index at: https://lifeinproduction.co.uk/llms.txt
> Use this file to discover other available public pages before exploring further.

# I've Got an Idea....
- URL: https://lifeinproduction.co.uk/building-my-media-stack-ive-got-an-idea/
- Published: 2026-09-26T16:24:46.000Z
- Updated: 2026-09-26T16:24:46.000Z
- Description: What started with an unused mini-PC and a dusty NAS somehow became my own automated streaming platform. Welcome to the rabbit hole — and the first part of Building My Media Stack.
- Author: Eric MacFarlane
- Tags: Life, Technology, Building My Media Stack

*Building My Media Stack — Part One*

There are certain words that usually mean I'm about to disappear down another rabbit hole.

**"I've got an idea....."**

It normally starts innocently enough.

I'll just install this.

I'll just change that.

I'll just see if this works.

And then several hours later I'm staring at a terminal window, I've got 14 browser tabs open, Docker containers are restarting, AI is telling me everything is going to be fine, and I'm wondering exactly where my evening went.

This particular rabbit hole started with one simple idea:

**I wanted my own Netflix.**

Not literally Netflix, obviously. Their lawyers can stand down.

I wanted something that *felt* like Netflix.

Something where I could sit down, pick something to watch and press play.

Something my kids could use when they're staying with me.

Something I could share with close friends and family.

And, most importantly, something that I built myself.

Turns out that last requirement changes things quite considerably.

## I've Been Here Before

This wasn't my first media server.

Years ago, I ran Plex.

I loved it.

There was something brilliant about having your own collection available whenever you wanted it. Films, TV and music sitting there waiting rather than having to work out which streaming service happened to own what this month.

Then life happened.

I changed jobs. Things changed. The server disappeared.

Eventually the whole thing became one of those projects from another part of your life.

Meanwhile, streaming changed too.

What originally felt like a fantastic alternative to traditional television slowly became another collection of subscriptions.

Netflix.

Disney+.

Amazon.

Apple.

And whatever other service happens to own the one programme you actually want to watch.

Individually they aren't necessarily ridiculous.

Add enough of them together and suddenly you're looking at another sizeable monthly bill.

And I'm a geek.

That's quite an important part of this story.

Because when most people look at that problem they probably think:

*"I'll cancel a couple of subscriptions."*

I apparently think:

**"I'll build an entire automated media platform."**

Perfectly reasonable.

## Some Forgotten Hardware

Fortunately, I already had the beginnings of a homelab sitting around doing very little.

The main machine was a TRIGKEY G5 mini-PC:

- Intel N100 quad-core processor
- 16 GB DDR4
- 500 GB PCIe SSD
- Dual Ethernet
- Wi-Fi 6
- Bluetooth 5.2

Nothing particularly exotic.

No enormous rack of enterprise servers humming away in the garage.

Just a little mini-PC that had previously been running Proxmox.

It had once handled things like DDNS, Let's Encrypt and reverse proxy access to my Home Assistant server. But I'd since moved much of that functionality elsewhere using things like Cloudflare and Tailscale.

The little box was basically redundant.

I also had an old five-bay QNAP NAS sitting powered off and collecting dust.

So naturally...

**I've got an idea.....**

I wiped the mini-PC, installed Debian 13 and decided to keep things simple.

No Proxmox.

No unnecessary layers.

Debian.

Docker.

Let's build something.

## Hello, Jellyfin

The first container was Jellyfin.

That was the point where the project really started.

Jellyfin gave me the core of what I wanted: my own media library with a proper interface that could be accessed from TVs, phones, consoles and other devices.

But almost immediately I started looking at it and thinking:

*"That's good... but what if..."*

Dangerous words.

Because Jellyfin wasn't really the destination.

It was the starting point.

I didn't just want somewhere to store movies.

I wanted the whole thing automated.

If I wanted a film or television programme, I wanted to request it and let the system handle everything else.

Search for it.

Find it.

Download it.

Rename it.

Organise it.

Add subtitles where needed.

Put it into the correct library.

And finally have it appear in Jellyfin ready to watch.

That was the idea.

The reality involved considerably more Docker containers.

## The Stack Begins to Grow

This is where I discovered the wonderful world of the \*Arr applications.

Sonarr handles television.

Radarr handles movies.

Lidarr handles music.

Prowlarr ties the search/indexer side of things together.

Bazarr handles subtitles.

Seerr provides the nice request interface sitting in front of everything.

Then there are the download clients, Usenet, indexers, VPN routing and all the other pieces required to make everything talk to everything else.

One container became two.

Two became five.

Five became...

Actually, I've stopped counting.

But that's half the fun.

The goal was that eventually the complicated stuff should disappear.

Someone shouldn't need to understand Sonarr, Radarr, Prowlarr, indexers or Usenet.

They should just say:

**"I want to watch this."**

The machinery behind the scenes can deal with the rest.

## Then Something Magical Happened

With a lot of help from AI, I started exploring things I'd never really used properly before.

Usenet.

Indexers.

Docker networking.

Automation.

We worked our way through setting up the different services, connecting everything together and fixing the inevitable problems along the way.

And then one day...

It worked.

A movie travelled through the system and appeared automatically in Jellyfin.

All that setup.

All those containers.

All that configuration.

All those terminal commands.

The first glorious proof that my automated media empire was alive.

And what cinematic masterpiece marked this historic occasion?

**Minions & Monsters.**

Not exactly *2001: A Space Odyssey*.

But there it was.

Sitting in Jellyfin.

Ready to play.

I may have been disproportionately pleased about this.

## Storage. Slight Problem.

There was, however, a fairly obvious problem.

The mini-PC only had a 500 GB SSD.

That's absolutely fine for Debian, Docker and application data.

It's considerably less impressive when you start feeding it movies and television programmes.

Fortunately, remember that dusty QNAP?

I had a plan.

Years ago I'd also owned a Drobo NAS containing around **16 TB of media I'd collected over roughly ten years**.

Movies.

Television.

Music.

The obvious idea was to reuse the drives.

There was just one slight problem.

I hadn't touched the Drobo in more than six years.

Passwords?

No idea.

Login details?

Gone.

Access to the data?

Nope.

Eventually I had to accept it.

The collection was gone.

**Sixteen terabytes.**

Fortunately, my photos weren't part of that loss. Those are safely stored elsewhere in iCloud.

But the media library I'd spent around a decade building?

Gone.

That hurt.

It also changed how I approached the new system.

If I was rebuilding this thing, I wanted to build it better.

## Enter the QNAP

The old QNAP suddenly had a purpose again.

It's a five-bay NAS, and eventually I managed to get hold of five 8 TB SATA drives.

Suddenly our little media server had some serious storage available.

The architecture was beginning to make sense.

The mini-PC could concentrate on running Docker and all the applications.

The QNAP could concentrate on storage.

And because the applications and media were separated, the whole thing became much easier to grow and maintain.

Of course, saying *"much easier to maintain"* is slightly ironic considering what happened next.

## The Rabbit Hole

At some point I made another mistake.

I looked at Jellyfin and thought:

*"I wonder if I could make this look more like Netflix?"*

And so began **Jellyfin Cinema**.

Initially it was just some interface customisation.

Then a little CSS.

Then some JavaScript.

Then a cinematic hero.

Then navigation.

Then redesigned media pages.

Then episode pages.

Then season selectors.

Then...

You can probably see where this is going.

I wasn't customising Jellyfin anymore.

I was effectively building another interface on top of it.

And it looked bloody good.

Until playback stopped working.

## I've Broken Jellyfin

This was one of those moments where your stomach sinks slightly.

Everything had been working.

Then suddenly it wasn't.

Naturally, my first assumption was:

**I've broken it.**

Considering the amount of custom JavaScript and CSS we'd been injecting into Jellyfin, that wasn't exactly an unreasonable conclusion.

So we started debugging.

We stripped things back.

We tested.

We checked the custom interface.

We checked JavaScript.

We checked Jellyfin.

At one point the evidence seemed to suggest we might actually be dealing with a bug in the version of Jellyfin itself.

That sent us even further down the rabbit hole.

Hours became days.

And somewhere during all this I started considering the nuclear option:

**Build a completely new Jellyfin server.**

Start again.

I really didn't want to do that.

Then came the breakthrough.

Hidden away inside Jellyfin's own settings was some custom CSS we'd added earlier in the project.

We'd forgotten about it.

Removed it.

Playback returned.

Everything worked.

Several days of debugging.

Numerous theories.

Potential Jellyfin bug.

Possible complete rebuild.

**CSS.**

Welcome to IT.

## My Other Engineer Is Artificial

There's something else worth mentioning about this entire project.

I've leaned heavily on AI.

I'm not hiding that.

In fact, it's one of the reasons I've been able to push the project as far as I have.

I've worked in IT for years, so I understand infrastructure, networking, troubleshooting and the general principles behind what we're building.

But that doesn't mean I know every Docker application, Linux command, configuration format or API off the top of my head.

AI has effectively become another engineer sitting beside me.

We discuss ideas.

Build configurations.

Write scripts.

Break things.

Read logs.

Work out why we broke things.

Fix them.

Then immediately come up with something else to change.

It's allowed me to explore technologies I might previously have looked at and thought:

*"I'll figure that out another day."*

Now I can dig in.

Ask questions.

Understand what's happening.

Experiment.

And learn while I'm doing it.

It's brilliant.

At least until Skynet becomes self-aware.

At which point I assume my first warning will be:

`docker compose down`

## Just One More Thing.....

The dangerous part about building something like this is that it never really ends.

Because once something works, I immediately start thinking about how it could work **better**.

Take the last few days.

We built a fully automated backup system.

It stops the relevant Docker containers.

Backs up their configuration.

Encrypts the backup.

Restarts everything.

Uploads the encrypted archive off-site to Google Drive.

Checks the hashes to make sure the backup arrived intact.

And then emails me to tell me whether everything worked.

Which required proper email delivery.

So...

**Just one more thing.....**

We built a central SMTP relay.

Then there was Caddy.

And maintenance pages.

And internal DNS.

And remote access.

And Jellyfin Cinema.

And monitoring.

And...

You get the idea.

Every time I think the stack is approaching "finished", I discover another thing that would make it slightly better.

Or safer.

Or easier.

Or simply cooler.

And apparently that's enough justification.

## So What Actually Makes Up the Media Stack?

I'm deliberately not going deep into the configuration in this post.

That comes next.

But at a high level, the system now looks something like this:

![](https://lifeinproduction.co.uk/content/images/2026/09/6d8834ed-7648-404d-a9e1-51e7c37ed7f1.png)

The current Building My Media Stack architecture. And yes, it started with just Jellyfin.

**Requests**

Seerr gives people a simple way of asking for something to watch.

↓

**Automation**

Sonarr, Radarr and the rest of the \*Arr stack work out what is needed.

↓

**Discovery**

Prowlarr and the configured indexers help find it.

↓

**Download**

The download layer handles actually retrieving the media.

↓

**Storage**

The finished media ends up on the QNAP.

↓

**Playback**

Jellyfin presents everything through the interface people actually see.

Around all of that are the supporting systems.

Docker.

Debian.

Networking.

DNS.

VPNs.

Remote access.

Reverse proxies.

Email.

Backups.

Monitoring.

And probably three other things I've forgotten I've installed.

The important thing is that the person sitting on the sofa doesn't need to understand any of it.

They see a poster.

They press play.

That's the whole point.

## Building My Media Stack

This post is really the beginning of another series for *Life in Production*.

I'm calling it:

**Building My Media Stack.**

Over the next posts I'm going to pull this thing apart and look at how the individual pieces actually work.

We'll look at Jellyfin.

Docker.

Sonarr and Radarr.

Prowlarr.

Seerr.

Usenet and indexers.

Storage and the QNAP.

Networking and remote access.

Backups.

And yes, we'll definitely return to Jellyfin Cinema and the adventure that nearly resulted in me rebuilding the whole bloody server because of some forgotten CSS.

I'll include configuration examples where they're useful.

I'll show some of the architecture.

I'll explain what worked.

And, probably more importantly, I'll show what **didn't** work.

What I won't publish are credentials, API keys, private addresses or configuration that could compromise my own infrastructure.

Because sharing knowledge is good.

Publishing the keys to your house on the internet is less good.

## Follow the Rabbit Hole

There's a bigger reason I want to write this series.

You don't have to know everything before you start.

I certainly didn't.

Sometimes the best way to learn something is simply to build it.

Read.

Experiment.

Break something.

Work out why it broke.

Fix it.

Understand a little more than you did yesterday.

Then go a bit deeper.

There will be moments when nothing makes sense.

There will be evenings where something that should take ten minutes consumes three hours.

There will almost certainly be a moment where you consider deleting everything and starting again.

Don't be afraid of that.

Follow the rabbit hole occasionally.

You might be surprised where you end up.

After all, this whole thing started with an unused mini-PC and an old NAS gathering dust.

Now I've effectively built my own automated streaming platform.

And it's working.

The storage is growing.

The automation works.

The backups work.

People can request something and watch it.

Maybe, finally, I can call it finished.

Although...

**I've got an idea.....**