Part 7: Long time no see

Hello again!

Yeah… it’s been a long time since part 6. Life got hectic, the conference prep was full-on, and my Pis didn’t make it any easier. But I’m back! A lot has happened, so let’s catch up.

Control Plane rescue mission

Last time we left off, the control plane Pi kept freezing because it was juggling way too much: k3s, networking, SSH - and basically being a Pi superhero. I was juggling ideas, but with time running out and realizing most would just shift the problem, I decided: Pi 4 to the rescue 🚀 More RAM, more stability.

NAT Gateway madness

Turns out, making the control plane also act as a NAT gateway is… not trivial. After a lot of head-scratching, fiddling with Masquerade, iptables, and NAT-rules, and feeling like I was writing modern art instead of networking rules, it finally worked. I honestly cannot recall the exact steps and it also felt a bit hacky in the end. But: It worked :)

Fun with k3s

k3s continued to surprise me… Some mysterious iptables issues popped up — I barely remember how I fixed them, but eventually all nodes were talking, and my tiny cluster was alive again 🎉

Overengineering the blog onto k3s

Yes, I did it again — deploying the blog to the cluster, just like in Part 4 .

And voilà: cluster-powered blogging! Adding a new article, updating the image, and seeing it live on the cluster (with NAT and a separate internal network) worked perfectly — exactly what I wanted for the conference. Felt fancy, not gonna lie 😇

At that point, I made what was perhaps my biggest mistake: I didn’t record the great success as a backup for demo time.

Aesthetics

Remember the chaos from Part 5 :

Some Lego + Pi housing later, my setup looked, well, … different:

When everything went to hell

Final piece for the live demo: WiFi. The tricky part was letting my control plane know the new WiFi network at the venue. I found plenty of advice online — almost all of it mentioned configuring wpa_supplicant. Spoiler: it did not work! And while fighting and juggling, SD cards were dying, time was running out - not to mention my nerves … So, the demo? Mostly theoretical… Which brings me to the last part:

Learnings

• Have spare SD cards at home !!!
• Record as soon as the demo reaches the point you want it to!

In case you are interested in the slides for the very first talk - feel free to have a look !

So, what’s next?

Good news: more conferences are coming up, so the adventure continues. Stay tuned!