Raspberry Pi Home Server, hardware update and an unintentional rebuild...

1 month ago by Sebastian

- - - - -

Being a little bit clumsy but eager tinkerer has it’s pros and cons. It’s nice when everything is working as it should, but I’m then becoming restless and the urge to meddle with stuff overtakes the reason. I guess my Raspberry Pi home server has been to stable...

Hardware updates!

For starters, despite my Raspberry Pi 4B setup being stable as a rock, I decided to replace it with something with more „oomph“. Since my dear Wife is weary of fan noise and increasing electricity bills, I opted for a newer iteration of the same - Raspberry Pi 5. I’ve ordered it just on premier and it took some time for the store to actually ship it. But anyways, it’s been here for some time and has been put to work. The second part of hardware update is kind of related to the network rebuild I’ve done recently. Since we now have PoE (Power over Ethernet) switches it’s possible to finally ditch that bulky power supply and use network cable to power on the miniature computer.

Of course Raspberry Pi doesn’t have that capability on it’s own. What makes it possible is a PoE HAT, one of the many additions you can equip your device with. I’ve got this Waveshare 26399 PoE HAT for Raspberry Pi, installed it and... It didn’t boot, at least not fully.

The tedious process of digging it out of the cabinet and connecting it to a display to diagnose the problem has began. After some tries, it became apparent it doesn’t get enough power to supply more demanding eternal devices, just like an USB hard drive I’ve had. After some research I’ve found out I need to edit config file:

/boot/firmware/config.txt

and add following line to it:

usb_max_current_enable=1

The board is limiting the amount of current passed through to USB ports (more details here). After a quick reboot it was back online! That’s more like it. 

With root power comes root responsibility...

The second part of the update is that aforementioned accidental need of rebuild. I’m not 100% sure what exactly happened, but my suspicion is it was a classic case of „the wheel is spinning but the hamster is dead“.

I’ve been trying out CasaOS. It’s a very nice interface for a home „server“ with a Docker service dressed up as App Store. It’s been designed for very interesting and tempting ZimaBoard, but you can run it on Raspberry Pi or any other hardware.

Some of the containers didn’t want to build correctly, so I started messing around with what they call „Install a customized app“. I think my error was I mapped / of the filesystem as one of the container’s volumes. The container didn’t start, so I decided to delete it.

Now, I think it would have not happened if I’ve used Docker CLI, but I didn’t. I decided to remove that container via CasaOS menu, which asked me if I wanted to delete all the container’s data. I clicked „yes“ and presumably everything that was possible to remove, has been removed.

Usually, that would be a tragedy, but knowing myself well enough I made sure of two things:
- it was not „mission critical“ (maybe except DNS sink hole)
- I’ve made sure DNS is redundant
I took the opportunity to rebuild the software part of my setup, so it’s stripped of months of bad decisions and poorly implemented solutions.

It came out to me, that things didn’t want to build for me in Docker because my system was 32bit... Did a clean install of 64bit Raspbian OS, installed CasaOS again and things stared to fall into place.

The changes are as follows:
- I’ve ditched Cockpit, because it’s Docker integration was spotty at best
- got rid of Home Assistant, because I’ve got no use for it and HomeBridge does everything I need
- Transmission and Plex now run in Docker containers
- installed Portainer to manage Docker containers in a better way
- I’m monitoring my stuff with Uptime Kuma, but I still need to figure out the notification part

If it works, leave it be...

Yes, that should be my motto. Fortunately, no important data has been lost and it didn’t affect anyone in a meaningful way (a brief DNS outage), as this is just my toys, and not the „mission critical“ stuff. In the process I’ve learned a lot about Docker and Raspbian and hopefully I won’t make the same mistakes again.

Until the next time, I guess... 😂

Comments

The comment section is currently empty.

Share your thoughts...

Your email address will not be published. Required fields are marked *



Designed by accident by Sebastian Samulczyk © 2023