Slackware ARM cluster computing ideas
 

Some time ago I started a project to create a Raspberry Pi cluster and would like to expand it. This week I was gifted some cash in the form of several gift cards for my birthday. I purchased a cluster case, a UPS, sd cards, etc and they should all be here over the coming weeks. Among other bits of peripherals, routers and switches I can cannibalize for the project. Additionally, I have 4 raspberry pis of different models, 3 orange pi r1 plus with the USB expansion boards, and a rockpro64.

Here is the case: https://www.amazon.com/dp/B07D5MJ7PQ

Here is my current cluster: https://exitstatus.one/slackware-powered-pi-cluster/

I used it to offload distributed tasks, such as building Slackware packages, educational penetration testing projects, a test bed for various Linux distributions, and a wireless intrusion detection + honeypot.

Does anyone have some good ideas for this cluster that are educational as well as have real world application?

Amazing :thumbsup: You're not using a PoE switch?

I already have all the power supplies that came with each board. So initially no PoE, but it is on the "list".

EDIT:

The orange pi's are going to be powered by my USB hub: https://www.amazon.com/dp/B0797NZFYP

I am always interested in network related projects and thought it best to get this OPi kit: https://www.amazon.com/dp/B08Q88H43W

The raspberry pi's were all bought over the last several years, for different reasons, and not with this purpose in mind. Initially I will be using the opi's and the raspberry pi 4, so that I have room for some SATA disks in the case.

Build a miniature render farm. Try Povray or Blender. This should be a good educational project.
Ed

Good idea! I know nothing about doing that sort of project. It is interesting.

I have 3 orange pi r1 plus (3gb ram), 2 raspberry pi 3s (2gb ram), and a raspberry pi 4B (4gb ram).

I will likely need more than 9GB of RAM and 24 cpu cores worth of resources. Or are those specs sufficient?

I do have all the network hardware but I would be short on hard disks. I have roughly 5 Terabytes worth of hard disks. One of which is a 2TB disk. The rest are 1TB or less. Doesn't video rendering require a lot of disk space too?

Povray is an old program. It had been run on much weaker hardware.

Blender is modern and likely more resource-intensive.

64-bit CPUs with a few GB of RAM should be adequate. Better hardware will enable more complex scenes and higher image quality.

(I have to give the disclaimer that I haven't tried either program since I write my own).

You can start with still images to see how much the hardware can handle.
Ed

Thanks! I would probably task the raspberry Pi's to do the heavy lifting. The orange pi r1's are fairly weak in the GPU department, and are basically designed to be routers (two Ethernet ports). I could easily fill the empty space in the cluster case with some SSD's for each raspberry pi.

I have my rockpro64 in the pine64 case and it is acting as a gateway. I added an antennae to the rockpro to boost the quality of the wifi signal. That way I can download updates to the arm32 and arm64 Slackware repositories. The cluster is on the ethernet port that has it's own separate subnet from the rest of my network.

I will be using containers to segregate the different services for whatever task(s) I am doing. Probably with LXC since it comes with Slackware.