A brief collection of bullet points highlighting some of the things I’ve been working on.

• Returned my iPhone SE 2022 to the Apple Store
  • I barely lasted a week.
  • I loved the device physically, but felt the software was too limiting
  • Some of the things I really missed: Loop Habit tracker, new pipe, fast lte speeds
  • For some reason the LTE on my device wouldn’t go higher than ~30 mbs where my S22 Ultra or S24 (what I ended up on) could hit 100 mbs or more
• Implement basic web app for carplayinstaller.com
  • Working with local providers to help boost their services, get more customers.
  • Basic ordering form; saves time on intake
  • Added analytics
    • I want to monitor site performance before investing too much time on the app itself
    • there’s no point developing the app if no one is looking at it
    • the basic functionality works
  • If you run into any issues please send me an email [email protected] and I’ll take a look 🙂
  • Deployed a basic nodejs backend for persisting orders
  • Tried to keep it stupid simple by using for storing orders as json – https://jsonbin.io/
  • Working on a new logo for this – my GIMP skills are rusty…
• Spent a better part of 3 days setting up a prod-like kubernetes cluster on local proxmox hardware
  • the main goal was to run EFK: elasticsearch, fluentd, and kibana
  • I want a centralized logging platform
  • lesson learned: running a centralized logging platform as a one person team is way more complicated than it should be
  • clearly there is a huge industry behind automating infra and moving it to the cloud or other managed solutions.
    • If large organizations are happy throwing a lot of money at this problem, more power to them
    • But for me personally, I think it is crucial we are able to self host software that is reliable (think 3 9s), simple (architecture should fit in a single persons head) and easy, one-click, zero conf install to cover 80% of use cases
• Learning about erlang
  • my issues with kubernetes and unbounded complexity while trying to deploy the EFK stack, got me thinking that there must be simpler ways to do this
  • Let’s cut the complexity, let’s make it simple, reliable and easy to use
  • So one of my goals now is to build a replacement for something like EFK, just for logs and just for 80% of the use cases. It should be much much much smaller than EFK.
  • Stay tuned for more
• Miscellaneous things while setting up local infra
  • Backup data from previous hyperv cluster – reusing an exisiting server
  • install proxmox on that server
  • create debian VMs to run k8s nodes
  • realize minikube can do multi node but not multi host multi node
    • this means it can run 3 k8s nodes in a single minikube on one debian VM
    • It cannot do 2 debian VMs running 3 nodes
  • Moved to k3s to overcome the single VM limitation of minikube
  • Lost a day configuring tls and auth between fluentd and elasticsearch
    • lesson: “community” or org sponsored helm charts such as those by elastic and bitnami are very complicated, the documentation is often out of date, and does not show how to get the common use cases done
      • Probably good for making money as consultants?
• GPTel and other local GPTs
  • got my emacs running on a dev debian vm to use gptel (emacs extension) to connect to a backend running ollama on my m3 macbook.
  • The macbook has better hardware for the AI workloads
  • install Jan as a local replacement for ChatGPT
    • trying to significantly reduce my use of external services
    • the internet is distracting
    • there are bad actors, its not clear who is collecting our data, what their collecting, what they’re doing with it and who they’re resharing it to

I wrote this post partially as a way to share and feel more connected as I dive deeper and deeper into my research. It was a rough format, mostly stream of conscious. I was originally intending to write this on Saturday, but here we are Tuesday at 5.30 am. I’d rather get something out than nothing.