Forging an optimal MetalLB configuration

As someone who’s been playing with a lot of Kubernetes on bare metal lately, I’ve come to appreciate MetalLB (a load balancer implementation for bare metal). Nothing is worse than blindly pasting YAML into your terminal, then seeing Pending next to all your newly created services expecting cloud load balancers. MetalLB was the last thing I needed to make my tiny home lab cluster feel like a real cloud. When I was first configuring it, the hardest thing to wrap my head around was how traffic flowed in the different modes and traffic policies. I spent a lot of time reading docs and experimenting, so hopefully this post will help you understand the different modes and how they work with service traffic policies. ...

May 30, 2022 · 6 min · Patrick Easters

Taking a whack at custom Prometheus alerting

After many years of being on-call under my belt, I never thought I’d say I have a favorite alerting method. But that changed after watching one of Justin Garrison’s videos which had an excellent depiction of how Linux’s Out-of-Memory Killer works. I was no stranger to the OOM Killer visiting my Kubernetes clusters, so this gave me a dumb idea for a (perhaps) fun alerting mechanism: the OOM Bonker. ...

May 11, 2022 · 4 min · Patrick Easters