Tagged: GCP

K8s: Stopping kube-system pods from preventing cluster scale-down

I’m now at the point where I run a fair few production services from Kubernetes, the most successful being departureboard.io and the departureboard.io API. To make sure I can meet the fluctuating demand for these services, in a cost-efficient fashion, I take advantage of the Kubernetes Cluster Autoscaler. I won’t go into too much detail about how this works, but in short, it enables my cluster to automatically scale up if the total resource requests from my application pods exceed the available capacity in the cluster.

This blog now runs on Kubernetes: Here's my architecture

As I alluded too in my previous post, I’ve recently been working in moving my blog (and all of my services) to Kubernetes. Up until this point, I’ve been running all of my services in containers, but on specific machines with Docker installed. I had a pretty sophisticated and automated deployment process but it was only ever an interim solution before I delved into Kubernetes. The decision to move to Kubernetes was further expedited by being given access to the Digital Ocean Kubernetes Limited Availability Preview.