Read one link per day, mull on it, and ask clarifying questions from your local SRE
- What is a deployment?
- Don't ready the whole doc - It's long and technical. Just read the preview, and know the primitives it mentions - Create, Update, Rollback, Scale, Pause, Status, Cleanup.
- Pods
- Persistent Volumes in k8s.
- Editorializing for a bit; there's a bunch here, its very complex and only adds more. There's very few good reasons to use volumes. It doesn't reduce complexity vs. provisioning your backing store in more traditional ways. K8s is great because we're moving to stateless microservices; for stateful services like database backends, generally a losing proposition. Except when you can rapidly and automatically rebuild from quorum; I'll discuss those later.
- What is a node?
- Kubernetes Service
- Kubernetes Basics
- Updating a deployment
- Scaling, p1: - Deployment; Scales pods
- Scaling, p2 - HPA; Sizes Deployments
- Scaling, p3 - Cluster Autoscaler; Scales nodes
- Our Kubernetes dashboard is available at: api.stg2.secrethouse.party/ui - Explore it, click around, see what's available.
- Kubernetes patterns and anti-patterns
- https://k8guard.github.io/about/
- On the origin of Kubernetes
- HPAs, again
- A cronjob creates jobs on a schedule
- Kubecfg is a templating language, based on jsonnet, for writing kubernetes configs.
- DNS in kubernetes
- There's a bunch of cool stuff you can do here. Service discovery using loadbalanced or headless services is fairly flexible, allowing you to get an IP that directs you to a loadbalanced pod, or a list of pods running under that service.