Containers & Kubernetes

Lean, reproducible images, a registry with rules, clusters laid out and upgraded on schedule, requests, probes, secrets and policies set properly, and the habits that keep workloads healthy.

Free · Mac (Apple silicon) and Windows · v1.0.5 · Other platforms

Then open the Marketplace in the app, search for Containers & Kubernetes, and click Install. It goes into your own files, and works offline from then on.

What it installs.

1

Pillar

1

Area

50

Projects

205

Starter tasks

⚙️ Software Engineering

Containers & Kubernetes

Packaging applications in containers and running them on orchestration platforms, including scaling, configuration, secrets and cluster upkeep.

  1. Inventory of containerised workloads and clusters
  2. Decision on whether Kubernetes is worth running
  3. Production Dockerfile for the main service
  4. Approved base image standard
  5. Container registry with access and retention rules
  6. Image tagging and digest convention
  7. Local development with Compose that mirrors production
  8. Namespace layout and naming conventions
  9. Least-privilege cluster access for engineers
  10. Cluster upgrade cadence and version tracking
  11. Monthly base image refresh and rebuild
  12. Resource requests and limits from real usage
  13. Node pool and capacity health check
  14. Registry clean-up and storage hygiene
  15. Secret rotation for cluster workloads
  16. Cluster backup and restore drill
  17. Certificate expiry watch for ingress and cluster components
  18. Weekly review of crashlooping and evicted pods
  19. Cluster add-on version register
  20. Kubernetes core objects study plan
  21. Debugging a failing pod step by step
  22. Container internals, namespaces, cgroups and layers
  23. Writing a Helm chart from scratch
  24. Kustomize overlays for each environment
  25. Kubernetes networking with services, DNS and ingress
  26. Liveness, readiness and startup probes done right
  27. Pod scheduling with affinity, taints and topology spread
  28. Image size reduction for the largest images
  29. Managed Kubernetes provider and distribution choice
  30. GitOps controller for cluster state
  31. Horizontal pod autoscaling on the right signal
  32. Ingress controller or Gateway API choice
  33. External secrets operator in place of manual secrets
  34. Pod disruption budgets and graceful shutdown
  35. Moving the first service onto the cluster
  36. Major Kubernetes version upgrade in production
  37. Migrating workloads to a new cluster
  38. Cluster headroom review before a peak
  39. Retiring a cluster and its leftovers
  40. Solo engineer running containers without a cluster
  41. Small team adopting Kubernetes without a platform specialist
  42. Kubernetes skills matrix for the team
  43. Inheriting a cluster nobody documented
  44. Quotas and tenancy on a cluster shared by teams
  45. Cluster help desk for product teams
  46. Pod security standards and admission policies
  47. Default-deny network policies between workloads
  48. Stateful workloads on Kubernetes decision
  49. Writing a custom operator or controller
  50. Image signing and provenance checks at admission

Where it starts.

Each project arrives with its purpose, milestones and first tasks written out. Here is how the first three read.

Inventory of containerised workloads and clusters

Most teams cannot say on demand how many clusters they run, which Kubernetes version each is on, or which images in production were built from a Dockerfile nobody has opened in two years. A single inventory of clusters, namespaces, workloads and their image sources turns guesswork into a list you can prioritise from.

First tasks

  • List every cluster and kubeconfig context you can reach in a shared sheet
  • Export deployments, statefulsets and daemonsets with their images from each cluster
  • Look up the end of support date for each cluster's Kubernetes version
  • Mark workloads using public images or the latest tag
  • Write the five most worrying findings at the top of the sheet

Decision on whether Kubernetes is worth running

Kubernetes solves scheduling, self-healing and multi-service networking at the price of a steep learning curve and a standing upgrade burden. For three services and a database, a managed container service or a couple of virtual machines running Compose is often cheaper in hours and money. Writing the decision down, with the trade-offs, saves a year of running a platform that nobody needed.

First tasks

  • Count your services, environments and the engineers who will operate them
  • Estimate monthly running cost for each of the three hosting options
  • Ask each engineer how confident they are operating a cluster unaided
  • Write the decision and the triggers that would reopen it

Production Dockerfile for the main service

Tutorial Dockerfiles usually run as root, ship the compiler and test fixtures to production and rebuild every dependency on any code change. Rewriting the main service's Dockerfile as a multi-stage build with a pinned base image, a non-root user and a sensible layer order makes images smaller, faster to build and safer to run.

First tasks

  • Record the current image size and cold and warm build times
  • Split the Dockerfile into a build stage and a minimal runtime stage
  • Pin the base image by version and digest
  • Add a non-root user and set it with the USER instruction
  • Write a .dockerignore covering secrets, git history and build output

Make it yours.

Install Containers & Kubernetes, then tell Aurum the date you're aiming for. It checks whether the plan fits the weeks you actually have.

Free · Mac (Apple silicon) and Windows · v1.0.5 · Other platforms

Blueprints are general information to help you organise your own work. They are not medical, financial, legal or other professional advice. See the Terms.