Phase 2 — Intermediate Operations (Days 31–60)

Week 5: Kubernetes — Advanced Workloads & Networking

Topics:

• StatefulSets, DaemonSets, Jobs, CronJobs
• Ingress controllers (NGINX Ingress, Traefik)
• Network policies
• Persistent Volumes, PVCs, StorageClasses
• Helm: install, upgrade, rollback, write your own chart
• Kustomize: base + overlays pattern

Hands-on Labs:

• Deploy a StatefulSet for Postgres with PVC
• Install NGINX Ingress; route traffic to two services by path
• Package the sample app as a Helm chart with values.yaml overrides
• Apply a NetworkPolicy that restricts pod-to-pod traffic

Week 6: Observability — Logs & Metrics

Goal: Follow what's happening in a live system; detect and diagnose issues.

Topics:

• The three pillars: Logs, Metrics, Traces
• Prometheus: scrape configs, PromQL basics, alerting rules
• Grafana: dashboards, data sources, alert notifications
• Node Exporter, kube-state-metrics, cAdvisor
• Log aggregation: Loki + Promtail stack (or EFK: Elasticsearch + Fluentd + Kibana)
• kubectl logs --previous, stern for multi-pod tailing
• Distributed tracing concept: OpenTelemetry, Jaeger (overview)
• SLIs, SLOs, error budgets — the language of reliability

Hands-on Labs:

• Deploy kube-prometheus-stack via Helm; import K8s dashboard in Grafana
• Write a PromQL query: CPU usage per pod, memory limits hit
• Set up an alert rule: fire when pod restarts > 3 in 10 minutes
• Deploy Loki + Promtail; search app logs from Grafana
• Instrument the sample app with a /metrics endpoint (use prom-client or prometheus_client)

Resources:

• Observability Engineering by Charity Majors, Liz Fong-Jones, George Miranda
• Prometheus docs: prometheus.io/docs

Week 7: CI/CD Pipelines

Goal: Automate build → test → containerize → deploy with no manual steps.

Topics:

• CI/CD concepts: pipeline stages, artifact promotion, environment gates
• GitHub Actions: workflows, jobs, steps, secrets, reusable workflows
• GitLab CI (optional if org uses it): .gitlab-ci.yml
• Pipeline patterns: build on PR, deploy to staging on merge to main, deploy to prod on tag
• Image tagging strategy: sha, semver, latest (avoid latest in prod)
• GitOps with ArgoCD: App of Apps pattern, sync policies, auto-rollback on drift
• Secret management in pipelines: GitHub Secrets, HashiCorp Vault, AWS Secrets Manager

Hands-on Labs:

• Build a GitHub Actions pipeline: lint → test → docker build/push → deploy to kind
• Add image vulnerability scanning step (trivy); fail on CRITICAL
• Set up ArgoCD; connect it to a GitOps repo; deploy sample app via ArgoCD
• Implement branch-based environment promotion: dev → staging → prod namespaces
• Store a DB password in Vault or AWS Secrets Manager; inject it via External Secrets Operator

Week 8: Rollback Strategies

Goal: Know exactly how to safely and quickly undo bad deployments.

Topics:

• Kubernetes rollout rollback: kubectl rollout undo
• Deployment history and revision tracking
• Blue/Green deployments
• Canary deployments with Argo Rollouts or Flagger
• Feature flags as a rollback mechanism (LaunchDarkly concept)
• Helm rollback: helm rollback <release> <revision>
• ArgoCD sync rollback: reverting Git commit + ArgoCD sync
• Database migration rollback strategies (Flyway, Liquibase)
• Incident runbooks: define your rollback decision tree

Hands-on Labs:

• Deploy a broken image version; execute kubectl rollout undo; verify recovery
• Install Argo Rollouts; implement a canary: 10% → 50% → 100% traffic shift
• Simulate a bad Helm release; run helm rollback
• Write a 1-page rollback runbook for the sample app