Autoscaling, Rolling Deploys with Drupal on AKS
summary
A production Drupal stack on AKS uses custom Docker PHP and Nginx images plus upstream Varnish, memcache and Elasticsearch, deployed with Helm and GitHub Actions CI/CD; autoscaling via Kubernetes HPA (requests/limits set, scale 1→10 on CPU 70%) handles bursts—70,000 daily logins since November with zero-downtime rolling/ramped deploys and canary shadow testing. Monitoring uses Dynatrace, Prometheus, Grafana and Loki; tests use Cypress and k6. Key lessons: hardcode image versions, set resource limits to avoid CPU throttling, add a unique deploy ID to avoid Drupal cache bugs.
AI-written summary · about
chapters (14)
- 0:00Introduction and background
- 2:56Zero downtime deployments with Drupal
- 4:00Why use Kubernetes for Drupal
- 7:15Repository structure and CI setup
- 9:03Creating Docker containers
- 16:01Kubernetes basics and concepts
- 23:06Autoscaling pods in production
- 27:49Deployment strategies for zero downtime
- 31:45Ingress and services explained
- 33:11Persistent storage and config maps
- 36:47Helm charts and infrastructure code
- 39:59Open-source Drupal Helm chart
- 41:05Common pitfalls and solutions
- 44:03Q&A and closing remarks
related talks
42mDrupal, Kubernetes, Production
45mHow companies are currently using Kubernetes with Drupal
01h 10mHosting Drupal on AWS
55mReturn of the Clustering: Kubernetes for Drupal
22mDrupal HA: a story from the trenches
30mAutoScaling WordPress with Docker & AWS
30mScaling to 5000+ Unique K8s Deployments, How We Did It
56mAzure Kubernetes Services 101 for Developers