Continouos Delivery in Kubernetesland

Transcript

1. 1 CONTINUOUS DELIVERY IN KUBERNETESLAND Entwicklertag Frankfurt 2018 Nicolas Byl, codecentric AG
2. 2 . 1 GETTING STARTED
3. 2 . 2 Mission Statement You can follow the pipeline on your device. Install prequisites
4. 2 . 3 Prequisites https://github.com/nbyl/cd-workshop-demo
5. 2 . 4 Organisational Stuff Ask questions anytime! Breaks? Slack
7. 2 . 5 3 . 1 CONTINUOUS DELIVERY
8. 3 . 2 Key Concepts The software is releasable at at any time The whole process from source to production is automated Decouple technical rollout from feature rollout
9. 3 . 3 Continuous Delivery vs. Continuous Deployment Continuous Delivery: Software is releasable at any time Continuous Deployment: Software is released on every change
10. 4 . 1 DOCKER
12. 4 . 2
13. 4 . 3
14. 4 . 4
15. 4 . 5
16. 4 . 6 4 . 7 Dockerfile FROM java:8 MAINTAINER Marcel Birkner <marcel.birkner@codecentric.de> ADD target/edmp-sample-app*.jar app.jar RUN bash -c 'touch /app.jar' ENTRYPOINT ["java","-jar","/app.jar"]}
18. 4 . 8 5 . 1 KUBERNETES
19. "Kubernetes is an open-source platform for automating deployment, scaling, and operations of application containers across clusters of hosts, providing container- centric infrastructure."
20. 5 . 2 portable: public, private, hybrid, multi-cloud extensible: modular, pluggable, hookable, composable self-healing: auto-placement, auto-restart, auto- replication, auto-scaling
21. 5 . 3
22. 5 . 4 5 . 5 Pods & Co.
23. 5 . 6 Pods
24. 5 . 7 Replication Controller  P P P
25. 5 . 8 Deployment combination of pod & replication controller edited as a unit
26. apiVersion: extensions/v1beta1 kind: Deployment metadata: name: frontend spec: replicas: 3 template: metadata: labels: app: guestbook tier: frontend spec: containers: - name: php-redis image: gcr.io/google-samples/gb-frontend:v4 resources: t
27. 5 . 9 5 . 10 Services
28. 5 . 11 endpoint for a set of pods an external endpoint can be resolved using DNS environment variables
30. 5 . 12 6 . 1 SHOWCASE PROJECT
31. 6 . 2 confy microservice to manage conference talks and speakers technology: REST-API UI using Gradle Spring Boot (Web, JPA) ng-admin https://github.com/nbyl/confy
32. 6 . 3 Target Platform kubernetes PostgreSQL
33. 6 . 4 Testing Requirements automatic testing using the target database user acceptance test before exposing new versions
34. 6 . 5 Build Pipeline build publish docker container integration test user acceptance test production
35. 6 . 6 Assumptions any manual interaction will be done in reasonable period our cluster will (for now) contain of only one host
36. 7 . 1 LAB 1: SETUP
37. 7 . 2 Recently on this Program... Have you installed the prerequisites?
38. 7 . 3 Create a cluster minikube start --memory 4096 kubectl apply -f minikube/storageclass.yml
39. 7 . 4 Install Helm helm init
40. 7 . 5 Install Jenkins helm install stable/jenkins --set Agent.Memory=1024Mi --name=cd minikube service cd-jenkins
41. 8 . 1 LAB 2: CONTINUOUS INTEGRATION
42. 8 . 2 Objectives build an artefact of the software run all unit tests and in-tree integration tests
43. 8 . 3 Jenkins Kubernetes Plugin creates a new pod for every job use the pod as a temporary build slave
44. 8 . 4 Let's Go create a new pipeline job use as SCM source for your Jenkinsfile https://github.com/nbyl/cd-workshop-demo.git
45. 8 . 5 caveat build cache is gone after every build
47. 8 . 6 8 . 7 persistent volumes create manually use storageclass with auto-provisioner
48. kind: StorageClass apiVersion: storage.k8s.io/v1beta1 metadata: name: generic provisioner: kubernetes.io/host-path
49. 8 . 8 kind: PersistentVolumeClaim apiVersion: v1 metadata: name: gradle-cache annotations: volume.beta.kubernetes.io/storage-class: "generic" spec: accessModes: - ReadWriteOnce resources: requests: t 8Gi
50. 8 . 9 9 . 1 LAB 3: BUILDING A DOCKER CONTAINER
51. 9 . 2 Objectives build a docker container containing the application make the container available for deployment
52. 9 . 3 Build Pod Revisited using build pods is scalable and reproducible the host docker daemon is not reachable
53. 9 . 4 Docker-in-Docker Builds mount the docker sockets inside the container /run/docker.sock /var/run/docker.sock image will be built in host docker Warning: possible security problem
54. 10 . 1 LAB 4: INTEGRATION TESTING
55. 10 . 2 Objectives create a deployment description for the application deployment the application in a configuration analog to production run an integration test verifying the basic functionality
56. 10 . 3 Helm Chart helm create helm/confy [edit] helm upgrade --install dev-confy helm-confy
57. 10 . 4 Configure the Application env: - name: SPRING_DATASOURCE_URL value: {{ .Values.database.url }} - name: SPRING_DATASOURCE_DRIVER value: {{ .Values.database.driver }} - name: SPRING_DATASOURCE_USERNAME value: {{ .Values.database.username }} - name: SPRING_DATASOURCE_PASSWORD value: {{ .Values.database.password }}
58. 10 . 5 Simulate roduction use the same database server as in production treat system as blackbox
59. 11 . 1 LAB 5: USER ACCEPTANCE TESTING
60. 11 . 2 Objectives deploy the application accessible for users allow manual testing continue to production after manual confirmation
61. 11 . 3 Ingress
62. 11 . 4 Ingress Controller Read Ingress state route accordingly Implementations: NGINX traefik F5 ...
63. 12 . 1 LAB 5: PRODUCTION DEPLOYMENT
64. 12 . 2 Objectives make application available for end users (nothing new to learn)
65. 13 . 1 WRAP UP
66. 13 . 2 Links https://kubernetes.io https://www.cncf.io/ https://www.openshift.com/promotions/kubernetes.html https://fabric8.io https://github.com/ramitsurana/awesome-kubernetes
67. 13 . 3 The End @NicolasByl Copyright 2017