What Devs Should Consider, When Coding Applications 

● Introduction ● Context ● Challenges and Approaches 

Thomas Philipona • CTO @ Puzzle ITC • OpenSource • Responsible for our CI/CD Processes • APPUiO – Swiss Container Platform @tphilipona phil-pona 

www.puzzle.ch @puzzleitc | | 

Context 

Container Hype ©Håkan Dahlström Source: https://www.flickr.com/photos/dahlstroms/3144199355 Container Hype Docker | Rocket | Dev - driven | Ecosystem 

Docker in Production Monitoring Log server Storage Jenkins Mail server DB Server Backup Network Lifecycle Container u. Infrastruktur 

How it all began Multiple Workloads | OpenShift 3 | Internal use | Platform to run our Apps 

Our Container Platform Cloudscale IaaS ✓ ISO 27001 ✓ FINMA ✓ -based ✓ 

Docker Kubernetes APPUiO is based on 

Our Application Stack Angular 2 Client DB External Services Wildfly 10 Java EE 7 . . . 

Challenges and Approaches 

Dynamic Container Platform Restarts of nodes due to maintenance Container gets moved due to ressource allocation → Start up time of the Application → Robustness Reconnect of Backends → at least two instances running 

Start Up Time of Applications 17:11:38,607 INFO [org.jboss.as] (MSC service thread 1-7) WFLYSRV0049: WildFly Full 10.0.0.Final (WildFly Core 2.0.10.Final) starting ... ... 17:13:10,297 INFO [org.jboss.as] (Controller Boot Thread) WFLYSRV0025: WildFly Full 10.0.0.Final (WildFly Core 2.0.10.Final) started in 93046ms - Started 1235 of 1510 services (409 services are lazy, passive or on-demand) 

Scalability Stateless architecture No session pinning (Sessions in the database or at the client) No local filesystem → Easier testable . . . 

Avoid Long Running Tasks Avoid them completely (if possible) Separate processes / applications running along the application Split them into parts, allow continuation t . . . 

Monitoring / Application Healthchecks View into platform | fast feedback | easily integrated into infrastructure 

Monitoring / Healthchecks Liveness check /health Process is running If not ok, process gets restarted Readiness check /health/ready Deployment ok Ready for requests Fully functional Health checks as simple JSON response on a REST endpoint 

Cascading Checks app DB backend GET /health GET /health select 1 from ... HTTP/1.1 200 OK { "status": "UP", "db": { "status": "UP" }, "backend": { "status": "UP" } } 

Cascading Errors app DB backend GET /health GET /health select 1 from ... HTTP/1.1 299 Warning { "status": "UP", "db": { "status": "DOWN" }, "backend": { "status": "UP" } } 

Advantages Easily integratable into external monitoring and alarming Kubernetes / OpenShift uses those heath checks 

Automation and CI/CD Git code Compile, Unit-tests Integration Tests, Package Deploy to Dev Acceptance Tests Deploy to UAT env Manual Tests Deploy to Prod 

keep everything you need to build, deploy, test, & release in version control 

Groovy Jenkins Pipeline Scripts stage 'Build' sh "${mvnHome}/bin/mvn versions:set -DnewVersion=${buildVersion}" sh "${mvnHome}/bin/mvn clean install" // prepare Dockerfile and Context sh "tar cvfz docker-context.tar.gz Dockerfile target/javaee7-wildfly-example.war" archive 'docker-context.tar.gz' 

If It Hurts, Do It More Frequently, and Bring the Pain Forward Source: Jez Humble, David Farley, Continuous Delivery, Pearson Education, Inc. 2011 

Automated Database Migrations DB migrationscripts are part of the deployed application and run during startup or deployment. $ php bin/console doctrine:schema:update $ rake db:migrate 

Conclusion 

Heterogeneous Team Devs | Ops | User Experience | Product Owner 

The Twelve-Factor App Framework / Manifest on how to build Software-as-a-Service Apps • To max the portability and decrease vendor lockin • Deployment on cloud-platforms • Ensure continuous deployment • Scalability https:/ /12factor.net/ 

No content