60% More Releases
Technical Product Manager
Implemented ArgoCD/Flux. Deploys: hours → 5 min. 60% more frequent releases
TL;DR
Led GitOps migration at Getir using ArgoCD and Flux. Reduced deployment time from hours to 5 minutes and increased release frequency by 60%.
Context
Getir's manual deployment processes couldn't keep pace with growth. Deployments took hours, were inconsistent across teams, and had no drift detection. Engineers were afraid to deploy because rollbacks were painful.
Problem
- Slow deployments taking hours
- Inconsistent deployment processes across teams
- No drift detection—configuration drift went unnoticed
- Manual processes prone to human error
- Engineers afraid to deploy frequently
What I Did
- Assessment: Mapped existing deployment workflows and pain points
- Tool Selection: Evaluated GitOps tools (ArgoCD, Flux, Spinnaker) against requirements
- Architecture Design: Designed GitOps workflow with ArgoCD for apps and Flux for infrastructure
- Rollout: Phased migration starting with willing teams
Key Decisions
GitOps-First
Declarative infrastructure as the single source of truth. No more manual kubectl commands or SSH deployments.
ArgoCD + Flux Combination
ArgoCD for application deployments (better UI, easier adoption), Flux for infrastructure (better Terraform integration).
Automatic Drift Detection
Continuous reconciliation ensures actual state matches desired state. Drift is automatically corrected.
Technical Details
- ArgoCD: Application deployment with automatic sync and rollback
- Flux: Infrastructure deployment with Terraform controller
- Git as Source of Truth: All configuration in version control
- Automatic Rollback: Failed deployments automatically roll back to last known good state
Results
- Consistent security configurations across all services
- Engineers deploy confidently knowing rollback is automatic
- Audit trail for every configuration change
Lessons Learned
- GitOps requires culture change: Technical implementation is easy; changing how teams think about deployments takes time.
- Start with willing teams: Early adopters prove value and become advocates.
- Automatic rollback builds confidence: Engineers deploy more often when they trust the safety net.
- Drift detection catches issues early: Problems that used to cause incidents are now caught immediately.