ArgoCD High Availability Setup
What you'll learn: How to deploy ArgoCD with high availability to eliminate single points of failure in production GitOps environments.
Why High Availability Matters
In production, your GitOps control plane can't afford downtime. If ArgoCD goes down, deployments stop, sync operations fail, and your GitOps workflow breaks. High Availability (HA) ensures ArgoCD remains operational even when individual components fail—like having backup generators for critical infrastructure.
Core HA Components
Multiple Replicas
ArgoCD consists of three main components you should scale:
- argocd-server: The API/UI server should run 2+ replicas behind a load balancer
- argocd-repo-server: Handles Git repository operations; scale to 2+ replicas for load distribution
- argocd-application-controller: Monitors cluster state; typically runs with 1 active replica (uses leader election)
Redis High Availability
ArgoCD uses Redis for caching. In HA mode, deploy Redis in sentinel or cluster mode with:
- Redis master and replica pods
- Redis Sentinel for automatic failover
- Persistent volumes to survive pod restarts
Resource Limits and Requests
Production setups need proper resource allocation:
resources:
requests:
memory: "256Mi"
cpu: "100m"
limits:
memory: "512Mi"
cpu: "500m"
Setting requests ensures scheduling guarantees, while limits prevent resource exhaustion that could cascade across nodes.
Additional HA Considerations
- Pod Anti-Affinity: Spread replicas across different nodes and availability zones
- Health Checks: Configure liveness and readiness probes
- Persistent Storage: Use persistent volumes for Redis and repository cache
- Monitoring: Track sync failures, API latency, and repository operations
Key Takeaway: ArgoCD HA requires running multiple replicas of server and repo-server components, implementing Redis HA with sentinel or clustering, and setting appropriate resource limits—ensuring your GitOps control plane remains resilient against component and infrastructure failures.