deploy-prod

name	deploy-prod
description	Deploy services to the prod EKS namespace via ArgoCD (manual sync) with safety checks
disable-model-invocation	true
argument-hint	[service|all]

/deploy-prod [service|all]

Deploy services to the petclinic-prod namespace with extra safety checks.

Arguments

• service — Specific service to deploy (e.g., config-server, api-gateway) or all
• Default: all

Steps

1. Pre-deployment checks:

   • Verify kubectl context is the prod cluster:

     kubectl config current-context
     

   • Check if ArgoCD is installed:

     kubectl get namespace argocd 2>/dev/null
     

2. Explicit confirmation required: Ask the user: "You are deploying to PRODUCTION (petclinic-prod). This will affect live services. Confirm?" Do NOT proceed without explicit "yes" from the user.

3. If ArgoCD is installed (standard path):

   • Show what ArgoCD will sync (diff):

     argocd app diff {service}-prod || true
     

   • all: Sync services one at a time (sequential, not bulk):

     # Startup order: config-server first, then discovery-server, then rest
     argocd app sync config-server-prod && argocd app wait config-server-prod --timeout 180
     argocd app sync discovery-server-prod && argocd app wait discovery-server-prod --timeout 180
     # Then remaining services one at a time
     for svc in api-gateway customers-service visits-service vets-service genai-service admin-server; do
       argocd app sync ${svc}-prod && argocd app wait ${svc}-prod --timeout 180
     done
     

   • specific service: Sync that service:

     argocd app sync {service}-prod
     argocd app wait {service}-prod --timeout 180
     

   • If a service fails, STOP and report — do not continue to the next service.

4. If ArgoCD is NOT installed (bootstrap path):

   • Apply namespace (idempotent):

     kubectl apply -f k8s/base/namespaces.yaml
     

   • Sequential deployment (one service at a time, verify each):

     helm upgrade --install {service} helm/petclinic-service/ \
       -f helm-values/{service}.yaml \
       -f helm-values/prod.yaml \
       -n petclinic-prod --create-namespace
     kubectl rollout status deployment/{service} -n petclinic-prod --timeout=180s
     

   • If a service fails, STOP and report — do not continue to the next service.

5. Startup order (when deploying all):

   1. config-server → wait for ready
   2. discovery-server → wait for ready
   3. All remaining services (one at a time)

6. Post-deployment verification:

   kubectl get pods -n petclinic-prod -o wide
   kubectl get svc -n petclinic-prod
   

7. Show a deployment summary:

   • Services deployed
   • Pod status (Running/Pending/CrashLoopBackOff)
   • Any issues encountered

Important

• PRODUCTION deployment — extra caution required
• ArgoCD prod apps require manual sync (no auto-sync)
• Always show diff before syncing
• Always get explicit user confirmation
• Deploy sequentially, verify each service before the next
• If any service fails, STOP immediately — do not cascade failures
• Suggest running /smoke-test prod after successful deployment