Skip to main content

How to deploy to dev

Deploy your changes to the AKS aucert-dev namespace for testing.

Prerequisites​

  • az login authenticated
  • kubectl configured with AKS context (az aks get-credentials --resource-group aucert-rg --name aucert-aks)
  • Docker running locally
  • ACR login: az acr login --name aucertacr41e0x5

Steps​

Step 1: Build Docker image​

docker build \
  -t aucertacr41e0x5.azurecr.io/backend:latest \
  -f backend/platform/Dockerfile \
  backend/platform/
tip

For faster builds during development, use the --target dev stage if the Dockerfile has a multi-stage build. For production-like testing, omit --target to build the full image.

Step 2: Push to ACR​

docker push aucertacr41e0x5.azurecr.io/backend:latest

Verify the image was pushed:

az acr repository show-tags --name aucertacr41e0x5 --repository backend --output table

Step 3: Deploy via Helm​

helm upgrade --install aucert-dev k8s/charts/aucert \
  -n aucert-dev \
  -f k8s/charts/aucert/values/dev.yaml

This applies the dev values overlay, which sets:

  • Replica count: 1
  • Resource limits: low (for cost)
  • Image pull policy: Always (to pick up :latest changes)
  • Dev-specific ConfigMaps and secrets

Step 4: Verify deployment​

# Check pod status (should show Running within ~60s)
kubectl get pods -n aucert-dev

# Watch pod startup
kubectl get pods -n aucert-dev -w

# Check recent logs
kubectl logs -n aucert-dev -l app=backend --tail=50

Expected: Pods transition from ContainerCreating → Running. Logs show Started AucertPlatformApplication.

Step 5: Smoke test​

# Port-forward the service
kubectl port-forward -n aucert-dev svc/aucert-backend 8080:8080

# Hit the health endpoint
curl http://localhost:8080/health
# Expected: {"status": "UP"}

Deploying frontend changes​

The frontend deploys as a separate container:

# Build
docker build \
  -t aucertacr41e0x5.azurecr.io/frontend:latest \
  -f frontend/apps/console/Dockerfile \
  frontend/apps/console/

# Push
docker push aucertacr41e0x5.azurecr.io/frontend:latest

# Deploy (same Helm chart, frontend is a separate deployment)
helm upgrade --install aucert-dev k8s/charts/aucert \
  -n aucert-dev \
  -f k8s/charts/aucert/values/dev.yaml

Rolling back​

If a deployment breaks, roll back to the previous version:

# Check rollout history
helm history aucert-dev -n aucert-dev

# Roll back to previous release
helm rollback aucert-dev -n aucert-dev

# Or roll back to a specific revision
helm rollback aucert-dev 3 -n aucert-dev

Troubleshooting​

ImagePullBackOff

The pod can't pull the image from ACR. Fix:

  1. Re-authenticate: az acr login --name aucertacr41e0x5
  2. Verify image exists: az acr repository show-tags --name aucertacr41e0x5 --repository backend
  3. Check AKS has ACR pull permission: az aks check-acr --resource-group aucert-rg --name aucert-aks --acr aucertacr41e0x5
CrashLoopBackOff

The pod starts but crashes immediately. Debug:

# Check logs from the crashed container
kubectl logs -n aucert-dev <pod-name> --previous

# Common causes:
# - Missing environment variables → check ConfigMap/Secret
# - Database connection failed → check port-forward or VNet connectivity
# - OOM kill → check resource limits in values/dev.yaml
Pods stuck in Pending

Usually a resource issue:

# Check events
kubectl describe pod -n aucert-dev <pod-name>

# Common causes:
# - Insufficient CPU/memory → scale down replicas or increase node pool
# - PVC not bound → check storage class and PV availability
Helm upgrade fails with "another operation in progress"

A previous Helm operation didn't clean up. Fix:

# Check Helm release status
helm status aucert-dev -n aucert-dev

# If stuck in "pending-upgrade" or "pending-install"
helm rollback aucert-dev -n aucert-dev

What's next​