chore: make health checks robuster

Author: scottgerring

.github/workflows/docker-compose-integration.yml

@@ -24,8 +24,32 @@ jobs:
     - name: Wait for services to be healthy
       run: |
         echo "Waiting for all services to be healthy..."
-        # Wait for services with health checks to be healthy (4 services) and all services to be up
-        timeout 300 bash -c 'until [ $(docker compose ps --format "{{.Health}}" | grep -c "healthy") -ge 4 ] && [ $(docker compose ps --format "{{.Status}}" | grep -c "Up") -ge 5 ]; do sleep 5; done'
+        # Wait for services with health checks to be healthy (7 services) and all services to be up
+        # Increased timeout to 10 minutes for CI environment
+        timeout 600 bash -c 'while true; do
+          echo "=== Detailed Service Status ==="
+          docker compose ps --format "table {{.Name}}\t{{.Status}}\t{{.Health}}"
+          echo "=== Unhealthy Services ==="
+          unhealthy_services=$(docker compose ps --format "{{.Name}} {{.Health}}" | grep -v healthy | grep -v "^[[:space:]]*$" || echo "")
+          if [ -n "$unhealthy_services" ]; then
+            echo "$unhealthy_services"
+            echo "=== Logs for unhealthy services ==="
+            for service in $(echo "$unhealthy_services" | awk "{print \$1}"); do
+              echo "--- Logs for $service ---"
+              docker compose logs --tail=10 "$service" || echo "No logs for $service"
+            done
+          else
+            echo "All services healthy or no health check"
+          fi
+          healthy_count=$(docker compose ps --format "{{.Health}}" | grep -c "healthy" || echo "0")
+          up_count=$(docker compose ps --format "{{.Status}}" | grep -c "Up" || echo "0")
+          echo "Progress: $healthy_count/7 healthy, $up_count services up"
+          if [ "$healthy_count" -ge 7 ] && [ "$up_count" -ge 7 ]; then
+            echo "All services are healthy!"
+            break
+          fi
+          sleep 10
+        done'
         docker compose ps
       
     - name: Test dashboard endpoints
@@ -45,15 +69,12 @@ jobs:
       run: |
         echo "=== Docker Compose Services Status ==="
         docker compose ps
-        echo "=== Traefik Logs ==="
-        docker compose logs traefik
-        echo "=== Sticker Award Logs ==="
-        docker compose logs sticker-award
-        echo "=== User Management Logs ==="
-        docker compose logs user-management
-        echo "=== Database Logs ==="
-        docker compose logs sticker-award-db
-        docker compose logs user-management-db
+        echo "=== All Service Logs ==="
+        docker compose logs --tail=50
+        echo "=== Health Check Details ==="
+        docker compose ps --format "table {{.Name}}\t{{.Status}}\t{{.Health}}"
+        echo "=== Resource Usage ==="
+        docker stats --no-stream
       
     - name: Cleanup
       if: always()

docker-compose.yml

@@ -185,6 +185,11 @@ services:
       QUARKUS_S3_AWS_CREDENTIALS_STATIC_PROVIDER_SECRET_ACCESS_KEY: minioadmin
       QUARKUS_S3_PATH_STYLE_ACCESS: "true"
       STICKER_IMAGES_BUCKET: sticker-images
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
+      interval: 10s
+      timeout: 5s
+      retries: 5
     labels:
       - "traefik.enable=true"
       - "traefik.http.routers.sticker-award.rule=PathPrefix(`/api/award`)"
@@ -210,6 +215,11 @@ services:
       KAFKA__BOOTSTRAPSERVERS: "redpanda:9092"
       KAFKA__SCHEMAREGISTRY: "http://redpanda:8082"
       KAFKA__GROUPID: "stickerlandia-user-management"
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8080/api/users/v1/health"]
+      interval: 10s
+      timeout: 5s
+      retries: 5
     labels:
       - "traefik.enable=true"
       - "traefik.http.routers.user-management.rule=PathPrefix(`/api/users`)"

user-management/src/Stickerlandia.UserManagement.Api/Dockerfile

@@ -10,9 +10,10 @@ RUN dotnet restore src/Stickerlandia.UserManagement.Api/Stickerlandia.UserManage
 COPY . ./
 
 # Publish the application
-RUN dotnet publish src/Stickerlandia.UserManagement.Api/Stickerlandia.UserManagement.Api.csproj -o out -c Release -r linux-arm64
+RUN dotnet publish src/Stickerlandia.UserManagement.Api/Stickerlandia.UserManagement.Api.csproj -o out -c Release
 
-FROM mcr.microsoft.com/dotnet/aspnet:8.0-noble-chiseled AS runtime
+FROM mcr.microsoft.com/dotnet/aspnet:8.0 AS runtime
+RUN apt-get update && apt-get install -y curl && rm -rf /var/lib/apt/lists/*
 WORKDIR /App
 COPY --from=build /App/out .
 EXPOSE 80