🔧 refactor: add fallback to cache check for reliability

.gitea/workflows/ci-cd.yaml

@@ -76,22 +76,29 @@ jobs:
           echo "Dependency hash: $DEPS_HASH"
           echo "deps_hash=$DEPS_HASH" >> $GITHUB_OUTPUT
 
-      - name: Check for existing cache (optimized)
+      - name: Check for existing cache (optimized with fallback)
         id: check_cache
         run: |
-          # Check if image exists in registry using optimized approach
+          # Check if image exists in registry using optimized approach with fallback
           IMAGE_NAME="${{ env.CI_REGISTRY }}/${{ env.GITEA_ORG }}/${{ env.GITEA_REPO }}-build-cache:${{ steps.calculate_hash.outputs.deps_hash }}"
           
           # Fast check using docker manifest inspect (lighter than pull)
           echo "🔍 Checking cache: $IMAGE_NAME"
           
-          # Try manifest inspect first (fastest method)
+          # Try manifest inspect first (fastest method, but experimental)
           if docker manifest inspect "$IMAGE_NAME" >/dev/null 2>&1; then
             echo "✅ Cache hit - using existing build cache (manifest inspect)"
             echo "cache_hit=true" >> $GITHUB_OUTPUT
           else
-            echo "⚠️  Cache miss - will build new cache image"
-            echo "cache_hit=false" >> $GITHUB_OUTPUT
+            # Fallback to docker pull if manifest inspect fails (more reliable)
+            echo "⚠️  Manifest inspect failed, falling back to docker pull..."
+            if docker pull "$IMAGE_NAME" >/dev/null 2>&1; then
+              echo "✅ Cache hit - using existing build cache (fallback: docker pull)"
+              echo "cache_hit=true" >> $GITHUB_OUTPUT
+            else
+              echo "⚠️  Cache miss - will build new cache image"
+              echo "cache_hit=false" >> $GITHUB_OUTPUT
+            fi
           fi
 
       - name: Login to Gitea Container Registry