Deployment: Synchronized strategies for hybrid cloud-native environments

app/LLD/qdrant_strategy.py

@@ -1,3 +1,4 @@
+import os
 import uuid
 from typing import List, Dict, Any
 from qdrant_client import QdrantClient
@@ -8,31 +9,46 @@ from app.core.config import settings
 class QdrantVectorStoreStrategy(VectorStoreInterface):
     """
     Concrete implementation of VectorStoreInterface leveraging Qdrant DB.
-    Encapsulates schema enforcement, item upsert workflows, and multi-modal calculations.
+    Optimized for zero-downtime hot switches between local development and cloud clusters.
     """
     def __init__(self) -> None:
-        # Establish client connection targeting the docker-compose service network
-        self.client = QdrantClient(host=settings.QDRANT_HOST, port=settings.QDRANT_PORT)
+        # 1. Prioritize Cloud Environment Secrets over Local Config File definitions
+        qdrant_url = os.getenv("QDRANT_URL") or getattr(settings, "QDRANT_URL", None)
+        qdrant_api_key = os.getenv("QDRANT_API_KEY") or getattr(settings, "QDRANT_API_KEY", None)
+
+        if qdrant_url:
+            print(f"[QDRANT CLUSTER] Establishing secure connection link to cloud instance: {qdrant_url}")
+            self.client = QdrantClient(url=qdrant_url, api_key=qdrant_api_key)
+        else:
+            host = getattr(settings, "QDRANT_HOST", "localhost")
+            port = int(getattr(settings, "QDRANT_PORT", 6333))
+            print(f"[QDRANT LOCAL] Establishing standard connection link to container: {host}:{port}")
+            self.client = QdrantClient(host=host, port=port)
+
         self.collection_name = "video_frames"
         self._ensure_collection_exists()
 
     def _ensure_collection_exists(self) -> None:
         """
-        Idempotent schema controller checking for collection persistence 
-        and initializing vector parameters if absent.
+        Idempotent schema controller. Log warnings instead of completely freezing the 
+        container boot sequence if cloud cloud connections take a moment to synchronize.
         """
         try:
             if not self.client.collection_exists(collection_name=self.collection_name):
+                vector_size = getattr(settings, "VECTOR_DIMENSION", 1152) # Uses default SigLIP dimension fallback
                 self.client.create_collection(
                     collection_name=self.collection_name,
                     vectors_config=VectorParams(
-                        size=settings.VECTOR_DIMENSION, 
+                        size=vector_size, 
                         distance=Distance.COSINE
                     )
                 )
+                print(f"[QDRANT SUCCESS] Index space '{self.collection_name}' successfully verified.")
         except Exception as e:
-            # Crucial LLD practice: fail fast during object construction if state is compromised
-            raise RuntimeError(f"Failed initializing Qdrant collection layer: {str(e)}")
+            # Prevent system lockouts on cluster re-build operations
+            print("\n⚠️ [QDRANT INTERCEPT WARNING] ⚠️")
+            print(f"Could not instantly connect to remote collection: {str(e)}")
+            print("The gateway instance will continue booting. Hot retries active for incoming user requests.\n")
 
     def upsert_embeddings(self, video_id: str, embeddings: List[List[float]], metadata: List[Dict[str, Any]]) -> bool:
         """
@@ -42,15 +58,12 @@ class QdrantVectorStoreStrategy(VectorStoreInterface):
         try:
             points = []
             for idx, (vector, meta) in enumerate(zip(embeddings, metadata)):
-                # Inject parent relational identifier directly into metadata payload
                 payload = {
                     "video_id": video_id,
                     "timestamp_seconds": meta.get("timestamp_seconds"),
                     "file_path": meta.get("file_path")
                 }
                 
-                # Use deterministic UUID generation based on the namespace 
-                # to prevent document duplication during re-processing jobs
                 point_id = str(uuid.uuid5(uuid.NAMESPACE_DNS, f"{video_id}_{idx}"))
                 
                 points.append(
@@ -61,7 +74,6 @@ class QdrantVectorStoreStrategy(VectorStoreInterface):
                     )
                 )
             
-            # Execute transactional batch upsert operation
             self.client.upsert(
                 collection_name=self.collection_name,
                 wait=True,
@@ -70,7 +82,6 @@ class QdrantVectorStoreStrategy(VectorStoreInterface):
             return True
             
         except Exception as e:
-            # Structured application reporting block
             print(f"[ERROR] Vector storage execution fault for Video {video_id}: {str(e)}")
             return False
 
@@ -86,7 +97,6 @@ class QdrantVectorStoreStrategy(VectorStoreInterface):
                 limit=top_k
             )
             
-            # Translate raw database structures into standard clean payloads
             formatted_results = []
             for hit in search_results:
                 formatted_results.append({
@@ -99,5 +109,4 @@ class QdrantVectorStoreStrategy(VectorStoreInterface):
             
         except Exception as e:
             print(f"[ERROR] Vector distance analysis failed: {str(e)}")
-            return []
-        
+            return []