Spaces:
Running
Running
Fix: Prevent data wipe in initialize and set correct port
Browse files- src/vector_db.py +56 -31
src/vector_db.py
CHANGED
|
@@ -22,47 +22,72 @@ class UnifiedQdrant:
|
|
| 22 |
url = os.getenv("QDRANT_URL", ":memory:")
|
| 23 |
api_key = os.getenv("QDRANT_API_KEY", None)
|
| 24 |
print(f"Connecting to Qdrant at {url}...")
|
| 25 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
self.is_local = url == ":memory:" or not url.startswith("http")
|
|
|
|
| 28 |
if self.is_local:
|
| 29 |
-
print("
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
|
| 31 |
-
|
| 32 |
-
if self.client.collection_exists(self.collection_name):
|
| 33 |
-
self.client.delete_collection(self.collection_name)
|
| 34 |
-
|
| 35 |
-
# Try to create collection with full clusters
|
| 36 |
-
try:
|
| 37 |
-
self._create_collection_and_shards(self.num_clusters)
|
| 38 |
-
print(f"Successfully created collection with {self.num_clusters} clusters.")
|
| 39 |
-
except Exception as e:
|
| 40 |
-
print(f"Failed to create {self.num_clusters} clusters: {e}")
|
| 41 |
-
print("Attempting fallback to 8 clusters (Free Tier limit mitigation)...")
|
| 42 |
-
# Fallback 1: 8 Clusters
|
| 43 |
try:
|
| 44 |
-
self.num_clusters = 8
|
| 45 |
-
if self.client.collection_exists(self.collection_name):
|
| 46 |
-
self.client.delete_collection(self.collection_name)
|
| 47 |
self._create_collection_and_shards(self.num_clusters)
|
| 48 |
-
print(f"
|
| 49 |
-
except Exception as
|
| 50 |
-
print(f"Failed to create
|
| 51 |
-
print("
|
| 52 |
-
# Fallback
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
self.client.
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
|
| 63 |
def _create_collection_and_shards(self, n_clusters):
|
| 64 |
print(f"Creating collection '{self.collection_name}' with custom sharding ({n_clusters} clusters)...")
|
| 65 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
if self.is_local:
|
| 67 |
# Local mode doesn't support sharding_method=CUSTOM
|
| 68 |
self.client.create_collection(
|
|
|
|
| 22 |
url = os.getenv("QDRANT_URL", ":memory:")
|
| 23 |
api_key = os.getenv("QDRANT_API_KEY", None)
|
| 24 |
print(f"Connecting to Qdrant at {url}...")
|
| 25 |
+
|
| 26 |
+
# Relaxed connection settings for HF Spaces
|
| 27 |
+
port = 443 if url.startswith("https") else 6333
|
| 28 |
+
self.client = QdrantClient(
|
| 29 |
+
location=url,
|
| 30 |
+
port=port,
|
| 31 |
+
api_key=api_key,
|
| 32 |
+
timeout=60,
|
| 33 |
+
check_compatibility=False,
|
| 34 |
+
verify=False # Passed to httpx
|
| 35 |
+
)
|
| 36 |
|
| 37 |
self.is_local = url == ":memory:" or not url.startswith("http")
|
| 38 |
+
|
| 39 |
if self.is_local:
|
| 40 |
+
print("Running in local/memory mode. Custom Sharding is NOT supported. Simulating behavior.")
|
| 41 |
+
self.num_clusters = 1
|
| 42 |
+
if self.client.collection_exists(collection_name=self.collection_name):
|
| 43 |
+
self.client.delete_collection(collection_name=self.collection_name)
|
| 44 |
+
self.client.create_collection(
|
| 45 |
+
collection_name=self.collection_name,
|
| 46 |
+
vectors_config=VectorParams(size=self.vector_size, distance=Distance.COSINE)
|
| 47 |
+
)
|
| 48 |
+
print(f"Created standard collection '{self.collection_name}'.")
|
| 49 |
+
else:
|
| 50 |
+
# Check if exists first to avoid accidental deletion
|
| 51 |
+
if self.client.collection_exists(self.collection_name):
|
| 52 |
+
print(f"Collection '{self.collection_name}' already exists. Skipping initialization.")
|
| 53 |
+
return
|
| 54 |
|
| 55 |
+
# Try to create collection with full clusters
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
try:
|
|
|
|
|
|
|
|
|
|
| 57 |
self._create_collection_and_shards(self.num_clusters)
|
| 58 |
+
print(f"Successfully created collection with {self.num_clusters} clusters.")
|
| 59 |
+
except Exception as e:
|
| 60 |
+
print(f"Failed to create {self.num_clusters} clusters: {e}")
|
| 61 |
+
print("Attempting fallback to 8 clusters (Free Tier limit mitigation)...")
|
| 62 |
+
# Fallback 1: 8 Clusters
|
| 63 |
+
try:
|
| 64 |
+
self.num_clusters = 8
|
| 65 |
+
if self.client.collection_exists(self.collection_name):
|
| 66 |
+
self.client.delete_collection(self.collection_name)
|
| 67 |
+
self._create_collection_and_shards(self.num_clusters)
|
| 68 |
+
print(f"Fallback successful: Created collection with {self.num_clusters} clusters.")
|
| 69 |
+
except Exception as e2:
|
| 70 |
+
print(f"Failed to create 8 clusters: {e2}")
|
| 71 |
+
print("CRITICAL: Custom Sharding not supported. Falling back to Standard Collection (No Sharding).")
|
| 72 |
+
# Fallback 2: Standard Collection
|
| 73 |
+
self.num_clusters = 1
|
| 74 |
+
if self.client.collection_exists(self.collection_name):
|
| 75 |
+
self.client.delete_collection(self.collection_name)
|
| 76 |
+
|
| 77 |
+
self.client.create_collection(
|
| 78 |
+
collection_name=self.collection_name,
|
| 79 |
+
vectors_config=VectorParams(size=self.vector_size, distance=Distance.COSINE)
|
| 80 |
+
)
|
| 81 |
+
print("Fallback successful: Created Standard Collection.")
|
| 82 |
|
| 83 |
def _create_collection_and_shards(self, n_clusters):
|
| 84 |
print(f"Creating collection '{self.collection_name}' with custom sharding ({n_clusters} clusters)...")
|
| 85 |
|
| 86 |
+
if self.client.collection_exists(self.collection_name):
|
| 87 |
+
print(f"Collection '{self.collection_name}' already exists. Skipping creation.")
|
| 88 |
+
return
|
| 89 |
+
|
| 90 |
+
|
| 91 |
if self.is_local:
|
| 92 |
# Local mode doesn't support sharding_method=CUSTOM
|
| 93 |
self.client.create_collection(
|