Merge branch 'postgres_vector_on_demand'

Author: danielaskdd

env.example

@@ -405,6 +405,10 @@ POSTGRES_MAX_CONNECTIONS=12
 ### POSTGRES_WORKSPACE=forced_workspace_name
 
 ### PostgreSQL Vector Storage Configuration
+### Enable/disable vector features (default: true for backward compatibility)
+### Set to false to disable pgvector extension and vector operations when using PostgreSQL
+### only for KV/Graph/DocStatus storage with a different vector backend (e.g., Milvus, Qdrant)
+POSTGRES_ENABLE_VECTOR=true
 ### Vector storage type: HNSW, IVFFlat, VCHORDRQ
 POSTGRES_VECTOR_INDEX_TYPE=HNSW
 POSTGRES_HNSW_M=16

lightrag/kg/postgres_impl.py

@@ -145,6 +145,12 @@ def __init__(self, config: dict[str, Any], **kwargs: Any):
         self.ssl_crl = config.get("ssl_crl")
 
         # Vector configuration
+        _ev = config.get("enable_vector", True)
+        self.enable_vector = (
+            _ev
+            if isinstance(_ev, bool)
+            else str(_ev).lower() in ("true", "1", "yes", "on")
+        )  # True for backward compatibility, can be set to False to disable vector features
         self.vector_index_type = config.get("vector_index_type")
         self.hnsw_m = config.get("hnsw_m")
         self.hnsw_ef = config.get("hnsw_ef")
@@ -332,32 +338,35 @@ async def _init_connection(connection: asyncpg.Connection) -> None:
             encode/decode vector columns, eliminating non-deterministic behavior
             where some connections have the codec and others don't.
             """
-            await register_vector(connection)
+            if self.enable_vector:
+                await register_vector(connection)
 
         async def _create_pool_once() -> None:
             # STEP 1: Bootstrap - ensure vector extension exists BEFORE pool creation.
             # On a fresh database, register_vector() in _init_connection will fail
             # if the vector extension doesn't exist yet, because the 'vector' type
             # won't be found in pg_catalog. We must create the extension first
             # using a standalone bootstrap connection.
-            bootstrap_conn = await asyncpg.connect(
-                user=self.user,
-                password=self.password,
-                database=self.database,
-                host=self.host,
-                port=self.port,
-                ssl=connection_params.get("ssl"),
-            )
-            try:
-                await self.configure_vector_extension(bootstrap_conn)
-            finally:
-                await bootstrap_conn.close()
+            # Skip this step if vector support is not enabled.
+            if self.enable_vector:
+                bootstrap_conn = await asyncpg.connect(
+                    user=self.user,
+                    password=self.password,
+                    database=self.database,
+                    host=self.host,
+                    port=self.port,
+                    ssl=connection_params.get("ssl"),
+                )
+                try:
+                    await self.configure_vector_extension(bootstrap_conn)
+                finally:
+                    await bootstrap_conn.close()
 
             # STEP 2: Now safe to create pool with register_vector callback.
-            # The vector extension is guaranteed to exist at this point.
+            # The vector extension is guaranteed to exist at this point (if enabled).
             pool = await asyncpg.create_pool(
                 **connection_params,
-                init=_init_connection,  # Register pgvector codec on every connection
+                init=_init_connection,  # Register pgvector codec on every connection (if enabled)
             )  # type: ignore
             self.pool = pool
 
@@ -464,7 +473,7 @@ async def _run_with_retry(
                         await self.configure_age(connection, graph_name)
                     elif with_age and not graph_name:
                         raise ValueError("Graph name is required when with_age is True")
-                    if self.vector_index_type == "VCHORDRQ":
+                    if self.enable_vector and self.vector_index_type == "VCHORDRQ":
                         await self.configure_vchordrq(connection)
                     return await operation(connection)
 
@@ -1750,6 +1759,12 @@ def get_config() -> dict[str, Any]:
                 "POSTGRES_SSL_CRL",
                 config.get("postgres", "ssl_crl", fallback=None),
             ),
+            # Vector configuration
+            "enable_vector": os.environ.get(
+                "POSTGRES_ENABLE_VECTOR",
+                config.get("postgres", "enable_vector", fallback="true"),
+            ).lower()
+            in ("true", "1", "yes", "on"),
             "vector_index_type": os.environ.get(
                 "POSTGRES_VECTOR_INDEX_TYPE",
                 config.get("postgres", "vector_index_type", fallback="HNSW"),
@@ -2828,6 +2843,11 @@ async def initialize(self):
                 # Use "default" for compatibility (lowest priority)
                 self.workspace = "default"
 
+            if not self.db.enable_vector:
+                raise ValueError(
+                    "Cannot use PGVectorStorage when POSTGRES_ENABLE_VECTOR=false. Configure an alternative vector backend."
+                )
+
             # Setup table (create if not exists and handle migration)
             await PGVectorStorage.setup_table(
                 self.db,