Merge pull request #244 from chroma-core/fixDuckDbLogging

create persist dir if not exist, correct logging of persistence

chromadb/__init__.py

@@ -37,12 +37,15 @@ def require(key):
         return chromadb.db.clickhouse.Clickhouse(settings)
     elif setting == "duckdb+parquet":
         require("persist_directory")
+        logger.warning(
+            f"Using embedded DuckDB with persistence: data will be stored in: {settings.persist_directory}"
+        )
         import chromadb.db.duckdb
 
         return chromadb.db.duckdb.PersistentDuckDB(settings)
     elif setting == "duckdb":
         require("persist_directory")
-        logger.info("Using DuckDB in-memory for database. Data will be transient.")
+        logger.warning("Using embedded DuckDB without persistence: data will be transient")
         import chromadb.db.duckdb
 
         return chromadb.db.duckdb.DuckDB(settings)

chromadb/db/duckdb.py

@@ -14,6 +14,7 @@
 import duckdb
 import uuid
 import time
+import os
 import itertools
 import logging
 
@@ -44,8 +45,6 @@ class DuckDB(Clickhouse):
     # duckdb has a different way of connecting to the database
     def __init__(self, settings):
 
-        logger.warning("Using embedded DuckDB without persistence: data will be transient")
-
         self._conn = duckdb.connect()
         self._create_table_collections()
         self._create_table_embeddings()
@@ -391,6 +390,9 @@ def persist(self):
         if self._conn is None:
             return
 
+        if not os.path.exists(self._save_folder):
+            os.makedirs(self._save_folder)
+
         # if the db is empty, dont save
         if self._conn.query(f"SELECT COUNT() FROM embeddings") == 0:
             return
@@ -417,7 +419,8 @@ def load(self):
         """
         Load the database from disk
         """
-        import os
+        if not os.path.exists(self._save_folder):
+            os.makedirs(self._save_folder)
 
         # load in the embeddings
         if not os.path.exists(f"{self._save_folder}/chroma-embeddings.parquet"):