refine

msr.py

@@ -284,16 +284,31 @@ def get_duckdb_connection():
     """
     Initialize DuckDB connection with OPTIMIZED memory settings.
     Uses persistent database and reduced memory footprint.
+    Automatically removes cache file if lock conflict is detected.
     """
-    conn = duckdb.connect(DUCKDB_CACHE_FILE)
+    try:
+        conn = duckdb.connect(DUCKDB_CACHE_FILE)
+    except Exception as e:
+        # Check if it's a locking error
+        error_msg = str(e)
+        if "lock" in error_msg.lower() or "conflicting" in error_msg.lower():
+            print(f"   ⚠ Lock conflict detected, removing {DUCKDB_CACHE_FILE}...")
+            if os.path.exists(DUCKDB_CACHE_FILE):
+                os.remove(DUCKDB_CACHE_FILE)
+                print(f"   ✓ Cache file removed, retrying connection...")
+            # Retry connection after removing cache
+            conn = duckdb.connect(DUCKDB_CACHE_FILE)
+        else:
+            # Re-raise if it's not a locking error
+            raise
 
     # OPTIMIZED SETTINGS
     conn.execute(f"SET threads TO {DUCKDB_THREADS};")
     conn.execute("SET preserve_insertion_order = false;")
     conn.execute("SET enable_object_cache = true;")
     conn.execute("SET temp_directory = '/tmp/duckdb_temp';")
-    conn.execute(f"SET memory_limit = '{DUCKDB_MEMORY_LIMIT}';")
-    conn.execute(f"SET max_memory = '{DUCKDB_MEMORY_LIMIT}';")
+    conn.execute(f"SET memory_limit = '{DUCKDB_MEMORY_LIMIT}';")  # Per-query limit
+    conn.execute(f"SET max_memory = '{DUCKDB_MEMORY_LIMIT}';")  # Hard cap
 
     return conn