fix

msr.py

@@ -51,8 +51,8 @@ PATCH_WANTED_LABELS = [
 GIT_SYNC_TIMEOUT = 300  # 5 minutes timeout for git pull
 
 # OPTIMIZED DUCKDB CONFIGURATION
-DUCKDB_THREADS = 8
-DUCKDB_MEMORY_LIMIT = "64GB"
+DUCKDB_THREADS = 16
+DUCKDB_MEMORY_LIMIT = "128GB"
 
 # Streaming batch configuration
 BATCH_SIZE_DAYS = 7  # Process 1 week at a time (~168 hourly files)
@@ -68,7 +68,7 @@ UPLOAD_DELAY_SECONDS = 5
 UPLOAD_MAX_BACKOFF = 3600
 
 # Scheduler configuration
-SCHEDULE_ENABLED = True
+SCHEDULE_ENABLED = False
 SCHEDULE_DAY_OF_WEEK = 'fri'  # Friday
 SCHEDULE_HOUR = 0
 SCHEDULE_MINUTE = 0
@@ -1294,7 +1294,6 @@ def mine_all_agents():
         wanted_open = results['wanted_open']
         wanted_resolved = results['wanted_resolved']
         agent_discussions = results['agent_discussions']
-
     except Exception as e:
         print(f"Error during DuckDB fetch: {str(e)}")
         traceback.print_exc()
@@ -1315,10 +1314,17 @@ def mine_all_agents():
         save_leaderboard_data_to_hf(
             leaderboard_dict, issue_monthly_metrics, wanted_open, discussion_monthly_metrics
         )
-
     except Exception as e:
         print(f"Error saving leaderboard: {str(e)}")
         traceback.print_exc()
+    finally:
+        # Clean up DuckDB cache file to save storage
+        if os.path.exists(DUCKDB_CACHE_FILE):
+            try:
+                os.remove(DUCKDB_CACHE_FILE)
+                print(f"   ✓ Cache file removed: {DUCKDB_CACHE_FILE}")
+            except Exception as e:
+                print(f"   ⚠ Failed to remove cache file: {str(e)}")
 
 
 # =============================================================================