fix

msr.py

@@ -37,8 +37,8 @@ LEADERBOARD_TIME_FRAME_DAYS = 180
 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)
@@ -54,7 +54,7 @@ UPLOAD_DELAY_SECONDS = 5
 UPLOAD_MAX_BACKOFF = 3600
 
 # Scheduler configuration
-SCHEDULE_ENABLED = True
+SCHEDULE_ENABLED = False
 SCHEDULE_DAY_OF_WEEK = 'mon'  # Monday
 SCHEDULE_HOUR = 0
 SCHEDULE_MINUTE = 0
@@ -956,7 +956,6 @@ def mine_all_agents():
         # Separate commits and PRs
         commit_metadata = results['commits']
         pr_metadata = results['prs']
-
     except Exception as e:
         print(f"Error during DuckDB fetch: {str(e)}")
         traceback.print_exc()
@@ -984,10 +983,17 @@ def mine_all_agents():
         save_leaderboard_data_to_hf(
             leaderboard_dict, commit_monthly_metrics, pr_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)}")
 
 
 # =============================================================================