Merge pull request #517 from ddps-lab/azure-collector-fix

Azure colloector 의 Lambda 로그중 불필요 SettingWithCopyWarning 문제 해결.

collector/spot-dataset/azure/lambda/current_collector/utils/compare_data.py

@@ -4,15 +4,18 @@
 # compare previous collected workload with current collected workload
 # return changed workload
 def compare(previous_df, current_df, workload_cols, feature_cols):
+    previous_df = previous_df.copy()
+    current_df = current_df.copy()
+
     previous_df['OndemandPrice'] = previous_df['OndemandPrice'].fillna(-1)
     current_df['OndemandPrice'] = current_df['OndemandPrice'].fillna(-1)
     previous_df['Savings'] = previous_df['Savings'].fillna(-1)
     current_df['Savings'] = current_df['Savings'].fillna(-1)
     previous_df['IF'] = previous_df['IF'].fillna(-1)
     current_df['IF'] = current_df['IF'].fillna(-1)
 
-    previous_df = previous_df.dropna(axis=0, inplace=False)
-    current_df = current_df.dropna(axis=0, inplace=False)
+    previous_df = previous_df.dropna(axis=0)
+    current_df = current_df.dropna(axis=0)
 
     previous_df.loc[:, 'Workload'] = previous_df[workload_cols].apply(lambda row: ':'.join(row.values.astype(str)), axis=1)
     previous_df.loc[:, 'Feature'] = previous_df[feature_cols].apply(lambda row: ':'.join(row.values.astype(str)), axis=1)
@@ -78,4 +81,4 @@ def compare(previous_df, current_df, workload_cols, feature_cols):
 
     current_df = current_df.loc[changed_indices].drop(['Workload', 'Feature'], axis=1)
 
-    return current_df
+    return current_df

collector/spot-dataset/azure/lambda/current_collector/utils/upload_data.py

@@ -40,6 +40,8 @@ def submit_batch(records, counter, recursive):
 # Check Database And Table Are Exist and Upload Data to Timestream
 def upload_timestream(data, time_datetime):
     data = data[['InstanceTier', 'InstanceType', 'Region', 'OndemandPrice', 'SpotPrice', 'IF']]
+    data = data.copy()
+
     data['OndemandPrice'] = data['OndemandPrice'].fillna(-1)
     data['SpotPrice'] = data['SpotPrice'].fillna(-1)
     data['IF'] = data['IF'].fillna(-1)
@@ -80,6 +82,8 @@ def upload_timestream(data, time_datetime):
 def update_latest(data, time_datetime):
     data['id'] = data.index + 1
     data = data[['id', 'InstanceTier', 'InstanceType', 'Region', 'OndemandPrice', 'SpotPrice', 'Savings', 'IF']]
+    data = data.copy()
+
     data['OndemandPrice'] = data['OndemandPrice'].fillna(-1)
     data['Savings'] = data['Savings'].fillna(-1)
     data['IF'] = data['IF'].fillna(-1)