Add flower logger similar to the worker

exareme2/algorithms/flower/__init__.py

@@ -0,0 +1,23 @@
+import logging
+import os
+
+from flwr.common.logger import FLOWER_LOGGER
+
+for handler in FLOWER_LOGGER.handlers:
+    FLOWER_LOGGER.removeHandler(handler)
+
+FLOWER_LOGGER.setLevel(logging.DEBUG)
+
+request_id = os.getenv("REQUEST_ID", "NO-REQUEST_ID")
+worker_role = os.getenv("WORKER_ROLE", "NO-ROLE")
+worker_identifier = os.getenv("WORKER_IDENTIFIER", "NO-IDENTIFIER")
+
+flower_formatter = logging.Formatter(
+    f"%(asctime)s - %(levelname)s - FLOWER - {worker_role} - {worker_identifier} - %(module)s - %(funcName)s(%(lineno)d) - {request_id} - %(message)s"
+)
+
+# Configure console logger
+console_handler = logging.StreamHandler()
+console_handler.setLevel(logging.DEBUG)
+console_handler.setFormatter(flower_formatter)
+FLOWER_LOGGER.addHandler(console_handler)

exareme2/algorithms/flower/flower_data_processing.py

@@ -7,6 +7,7 @@
 import pandas as pd
 import pymonetdb
 import requests
+from flwr.common.logger import FLOWER_LOGGER
 from pydantic import BaseModel
 from sklearn import preprocessing
 from sklearn.impute import SimpleImputer
@@ -53,7 +54,6 @@ def _fetch_data_from_db(data_model, datasets) -> pd.DataFrame:
 
 def _fetch_data_from_csv(data_model, datasets) -> pd.DataFrame:
     data_folder = Path(f"{os.getenv('DATA_PATH')}/{data_model.split(':')[0]}_v_0_1")
-    print(f"Loading data from folder: {data_folder}")
     dataframes = [
         pd.read_csv(data_folder / f"{dataset}.csv")
         for dataset in datasets
@@ -85,21 +85,21 @@ def preprocess_data(inputdata, full_data):
 
 def error_handling(error):
     error_msg = {"error": str(error)}
-    print(
+    FLOWER_LOGGER.error(
         f"Error will try to save error message: {error_msg}! Running: {RESULT_URL}..."
     )
     requests.post(RESULT_URL, data=json.dumps(error_msg), headers=HEADERS)
 
 
 def post_result(result: dict) -> None:
-    print(f"Running: {RESULT_URL}...")
+    FLOWER_LOGGER.debug(f"Posting result at: {RESULT_URL} ...")
     response = requests.post(RESULT_URL, data=json.dumps(result), headers=HEADERS)
     if response.status_code != 200:
         error_handling(response.text)
 
 
 def get_input() -> Inputdata:
-    print(f"Running: {INPUT_URL}...")
+    FLOWER_LOGGER.debug(f"Getting inputdata from: {INPUT_URL} ...")
     response = requests.get(INPUT_URL)
     if response.status_code != 200:
         error_handling(response.text)
@@ -109,7 +109,7 @@ def get_input() -> Inputdata:
 
 def get_enumerations(data_model: str, variable_name: str) -> list:
     try:
-        print(f"Running: {CDES_URL}...")
+        FLOWER_LOGGER.debug(f"Getting enumerations from: {CDES_URL} ...")
         response = requests.get(CDES_URL)
         if response.status_code != 200:
             error_handling(response.text)
@@ -126,8 +126,4 @@ def get_enumerations(data_model: str, variable_name: str) -> list:
         else:
             raise KeyError(f"'enumerations' key not found in {variable_name}")
     except (requests.RequestException, KeyError, json.JSONDecodeError) as e:
-        error_msg = {"error": str(e)}
-        print(
-            f"Error will try to save error message: {error_msg}! Running: {RESULT_URL}..."
-        )
-        requests.post(RESULT_URL, data=json.dumps(error_msg), headers=HEADERS)
+        error_handling(str(e))

exareme2/algorithms/flower/logistic_regression/server.py

@@ -27,7 +27,6 @@ def evaluate(server_round, parameters, config):
         accuracy = model.score(X_test, y_test)
         if server_round == NUM_OF_ROUNDS:
             post_result({"accuracy": accuracy})
-            print({"accuracy": accuracy})
         return loss, {"accuracy": accuracy}
 
     return evaluate

exareme2/algorithms/flower/mnist_logistic_regression/client.py

@@ -3,6 +3,7 @@
 
 import flwr as fl
 import numpy as np
+from flwr.common.logger import FLOWER_LOGGER
 from sklearn.linear_model import LogisticRegression
 from sklearn.metrics import log_loss
 
@@ -49,14 +50,13 @@ def fit(self, parameters, config):
                 ]
                 return_data = (params, len(X_train), {"accuracy": accuracy})
             except Exception as e:
-                print(f"Error during model fitting: {e}")
+                FLOWER_LOGGER.error(f"Error during model fitting: {e}")
                 # On error, default to zero-initialized parameters, no training examples, and zero accuracy
                 zero_params = [
                     np.zeros_like(param) for param in utils.get_model_parameters(model)
                 ]
                 return_data = (zero_params, 0, {"accuracy": 0.0})
 
-            print(f"Returning from fit: {return_data}")
             return return_data
 
         def evaluate(self, parameters, config):

exareme2/algorithms/flower/mnist_logistic_regression/server.py

@@ -29,7 +29,6 @@ def evaluate(server_round, parameters: fl.common.NDArrays, config):
         accuracy = model.score(X_test, y_test)
         if server_round == NUM_OF_ROUNDS:
             post_result({"accuracy": accuracy})
-            print({"accuracy": accuracy})
         return loss, {"accuracy": accuracy}
 
     return evaluate

exareme2/worker/flower/starter/starter_service.py

@@ -12,6 +12,9 @@ def start_flower_client(request_id: str, algorithm_name, server_address) -> int:
         "MONETDB_USERNAME": worker_config.monetdb.local_username,
         "MONETDB_PASSWORD": worker_config.monetdb.local_password,
         "MONETDB_DB": worker_config.monetdb.database,
+        "REQUEST_ID": request_id,
+        "WORKER_ROLE": worker_config.role,
+        "WORKER_IDENTIFIER": worker_config.identifier,
         "SERVER_ADDRESS": server_address,
         "NUMBER_OF_CLIENTS": worker_config.monetdb.database,
         "CONTROLLER_IP": worker_config.controller.ip,
@@ -32,6 +35,9 @@ def start_flower_server(
     request_id: str, algorithm_name: str, number_of_clients: int, server_address
 ) -> int:
     env_vars = {
+        "REQUEST_ID": request_id,
+        "WORKER_ROLE": worker_config.role,
+        "WORKER_IDENTIFIER": worker_config.identifier,
         "SERVER_ADDRESS": server_address,
         "NUMBER_OF_CLIENTS": number_of_clients,
         "CONTROLLER_IP": worker_config.controller.ip,