update statshandler logging message

monai/handlers/stats_handler.py

@@ -19,7 +19,7 @@
 import torch
 
 from monai.config import IgniteInfo
-from monai.utils import is_scalar, min_version, optional_import
+from monai.utils import deprecated_arg_default, is_scalar, min_version, optional_import
 
 Events, _ = optional_import("ignite.engine", IgniteInfo.OPT_IMPORT_VERSION, min_version, "Events")
 if TYPE_CHECKING:
@@ -41,9 +41,8 @@ class StatsHandler:
 
     Note that if `name` arg is None, will leverage `engine.logger` as default logger directly, otherwise,
     get logger from `logging.getLogger(name)`, we can setup a logger outside first with the same `name`.
-    As the default log level of `RootLogger` is `WARNING`, may need to call
-    `logging.basicConfig(stream=sys.stdout, level=logging.INFO)` before running this handler to enable
-    the stats logging.
+    As the default log level is `WARNING`, it's recommended to call
+    `logging.getLogger(name).setLevel(logging.INFO)` before running this handler to enable the stats logging.
 
     Default behaviors:
         - When EPOCH_COMPLETED, logs ``engine.state.metrics`` using ``self.logger``.
@@ -52,7 +51,7 @@ class StatsHandler:
 
     Usage example::
 
-        logging.basicConfig(stream=sys.stdout, level=logging.INFO)
+        logging.getLogger("train_stats").setLevel(logging.INFO)
 
         trainer = SupervisedTrainer(...)
         StatsHandler(name="train_stats").attach(trainer)
@@ -64,6 +63,14 @@ class StatsHandler:
 
     """
 
+    @deprecated_arg_default(
+        "name",
+        old_default=None,
+        new_default="StatsHandler",
+        since="1.1",
+        replaced="1.3",
+        msg_suffix="the default logger is 'StatsHandler'.",
+    )
     def __init__(
         self,
         iteration_log: bool | Callable[[Engine, int], bool] = True,
@@ -135,10 +142,14 @@ def attach(self, engine: Engine) -> None:
         """
         if self.name is None:
             self.logger = engine.logger
-        if self.logger.getEffectiveLevel() > logging.INFO or logging.root.getEffectiveLevel() > logging.INFO:
+        if self.logger.getEffectiveLevel() > logging.INFO:
+            suggested = f"\n\nimport logging\nlogging.getLogger('{self.logger.name}').setLevel(logging.INFO)"
+            if self.logger.name != engine.logger.name:
+                suggested += f"\nlogging.getLogger('{engine.logger.name}').setLevel(logging.INFO)"
+            suggested += "\n\n"
             warnings.warn(
-                "the effective log level of engine logger or RootLogger is higher than INFO, may not record log,"
-                " please call `logging.basicConfig(stream=sys.stdout, level=logging.INFO)` to enable it."
+                f"the effective log level of {self.logger.name} higher than INFO, StatsHandler may not generate logs,"
+                f"\nplease use the following code before running the engine to enable it: {suggested}"
             )
         if self.iteration_log and not engine.has_event_handler(self.iteration_completed, Events.ITERATION_COMPLETED):
             event = Events.ITERATION_COMPLETED

tests/test_handler_stats.py

@@ -254,7 +254,8 @@ def _train_func(engine, batch):
 
         # set up testing handler
         stats_handler = StatsHandler(name=None, tag_name=key_to_print)
-        stats_handler.attach(engine)
+        with self.assertWarns(Warning):  # engine logging level warn
+            stats_handler.attach(engine)
         # leverage `engine.logger` to print info
         engine.logger.setLevel(logging.INFO)
         level = logging.root.getEffectiveLevel()