Use `default_root_dir` as the `log_dir` with `LoggerCollection`s (#8187)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>

pytorch_lightning/trainer/properties.py

@@ -28,6 +28,7 @@ from pytorch_lightning.callbacks.base import Callback
 from pytorch_lightning.callbacks.prediction_writer import BasePredictionWriter
 from pytorch_lightning.core.optimizer import LightningOptimizer
 from pytorch_lightning.loggers import LightningLoggerBase
+from pytorch_lightning.loggers.base import LoggerCollection
 from pytorch_lightning.loggers.tensorboard import TensorBoardLogger
 from pytorch_lightning.loops import PredictionLoop
 from pytorch_lightning.loops.dataloader.evaluation_loop import EvaluationLoop
@@ -226,8 +227,12 @@ class TrainerProperties(ABC):
     def log_dir(self) -> Optional[str]:
         if self.logger is None:
             dirpath = self.default_root_dir
+        elif isinstance(self.logger, TensorBoardLogger):
+            dirpath = self.logger.log_dir
+        elif isinstance(self.logger, LoggerCollection):
+            dirpath = self.default_root_dir
         else:
-            dirpath = getattr(self.logger, 'log_dir' if isinstance(self.logger, TensorBoardLogger) else 'save_dir')
+            dirpath = self.logger.save_dir
 
         dirpath = self.accelerator.broadcast(dirpath)
         return dirpath

pytorch_lightning/trainer/trainer.py

@@ -233,7 +233,10 @@ class Trainer(
             limit_predict_batches: How much of prediction dataset to check (float = fraction, int = num_batches)
 
             logger: Logger (or iterable collection of loggers) for experiment tracking. A ``True`` value uses
-                the default ``TensorBoardLogger``. ``False`` will disable logging.
+                the default ``TensorBoardLogger``. ``False`` will disable logging. If multiple loggers are
+                provided and the `save_dir` property of that logger is not set, local files (checkpoints,
+                profiler traces, etc.) are saved in ``default_root_dir`` rather than in the ``log_dir`` of any
+                of the individual loggers.
 
             log_gpu_memory: None, 'min_max', 'all'. Might slow performance
 

tests/profiler/test_profiler.py

@@ -23,6 +23,8 @@ import torch
 from packaging.version import Version
 
 from pytorch_lightning import Callback, Trainer
+from pytorch_lightning.loggers.base import LoggerCollection
+from pytorch_lightning.loggers.tensorboard import TensorBoardLogger
 from pytorch_lightning.profiler import AdvancedProfiler, PassThroughProfiler, PyTorchProfiler, SimpleProfiler
 from pytorch_lightning.profiler.pytorch import RegisterRecordFunction
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
@@ -430,6 +432,36 @@ def test_pytorch_profiler_nested(tmpdir):
     assert events_name == expected, (events_name, torch.__version__, platform.system())
 
 
+def test_pytorch_profiler_logger_collection(tmpdir):
+    """
+    Tests whether the PyTorch profiler is able to write its trace locally when
+    the Trainer's logger is an instance of LoggerCollection. See issue #8157.
+    """
+
+    def look_for_trace(trace_dir):
+        """ Determines if a directory contains a PyTorch trace """
+        return any("trace.json" in filename for filename in os.listdir(trace_dir))
+
+    # Sanity check
+    assert not look_for_trace(tmpdir)
+
+    model = BoringModel()
+
+    # Wrap the logger in a list so it becomes a LoggerCollection
+    logger = [TensorBoardLogger(save_dir=tmpdir)]
+    trainer = Trainer(
+        default_root_dir=tmpdir,
+        profiler="pytorch",
+        logger=logger,
+        limit_train_batches=5,
+        max_epochs=1,
+    )
+
+    assert isinstance(trainer.logger, LoggerCollection)
+    trainer.fit(model)
+    assert look_for_trace(tmpdir)
+
+
 @RunIf(min_gpus=1, special=True)
 def test_pytorch_profiler_nested_emit_nvtx(tmpdir):
     """

tests/trainer/properties/log_dir.py

@@ -15,7 +15,7 @@ import os
 
 from pytorch_lightning import Trainer
 from pytorch_lightning.callbacks import ModelCheckpoint
-from pytorch_lightning.loggers import TensorBoardLogger
+from pytorch_lightning.loggers import LoggerCollection, TensorBoardLogger
 from tests.helpers.boring_model import BoringModel
 
 
@@ -140,3 +140,20 @@ def test_logdir_custom_logger(tmpdir):
     assert trainer.log_dir == expected
     trainer.fit(model)
     assert trainer.log_dir == expected
+
+
+def test_logdir_logger_collection(tmpdir):
+    """Tests that the logdir equals the default_root_dir when the logger is a LoggerCollection"""
+    default_root_dir = tmpdir / "default_root_dir"
+    save_dir = tmpdir / "save_dir"
+    model = TestModel(default_root_dir)
+    trainer = Trainer(
+        default_root_dir=default_root_dir,
+        max_steps=2,
+        logger=[TensorBoardLogger(save_dir=save_dir, name='custom_logs')]
+    )
+    assert isinstance(trainer.logger, LoggerCollection)
+    assert trainer.log_dir == default_root_dir
+
+    trainer.fit(model)
+    assert trainer.log_dir == default_root_dir