Fix pickling error with CSVLogger (#10388)

* Don't store csv.Dictwriter in ExperimentWriter

* Add test for pickle after .save()

* Add entry in changelog

CHANGELOG.md

@@ -12,6 +12,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed `apply_to_collection(defaultdict)` ([#10316](https://github.com/PyTorchLightning/pytorch-lightning/issues/10316))
 - Fixed failure when `DataLoader(batch_size=None)` is passed ([#10345](https://github.com/PyTorchLightning/pytorch-lightning/issues/10345))
 - Fixed interception of `__init__` arguments for sub-classed DataLoader re-instantiation in Lite ([#10334](https://github.com/PyTorchLightning/pytorch-lightning/issues/10334))
+- Fixed issue with pickling `CSVLogger` after a call to `CSVLogger.save` ([#10388](https://github.com/PyTorchLightning/pytorch-lightning/pull/10388))
 
 
 ## [1.5.0] - 2021-11-02

pytorch_lightning/loggers/csv_logs.py

@@ -95,9 +95,9 @@ def save(self) -> None:
         metrics_keys = list(last_m.keys())
 
         with open(self.metrics_file_path, "w", newline="") as f:
-            self.writer = csv.DictWriter(f, fieldnames=metrics_keys)
-            self.writer.writeheader()
-            self.writer.writerows(self.metrics)
+            writer = csv.DictWriter(f, fieldnames=metrics_keys)
+            writer.writeheader()
+            writer.writerows(self.metrics)
 
 
 class CSVLogger(LightningLoggerBase):

tests/loggers/test_all.py

@@ -263,6 +263,10 @@ def _test_loggers_pickle(tmpdir, monkeypatch, logger_class):
     # the logger needs to remove it from the state before pickle
     _ = logger.experiment
 
+    # logger also has to avoid adding un-picklable attributes to self in .save
+    logger.log_metrics({"a": 1})
+    logger.save()
+
     # test pickling loggers
     pickle.dumps(logger)