Make delayed/chunked export configurable and off by default

docs/changes/newsfragments/5618.improved

@@ -0,0 +1,4 @@
+The feature of exporting large DataSets to netcdf by writing individual small files and combining them, introduced in QCoDeS 0.41.0 has been made configurable
+and turned off by default due to a number of corner cases where the export did not work correctly. The
+feature can be enabled when required by setting the config variable `qcodes.config.dataset.export_chunked_export_of_large_files_enabled`
+to True and the threshold controlled using `qcodes.config.dataset.export_chunked_threshold`

src/qcodes/configuration/qcodesrc.json

@@ -76,6 +76,8 @@
         "export_prefix": "qcodes_",
         "export_path": "{db_location}",
         "export_name_elements": ["captured_run_id", "guid"],
+        "export_chunked_export_of_large_files_enabled": false,
+        "export_chunked_threshold": 1000,
         "in_memory_cache": true
     },
     "telemetry":

src/qcodes/configuration/qcodesrc_schema.json

@@ -363,6 +363,16 @@
                     },
                     "default": ["captured_run_id", "guid"]
                 },
+                "export_chunked_export_of_large_files_enabled": {
+                    "type": "boolean",
+                    "default": false,
+                    "description": "Should large dataset be exported to netcdf in chuncks and then recombined into one file. This reduces the memory requirements for exporting the dataset but may be slower and fail in some corner cases"
+                },
+                "export_chunked_threshold": {
+                    "type": "integer",
+                    "default": 1000,
+                    "description": "Estimated size in MB above which the dataset will be exported in chuncks and recombined."
+                },
                 "in_memory_cache": {
                     "type": "boolean",
                     "default": true,

src/qcodes/dataset/data_set.py

@@ -247,7 +247,6 @@ def __init__(
         self._cache: DataSetCacheWithDBBackend = DataSetCacheWithDBBackend(self)
         self._results: list[dict[str, VALUE]] = []
         self._in_memory_cache = in_memory_cache
-        self._export_limit = 1000
 
         if run_id is not None:
             if not run_exists(self.conn, run_id):
@@ -1487,15 +1486,19 @@ def _export_as_netcdf(self, path: Path, file_name: str) -> Path:
         import xarray as xr
 
         file_path = path / file_name
-        if self._estimate_ds_size() > self._export_limit:
+        if (
+            qcodes.config.dataset.export_chunked_export_of_large_files_enabled
+            and self._estimate_ds_size()
+            > qcodes.config.dataset.export_chunked_threshold
+        ):
             log.info(
                 "Dataset is expected to be larger that threshold. Using distributed export.",
                 extra={
                     "file_name": str(file_path),
                     "qcodes_guid": self.guid,
                     "ds_name": self.name,
                     "exp_name": self.exp_name,
-                    "_export_limit": self._export_limit,
+                    "_export_limit": qcodes.config.dataset.export_chunked_threshold,
                     "_estimated_ds_size": self._estimate_ds_size(),
                 },
             )

tests/dataset/test_dataset_export.py

@@ -826,11 +826,24 @@ def test_export_dataset_small_no_delated(
     assert "Writing netcdf file directly" in caplog.records[0].msg
 
 
+def test_export_dataset_delayed_off_by_default(
+    tmp_path_factory: TempPathFactory, mock_dataset_grid: DataSet, caplog
+) -> None:
+    tmp_path = tmp_path_factory.mktemp("export_netcdf")
+    qcodes.config.dataset.export_chunked_threshold = 0
+    assert qcodes.config.dataset.export_chunked_export_of_large_files_enabled is False
+    with caplog.at_level(logging.INFO):
+        mock_dataset_grid.export(export_type="netcdf", path=tmp_path, prefix="qcodes_")
+
+    assert "Writing netcdf file directly." in caplog.records[0].msg
+
+
 def test_export_dataset_delayed_numeric(
     tmp_path_factory: TempPathFactory, mock_dataset_grid: DataSet, caplog
 ) -> None:
     tmp_path = tmp_path_factory.mktemp("export_netcdf")
-    mock_dataset_grid._export_limit = 0
+    qcodes.config.dataset.export_chunked_threshold = 0
+    qcodes.config.dataset.export_chunked_export_of_large_files_enabled = True
     with caplog.at_level(logging.INFO):
         mock_dataset_grid.export(export_type="netcdf", path=tmp_path, prefix="qcodes_")
 
@@ -863,7 +876,8 @@ def test_export_dataset_delayed(
     tmp_path_factory: TempPathFactory, mock_dataset_numpy: DataSet, caplog
 ) -> None:
     tmp_path = tmp_path_factory.mktemp("export_netcdf")
-    mock_dataset_numpy._export_limit = 0
+    qcodes.config.dataset.export_chunked_threshold = 0
+    qcodes.config.dataset.export_chunked_export_of_large_files_enabled = True
     with caplog.at_level(logging.INFO):
         mock_dataset_numpy.export(export_type="netcdf", path=tmp_path, prefix="qcodes_")
 
@@ -896,7 +910,8 @@ def test_export_dataset_delayed_complex(
     tmp_path_factory: TempPathFactory, mock_dataset_numpy_complex: DataSet, caplog
 ) -> None:
     tmp_path = tmp_path_factory.mktemp("export_netcdf")
-    mock_dataset_numpy_complex._export_limit = 0
+    qcodes.config.dataset.export_chunked_threshold = 0
+    qcodes.config.dataset.export_chunked_export_of_large_files_enabled = True
     with caplog.at_level(logging.INFO):
         mock_dataset_numpy_complex.export(
             export_type="netcdf", path=tmp_path, prefix="qcodes_"