diff --git a/plesna/storage/metadata_repository/fs_metadata_repository.py b/plesna/storage/metadata_repository/fs_metadata_repository.py
index 3714a49..a2a1de3 100644
--- a/plesna/storage/metadata_repository/fs_metadata_repository.py
+++ b/plesna/storage/metadata_repository/fs_metadata_repository.py
@@ -1,5 +1,6 @@
 from pathlib import Path
 import csv
+import json
 
 from plesna.storage.metadata_repository.metadata_repository import (
     ExecutionMetaData,
@@ -37,7 +38,7 @@ class FSMetaDataRepository(MetaDataRepository):
         filepath = self._basepath / self.EXECUTION_FILEMODEL.format(flux_id=flux_id)
         filepath.touch()
         with open(filepath, "a") as csvfile:
-            writer = csv.DictWriter(csvfile, fieldnames=ExecutionMetaData.__fields__.keys())
+            writer = csv.DictWriter(csvfile, fieldnames=ExecutionMetaData.model_fields.keys())
             writer.writeheader()
         return flux_id
 
@@ -49,7 +50,7 @@ class FSMetaDataRepository(MetaDataRepository):
             raise FSMetaDataRepositoryError(f"The flux {flux_id} hasn't been added yet.")
 
         with open(filepath, "a") as csvfile:
-            writer = csv.DictWriter(csvfile, fieldnames=ExecutionMetaData.__fields__.keys())
+            writer = csv.DictWriter(csvfile, fieldnames=ExecutionMetaData.model_fields.keys())
             writer.writerow(metadata_.to_flat_dict())
 
         return metadata_
@@ -65,8 +66,9 @@ class FSMetaDataRepository(MetaDataRepository):
             raise FSMetaDataRepositoryError(f"The flux {flux_id} hasn't been added yet.")
         with open(filepath, "r") as csvfile:
             logs = []
-            reader = csv.DictReader(csvfile, fieldnames=ExecutionMetaData.__fields__.keys())
+            reader = csv.DictReader(csvfile)
             for row in reader:
+                row["output"] = json.loads(row["output"])
                 logs.append(ExecutionMetaData(**row))
         return logs
 
diff --git a/plesna/storage/metadata_repository/metadata_repository.py b/plesna/storage/metadata_repository/metadata_repository.py
index 7ea66e5..151749d 100644
--- a/plesna/storage/metadata_repository/metadata_repository.py
+++ b/plesna/storage/metadata_repository/metadata_repository.py
@@ -16,7 +16,7 @@ class ExecutionMetaData(BaseModel):
     output: FluxMetaData
 
     def to_flat_dict(self):
-        return {"datetime": self.datetime.timestamp(), "output": self.output.model_dump_json()}
+        return {"datetime": self.datetime.isoformat(), "output": self.output.model_dump_json()}
 
 
 class MetaDataRepository:
diff --git a/tests/storage/test_fs_metadata_repository.py b/tests/storage/test_fs_metadata_repository.py
index 788425e..a2d18b2 100644
--- a/tests/storage/test_fs_metadata_repository.py
+++ b/tests/storage/test_fs_metadata_repository.py
@@ -4,7 +4,9 @@ from pathlib import Path
 
 import pytest
 
+from plesna.models.flux import FluxMetaData
 from plesna.storage.metadata_repository.fs_metadata_repository import FSMetaDataRepository
+from plesna.storage.metadata_repository.metadata_repository import ExecutionMetaData
 
 
 @pytest.fixture
@@ -51,4 +53,38 @@ def test_register_flux_execution(location, metadata_repository):
     metadata_filepath = location / metadata_repository.EXECUTION_FILEMODEL.format(flux_id=flux_id)
     with open(metadata_filepath, "r") as csvfile:
         content = csvfile.read()
-        assert content == 'datetime,output\n1678887000.0,"{""data"":{""truc"":""machin""}}"\n'
+        assert (
+            content == 'datetime,output\n2023-03-15T14:30:00,"{""data"":{""truc"":""machin""}}"\n'
+        )
+
+
+def test_register_and_get_flux_execution(location, metadata_repository):
+    flux_id = "my_flux"
+    metadata_repository.add_flux(flux_id)
+
+    metadata_repository.register_flux_execution(
+        flux_id,
+        metadata={
+            "datetime": datetime(2023, 3, 15, 14, 30),
+            "output": {"data": {"truc": "machin"}},
+        },
+    )
+    metadata_repository.register_flux_execution(
+        flux_id,
+        metadata={
+            "datetime": datetime(2024, 3, 15, 14, 30),
+            "output": {"data": {"truc": "chose"}},
+        },
+    )
+
+    logs = metadata_repository.flux_logs(flux_id)
+    assert logs == [
+        ExecutionMetaData(
+            datetime=datetime(2023, 3, 15, 14, 30),
+            output=FluxMetaData(data={"truc": "machin"}),
+        ),
+        ExecutionMetaData(
+            datetime=datetime(2024, 3, 15, 14, 30),
+            output=FluxMetaData(data={"truc": "chose"}),
+        ),
+    ]