Fix cache race condition when reading time series (fixes DEV-22)

enhydris/models/timeseries.py

@@ -209,8 +209,17 @@ def _get_data_from_cache(self, start_date, end_date):
             raise DataNotInCache()
         if self.start_date is None:
             return data  # Data should be empty in that case; just return it
-        start_date = max(start_date, self.start_date).astimezone(data.index.tzinfo)
-        end_date = min(end_date, self.end_date).astimezone(data.index.tzinfo)
+        try:
+            start_date = max(start_date, self.start_date).astimezone(data.index.tzinfo)
+            end_date = min(end_date, self.end_date).astimezone(data.index.tzinfo)
+        except TypeError:
+            # A TypeError will occur if self.start_date or self.end_date above is none.
+            # This should normally not happen, because self.start_date had already
+            # been checked immediately before, and self.end_date is not none whenever
+            # self.start_date is not none. However there's a race condition where
+            # another thread or process could have updated the time series (and
+            # invalidated the cache) in between these statements.
+            raise DataNotInCache()
         if data.index.min() > start_date or data.index.max() < end_date:
             raise DataNotInCache()
         return data.loc[start_date:end_date]

enhydris/tests/test_models/test_timeseries.py

@@ -497,6 +497,25 @@ def test_cached_empty_dataframe(self):
         data = self.timeseries.get_data(start_date=self.timeseries.end_date)
         self.assertEqual(len(data.data), 1)
 
+    @mock.patch("enhydris.models.timeseries.Timeseries._invalidate_cached_data")
+    def test_race_condition(self, m):
+        # Populate cache
+        self._get_data_and_check_num_queries(1, start_date=None, end_date=None)
+
+        # Pretend there's a race condition. We delete the time series data, but
+        # because we've mocked _invalidate_cached_data, the cache will not be deleted.
+        # However, we manually delete the cached end date. This simulates the case
+        # where another thread or process deletes the data (and invalidates the
+        # cache) between examining the start date and end date.
+        self.timeseries.set_data(HTimeseries())
+        cache.delete(f"timeseries_end_date_{self.timeseries.id}")
+
+        # Try getting the data again and see that it works. There are two queries now;
+        # one is for the end date.
+        with self.assertNumQueries(2):
+            data = self.timeseries.get_data(start_date=None, end_date=None)
+        self.assertEqual(len(data.data), 0)
+
 
 class TimeseriesSetDataTestCase(TestTimeseriesMixin, TestCase):
     def setUp(self):