Use _stop_event consistently across all threads as the way to signal that it should stop. Add thread joins with timeouts to allow the program to exit cleanly

providers/sigrefdata/file_download_sig_ref_data_provider.py

@@ -1,6 +1,6 @@
 import logging
 from datetime import datetime
-from threading import Event, Thread
+from threading import Thread
 
 import pytz
 from requests.exceptions import ConnectionError, ConnectTimeout, ReadTimeout
@@ -21,19 +21,21 @@ class FileDownloadSIGRefDataProvider(SIGRefDataProvider):
         self._url = url
         self._poll_interval = poll_interval
         self._thread = None
-        self._stop_event = Event()
         self._url_data_cache = URLDataCache(f"sigrefdata_{sig_name}")
 
     def start(self):
         # Fire off the polling thread. It will poll immediately on startup, then sleep for poll_interval between
         # subsequent polls, so start() returns immediately and the application can continue starting.
         logger.info(f"Set up query of {self.sig_name} SIG ref data every {self._poll_interval!s} days.")
-        self._thread = Thread(target=self._run, name=f"FileDownloadSIGRefDataProvider-{self.sig_name}")
+        self._thread = Thread(target=self._run, name=f"FileDownloadSIGRefDataProvider-{self.sig_name}", daemon=True)
         self._thread.start()
 
     def stop(self):
         super().stop()
-        self._stop_event.set()
+        if self._thread:
+            self._thread.join(timeout=35)
+            if self._thread.is_alive():
+                logger.warning(f"{self.sig_name} SIG ref data worker thread did not exit on time and will be killed.")
 
     def _run(self):
         while True:

providers/sigrefdata/sig_ref_data_provider.py

@@ -1,5 +1,6 @@
 import logging
 from datetime import datetime
+from threading import Event
 
 import pytz
 
@@ -19,7 +20,7 @@ class SIGRefDataProvider:
         self.last_update_time = datetime.min.replace(tzinfo=pytz.UTC)
         self.status = "Not Started" if self.enabled else "Disabled"
         self.reference_count = 0
-        self._stop = False
+        self._stop_event = Event()
 
     def start(self):
         """Start the provider. This should return immediately after spawning threads to access the remote resources"""
@@ -30,7 +31,7 @@ class SIGRefDataProvider:
         """Stop any threads and prepare for application shutdown. Subclasses should implement this method and call
         super()."""
 
-        self._stop = True
+        self._stop_event.set()
 
     def _add_data(self, new_data):
         """Add all the provided reference data objects to the data store."""
@@ -43,7 +44,7 @@ class SIGRefDataProvider:
                 # For the big data sources, loading will take a few minutes. If we want to shut down the software neatly
                 # within the first few minutes of startup, we need a way to abort this expensive process of filling up the
                 # disk cache.
-                if self._stop:
+                if self._stop_event.is_set():
                     break
 
         self.reference_count = len(new_data)