Fixed that dispatches are never retried, but forever logged.

Signed-off-by: Mathias L. Baumann <[email protected]>

Author: Marenz

RELEASE_NOTES.md

@@ -14,4 +14,4 @@
 
 ## Bug Fixes
 
-<!-- Here goes notable bug fixes that are worth a special mention or explanation -->
+* Fixed that dispatches are never retried on failure, but instead an infinite loop of retry logs is triggered.

src/frequenz/dispatch/_actor_dispatcher.py

@@ -323,7 +323,14 @@ async def _run(self) -> None:
 
             async for selected in select(retry_recv, self._dispatch_rx):
                 if retry_recv.triggered(selected):
-                    self._retrier.retry(selected.message)
+                    dispatch = selected.message
+                    if not dispatch.started:
+                        _logger.info(
+                            "Giving up on dispatch %s, duration was exceeded.",
+                            dispatch.id,
+                        )
+                    else:
+                        await self._handle_dispatch(selected.message)
                 elif self._dispatch_rx.triggered(selected):
                     await self._handle_dispatch(selected.message)