From d8655e3b3d856af32a0102782aa55db01df845bc Mon Sep 17 00:00:00 2001
From: Elzbieta Kotulska <elzbieta.kotulska@frequenz.com>
Date: Thu, 28 Nov 2024 12:44:09 +0100
Subject: [PATCH] Fix ConfigManagingActor to handle missing config files
 gracefully

Eliminate recursive actor crashes when config files are missing.
The previous implementation suffered from a critical error pattern:
- The `_read_config` method is called at the start of the `_run` method
  and on FileWatcher event.
- When no config files existed, an exception would be raised
- This exception caused the actor to crash and immediately restart
- Restarting triggered the same `_read_config` method
- The cycle repeated, creating a persistent crash loop

This fix introduces a more robust approach:
- Detect missing config files without throwing exceptions
- Set up a FileWatcher to monitor for future config file creation
- call `_read_config` method as soon as any config file is crated.

Signed-off-by: Elzbieta Kotulska <elzbieta.kotulska@frequenz.com>
---
 src/frequenz/sdk/config/_config_managing.py | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/src/frequenz/sdk/config/_config_managing.py b/src/frequenz/sdk/config/_config_managing.py
index e66e2e461..b54d07890 100644
--- a/src/frequenz/sdk/config/_config_managing.py
+++ b/src/frequenz/sdk/config/_config_managing.py
@@ -107,14 +107,11 @@ def __init__(
         self._force_polling: bool = force_polling
         self._polling_interval: timedelta = polling_interval
 
-    def _read_config(self) -> abc.Mapping[str, Any]:
+    def _read_config(self) -> abc.Mapping[str, Any] | None:
         """Read the contents of the configuration file.
 
         Returns:
             A dictionary containing configuration variables.
-
-        Raises:
-            ValueError: If config file cannot be read.
         """
         error_count = 0
         config: dict[str, Any] = {}
@@ -138,14 +135,18 @@ def _read_config(self) -> abc.Mapping[str, Any]:
                 error_count += 1
 
         if error_count == len(self._config_paths):
-            raise ValueError(f"{self}: Can't read any of the config files")
+            _logger.error(
+                "%s: Can't read any of the config files, ignoring config update.", self
+            )
+            return None
 
         return config
 
     async def send_config(self) -> None:
         """Send the configuration to the output sender."""
         config = self._read_config()
-        await self._output.send(config)
+        if config is not None:
+            await self._output.send(config)
 
     async def _run(self) -> None:
         """Monitor for and send configuration file updates.