edda-framework 0.14.0__tar.gz → 0.14.1__tar.gz

{edda_framework-0.14.0 → edda_framework-0.14.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: edda-framework
-Version: 0.14.0
+Version: 0.14.1
 Summary: Lightweight Durable Execution Framework
 Project-URL: Homepage, https://github.com/i2y/edda
 Project-URL: Documentation, https://github.com/i2y/edda#readme

{edda_framework-0.14.0 → edda_framework-0.14.1}/edda/app.py

@@ -583,7 +583,6 @@ class EddaApp:
                 auto_resume_stale_workflows_periodically(
                     self.storage,
                     self.replay_engine,
-                    self.worker_id,
                     interval=60,
                 ),
                 name="leader_stale_workflow_resume",
@@ -628,7 +627,6 @@ class EddaApp:
                 auto_resume_stale_workflows_periodically(
                     self.storage,
                     self.replay_engine,
-                    self.worker_id,
                     interval=60,
                 ),
                 name="leader_stale_workflow_resume",
@@ -1411,7 +1409,8 @@ class EddaApp:
         from growing indefinitely with orphaned messages (messages that were
         published but never received by any subscriber).
 
-        Uses system-level locking to ensure only one pod executes cleanup at a time.
+        Important: This task should only be run by a single worker (e.g., via leader
+        election). It does not perform its own distributed coordination.
 
         Args:
             interval: Cleanup interval in seconds (default: 3600 = 1 hour)
@@ -1422,27 +1421,13 @@ class EddaApp:
         """
         while True:
             try:
-                # Add jitter to prevent thundering herd in multi-pod deployments
+                # Add jitter to prevent thundering herd
                 jitter = random.uniform(0, interval * 0.3)
                 await asyncio.sleep(interval + jitter)
 
-                # Try to acquire global lock for this task
-                lock_acquired = await self.storage.try_acquire_system_lock(
-                    lock_name="cleanup_old_messages",
-                    worker_id=self.worker_id,
-                    timeout_seconds=interval,
-                )
-
-                if not lock_acquired:
-                    # Another pod is handling this task
-                    continue
-
-                try:
-                    deleted_count = await self.storage.cleanup_old_channel_messages(retention_days)
-                    if deleted_count > 0:
-                        logger.info("Cleaned up %d old channel messages", deleted_count)
-                finally:
-                    await self.storage.release_system_lock("cleanup_old_messages", self.worker_id)
+                deleted_count = await self.storage.cleanup_old_channel_messages(retention_days)
+                if deleted_count > 0:
+                    logger.info("Cleaned up %d old channel messages", deleted_count)
             except Exception as e:
                 logger.error("Error cleaning up old messages: %s", e, exc_info=True)
 

{edda_framework-0.14.0 → edda_framework-0.14.1}/edda/locking.py

@@ -192,7 +192,6 @@ async def _refresh_lock_periodically(
 
 async def cleanup_stale_locks_periodically(
     storage: StorageProtocol,
-    worker_id: str,
     interval: int = 60,
 ) -> None:
     """
@@ -204,49 +203,37 @@ async def cleanup_stale_locks_periodically(
     Note: This function only cleans up locks without resuming workflows.
     For automatic workflow resumption, use auto_resume_stale_workflows_periodically().
 
-    Uses system-level locking to ensure only one pod executes cleanup at a time.
+    Important: This function should only be run by a single worker (e.g., via leader
+    election). It does not perform its own distributed coordination.
 
     Example:
         >>> asyncio.create_task(
-        ...     cleanup_stale_locks_periodically(storage, worker_id, interval=60)
+        ...     cleanup_stale_locks_periodically(storage, interval=60)
         ... )
 
     Args:
         storage: Storage backend
-        worker_id: Unique identifier for this worker (for global lock coordination)
         interval: Cleanup interval in seconds (default: 60)
     """
     with suppress(asyncio.CancelledError):
         while True:
-            # Add jitter to prevent thundering herd in multi-pod deployments
+            # Add jitter to prevent thundering herd
             jitter = random.uniform(0, interval * 0.3)
             await asyncio.sleep(interval + jitter)
 
-            # Try to acquire global lock for this task
-            lock_acquired = await storage.try_acquire_system_lock(
-                lock_name="cleanup_stale_locks",
-                worker_id=worker_id,
-                timeout_seconds=interval,
-            )
-
-            if not lock_acquired:
-                # Another pod is handling this task
-                continue
-
             try:
                 # Clean up stale locks
                 workflows = await storage.cleanup_stale_locks()
 
                 if len(workflows) > 0:
                     logger.info("Cleaned up %d stale locks", len(workflows))
-            finally:
-                await storage.release_system_lock("cleanup_stale_locks", worker_id)
+            except Exception as e:
+                logger.error("Failed to cleanup stale locks: %s", e, exc_info=True)
 
 
 async def auto_resume_stale_workflows_periodically(
     storage: StorageProtocol,
     replay_engine: Any,
-    worker_id: str,
     interval: int = 60,
 ) -> None:
     """
@@ -255,39 +242,27 @@ async def auto_resume_stale_workflows_periodically(
     This combines lock cleanup with automatic workflow resumption, ensuring
     that workflows interrupted by worker crashes are automatically recovered.
 
-    Uses system-level locking to ensure only one pod executes this task at a time,
-    preventing duplicate workflow execution (CRITICAL for safety).
+    Important: This function should only be run by a single worker (e.g., via leader
+    election). It does not perform its own distributed coordination.
 
     Example:
         >>> asyncio.create_task(
         ...     auto_resume_stale_workflows_periodically(
-        ...         storage, replay_engine, worker_id, interval=60
+        ...         storage, replay_engine, interval=60
         ...     )
         ... )
 
     Args:
         storage: Storage backend
         replay_engine: ReplayEngine instance for resuming workflows
-        worker_id: Unique identifier for this worker (for global lock coordination)
         interval: Cleanup interval in seconds (default: 60)
     """
     with suppress(asyncio.CancelledError):
         while True:
-            # Add jitter to prevent thundering herd in multi-pod deployments
+            # Add jitter to prevent thundering herd
             jitter = random.uniform(0, interval * 0.3)
             await asyncio.sleep(interval + jitter)
 
-            # Try to acquire global lock for this task
-            lock_acquired = await storage.try_acquire_system_lock(
-                lock_name="auto_resume_stale_workflows",
-                worker_id=worker_id,
-                timeout_seconds=interval,
-            )
-
-            if not lock_acquired:
-                # Another pod is handling this task
-                continue
-
             try:
                 # Clean up stale locks and get workflows to resume
                 workflows_to_resume = await storage.cleanup_stale_locks()
@@ -369,8 +344,8 @@ async def auto_resume_stale_workflows_periodically(
                                 e,
                                 exc_info=True,
                             )
-            finally:
-                await storage.release_system_lock("auto_resume_stale_workflows", worker_id)
+            except Exception as e:
+                logger.error("Failed to cleanup stale locks: %s", e, exc_info=True)
 
 
 class LockNotAcquiredError(Exception):

{edda_framework-0.14.0 → edda_framework-0.14.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "edda-framework"
-version = "0.14.0"
+version = "0.14.1"
 description = "Lightweight Durable Execution Framework"
 authors = [
     { name = "Yasushi Itoh", email = "6240399+i2y@users.noreply.github.com" }