- error handling now uses a more comprehensive error communication system.

eveai_chat_workers/tasks.py

@@ -5,6 +5,7 @@ import traceback
 from flask import current_app
 from celery import states
 from sqlalchemy.exc import SQLAlchemyError, InterfaceError, OperationalError
+from redis.exceptions import ConnectionError as RedisConnectionError, TimeoutError as RedisTimeoutError
 
 from common.utils.config_field_types import TaggingFields
 from common.utils.database import Database
@@ -21,7 +22,9 @@ from common.utils.execution_progress import ExecutionProgressTracker
 
 
 # Healthcheck task
-@current_celery.task(name='ping', queue='llm_interactions')
+@current_celery.task(bind=True, name='ping', queue='llm_interactions',
+                     autoretry_for=(InterfaceError, OperationalError, RedisConnectionError, RedisTimeoutError, OSError),
+                     retry_backoff=True, retry_jitter=True, max_retries=5)
 def ping():
     return 'pong'
 
@@ -215,7 +218,9 @@ def prepare_arguments(specialist: Any, arguments: Dict[str, Any]) -> Dict[str, A
         raise ArgumentPreparationError(str(e))
 
 
-@current_celery.task(bind=True, name='execute_specialist', queue='llm_interactions', autoretry_for=(InterfaceError, OperationalError), retry_backoff=True, retry_jitter=True, max_retries=5)
+@current_celery.task(bind=True, name='execute_specialist', queue='llm_interactions',
+                     autoretry_for=(InterfaceError, OperationalError),
+                     retry_backoff=True, retry_jitter=True, max_retries=5)
 def execute_specialist(self, tenant_id: int, specialist_id: int, arguments: Dict[str, Any],
                        session_id: str, user_timezone: str) -> dict:
     """