Old md-cloud fixes that should better go to master

mediaire_toolbox/queue/daemon.py

@@ -54,8 +54,15 @@ def __init__(self,
         self.stopped = False
         self.processing_t_id = None
 
-        signal.signal(signal.SIGINT, self.exit_gracefully)
-        signal.signal(signal.SIGTERM, self.exit_gracefully)
+        try:
+            signal.signal(signal.SIGINT, self.exit_gracefully)
+            signal.signal(signal.SIGTERM, self.exit_gracefully)
+        except ValueError:
+            logger.warn('Catching ValueError from signal, are you using '
+                        'daemons outside of the main thread, and if so, '
+                        'do you know what you are doing?')
+            # (signal only works in main thread)
+            pass
 
     @abstractmethod
     def process_task(self, task):

mediaire_toolbox/transaction_db/t_db_retry.py

@@ -1,10 +1,8 @@
 import logging
 
-from tenacity.retry import retry_if_exception_type
+from tenacity.retry import retry_if_exception
 from tenacity import retry, stop_after_attempt, wait_fixed
 
-from sqlite3 import OperationalError as Sqlite3OperationalError
-from sqlalchemy.exc import OperationalError
 
 from mediaire_toolbox.constants import (
     RETRY_DATABASE_OP_SECONDS,
@@ -28,11 +26,17 @@ def before_sleep_log(retry_state):
                 retry_state.outcome))
 
 
+def is_operational_error(e):
+    return (
+        'OperationalError' in str(e) or
+        'OperationalError' in e.__class__.__name__
+    )
+
+
 def t_db_retry(f):
     return retry(
         retry=(
-            retry_if_exception_type(OperationalError)
-            | retry_if_exception_type(Sqlite3OperationalError)),
+            retry_if_exception(is_operational_error)),
         stop=stop_after_attempt(RETRY_DATABASE_OP_TIMES),
         wait=wait_fixed(RETRY_DATABASE_OP_SECONDS),
         before_sleep=before_sleep_log)(f)

mediaire_toolbox/transaction_db/transaction_db.py

@@ -236,12 +236,12 @@ def create_transaction(self,
     @lock
     def get_transaction(self, id_: int) -> Transaction:
         try:
-            return self._get_transaction_or_raise_exception(id_)
-        finally:
-            # we should always complete the lifetime of the connection,
-            # otherwise we might run into timeout errors
-            # (see https://docs.sqlalchemy.org/en/latest/orm/session_transaction.html)  # noqa: 501
+            t = self._get_transaction_or_raise_exception(id_)
             self.session.commit()
+            return t
+        except Exception:
+            self.session.rollback()
+            raise
 
     def _get_transaction_or_raise_exception(self, id_: int):
         t = self.session.query(Transaction).get(id_)
@@ -723,8 +723,9 @@ def get_study_metadata(self, study_id: str) -> StudiesMetadata:
         try:
             return self.session.query(StudiesMetadata)\
                 .filter_by(study_id=study_id).first()
-        finally:
             self.session.commit()
+        except Exception:
+            self.session.rollback()
 
     @t_db_retry
     def get_user_sites(self, user_id: int):  # TODO -> Query[UserSite]: