khyurri commented on a change in pull request #10254:
URL: https://github.com/apache/airflow/pull/10254#discussion_r496130366
##########
File path: airflow/utils/db.py
##########
@@ -606,6 +606,59 @@ def check_migrations(timeout):
log.info('Waiting for migrations... %s second(s)', ticker)
+def check_conn_id_duplicates(session=None):
+ """
+ Check unique conn_id in connection table
+ @param session: session of the sqlalchemy
+ @return: str
+ """
+ dups = session.query(Connection, func.count(Connection.conn_id)) \
+ .group_by(Connection.conn_id) \
+ .having(func.count(Connection.conn_id) > 1).all()
+ if dups:
+ return f'Seems you have non unique conn_id in connection table.\n' \
+ f'You have to manage those duplicate connections ' \
+ f'before upgrading the database.\n' \
+ f'Duplicated conn_id: {[dup[0] for dup in dups]}'
+ return ''
+
+
+def check_conn_type_null(session=None):
+ """
+ Check nullable conn_type column in connection table
+ @param session: session of the sqlalchemy
+ @return: str
+ """
+ n_nulls = session.query(Connection)\
+ .filter(Connection.conn_type.is_(None)).all()
+ if n_nulls:
+ return f'The conn_type column in the connection ' \
+ f'table must contain content.\n' \
+ f'Make sure you don\'t have null ' \
+ f'in the conn_type column.\n' \
+ f'Null conn_type conn_id: {list(n_nulls)}'
+ return ''
+
+
+@provide_session
+def auto_migrations_available(session=None):
+ """
+ @session: session of the sqlalchemy
+ @return: list[str]
+ """
+ errors_ = []
+ try:
+ for check_fn in (check_conn_id_duplicates, check_conn_type_null):
+ err = check_fn(session)
+ if err:
+ errors_.append(err)
+ except (exc.OperationalError, exc.ProgrammingError):
+ # fallback if tables hasn't been created yet
+ pass
Review comment:
> Of course, at the moment both check_conn_id_duplicates and
check_conn_type_null will raise the exception in the same scenario (Connection
table not existing), but if more check functions are added, then it might be
important to execute them all.
Is it not better to rewrite this part, if `except` will catch excessively?
Right now loop cancels, if database is not ready to upgrade.
Maybe I can catch more directly, only when table `Connection` is not exists?
But I have not found proper way to achieve that without creating custom
Exception class.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
