openstack/barbican
Code Issues Proposed changes

Fix for order obj expiration issue with SQLAlchemy

Browse Source 
By default, SQLAlchemy expires object state on commit. In some cases
(usually testing), the new_order object expires causing the whole
transaction to fail because SQLAlchemy raises an exception saying that
the object isn't bound to a session. This is because we're trying to
access the id attribute on a expired object. This patch pulls a copy
of the order_id off before we commit to the db; which can expire the
new_order object.

Change-Id: I799f6a52c20304abb6e537f5cbe69926efe38ead
This commit is contained in:
John Vrbanac 2015-03-07 19:20:18 -06:00
parent 926cbc18b5
commit b0bb65d59c
1 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
barbican/api/controllers/orders.py
View File

@ -195,13 +195,17 @@ class OrdersController(object):
new_order.project_id = project.id
self.order_repo.create_from(new_order)
# Grab our id before commit due to obj expiration from sqlalchemy
order_id = new_order.id
# Force commit to avoid async issues with the workers
repo.commit()
self.queue.process_type_order(order_id=new_order.id,
self.queue.process_type_order(order_id=order_id,
project_id=external_project_id)
url = hrefs.convert_order_to_href(new_order.id)
url = hrefs.convert_order_to_href(order_id)
pecan.response.status = 202
pecan.response.headers['Location'] = url